Tuned Graft Copolymers as Controlled Coatings for DNA Microarrays

ETH-Zürich, Wolfgang-Pauli-Strasse 10, CH-8093 Zürich, Switzerland, and Zeptosens AG, Benkenstrasse 254,. CH-4108 Witterswil, Switzerland...
4 downloads 0 Views 180KB Size
Download PDF
Anal. Chem. 2005, 77, 5831-5838

Tuned Graft Copolymers as Controlled Coatings for DNA Microarrays Susan M. De Paul,†,‡ Didier Falconnet,† Ste´phanie Pasche,† Marcus Textor,*,† Andreas P. Abel,§,| Ekkehard Kauffmann,§ Roman Liedtke,§ and Markus Ehrat§

Laboratory for Surface Science and Technology, Department of Materials, Swiss Federal Institute of Technology, ETH-Zu¨rich, Wolfgang-Pauli-Strasse 10, CH-8093 Zu¨rich, Switzerland, and Zeptosens AG, Benkenstrasse 254, CH-4108 Witterswil, Switzerland

DNA microarrays have become a powerful tool for expression profiling and other genomics applications. A critical factor for their sensitivity is the interfacial coating between the chip substrate and the bound DNA. Such a coating has to embrace the divergent requirements of tightly binding the capture probe DNA during the spotting process and of minimizing the nonspecific binding of target DNA during the hybridization assay. To fulfill these conditions, most coatings require a passivation step. Here we demonstrate how the chain density of a graft copolymer with a polycationic backbone, poly(L-lysine)-graft-poly(ethylene glycol), can be tuned such that the binding capacity during capture probe deposition is maximized while the nonspecific binding during hybridization assays is kept to a minimum, thus alleviating the requirement for a separate passivation procedure. Evidence for the superior performance of such coatings in terms of signalto-noise ratio and spot quality is presented using an evanescent field-based fluorescent sensing technique (the ZeptoREADER). The surface architecture is further characterized using optical waveguide lightmode spectroscopy and time-of-flight secondary ion mass spectrometry. Finally, in a model assay, we demonstrate that expression changes can be detected from 1 µg of total mRNA sample material with a limit of detectable differential expression of (1.5. Many areas of interest in current genomics research (including gene expression, genotyping, and analysis of single-nucleotide polymorphism) rely on the detection of low levels of analyte. For instance, when using a DNA microarray for gene expression analysis,1,2 a common approach is to detect cDNA that was reverse transcribed from mRNA. However, less than 5% of the total RNA in a typical cell is, in fact, messenger RNA. Furthermore, of these * To whom correspondence should be addressed, E-mail: [email protected]. Tel.: +41 44 632 6451. Fax: +41 44 633 1027. † Swiss Federal Institute of Technology. ‡ Current address: Solvias AG, Klybeckstrasse 191, Postfach, CH-4002 Basel, Switzerland. § Zeptosens AG. | Current address: Zu ¨ h˙ lke Engineering AG, Gescha¨ftsstelle Basel, Florenzstr. 9, Postfach, CH-4023 Basel, Switzerland. (1) Schena, M.; Shalon, D.; Davis, R. W.; Brown, P. O. Science 1995, 270, 467470. (2) DeRisi, J. L.; Iyer, V. R.; Brown, P. O. Science 1997, 278, 680-686. 10.1021/ac0504666 CCC: $30.25 Published on Web 08/13/2005

© 2005 American Chemical Society

genes, only 0.1% are highly abundant, with over 5000 copies/cell.3 Approximately 95% of the mRNA in a cell consists of so-called low-abundant genes, with fewer than 10 transcripts/cell. Detection and accurate quantification of such small amounts of material remain major challenges in DNA microarray analysis. Although biochemical amplification methods such as the polymerase chain reaction4 and T-7 amplification5 can be used to enhance the number of target molecules, the potential sequence-dependent bias in such approaches introduces ambiguities and often leads to poor reproducibility.6 Similarly, indirect signal amplification methods also have their drawbacks. Thus, a critical need exists for a sophisticated, ultrasensitive, and quantitative analytical detection approach that is capable of being used with unamplified genomic material. Recently, a novel platform that uses the evanescent field generated in a planar waveguide to excite fluorescence in a surface-specific and sensitive way has been developed and commercialized by Zeptosens AG7,8 as the ZeptoREADER. Such an approach has several advantages compared to conventional epifluorescent configurations. First, since the excitation is restricted to within 200 nm of the surface, only fluorophores within this distance from the surface contribute to the signal, allowing measurements to be made in the presence of bulk analyte solution and decreasing contributions from background fluorescence. Elimination of rinsing steps not only saves time and material but also permits in situ studies of kinetics and exchange, and the use of a liquid environment during image acquisition, as is typically done with the ZeptoREADER, significantly reduces light scattering. An additional advantage is that, unlike in a confocal setup, the evanescent field is uniformly and simultaneously available across the entire propagation path with a high-field strength in (3) Velculescu, V. E.; Madden, S. L.; Zhang, L.; Lash, A. E.; Yu, J.; Rago, C.; Lal, A.; Wang, C. J.; Beaudry, G. A.; Ciriello, K. M.; Cook, B. P.; Dufault, M. R.; Ferguson, A. T.; Gao, Y.; He, T.-C.; Hermeking, H.; Hiraldo, S. K.; Hwang, P. M.; Lopez, M. A.; Luderer, H. F.; Mathews, B.; Petroziello, J. M.; Polyak, K.; Zawel, L.; Zhang, W.; Zhang, X.; Zhou, W.; Haluska, F. G.; Jen, J.; Sukumar, S.; Landes, G. M.; Riggins, G. J.; Vogelstein, B.; Kinzler, K. W. Nat. Genet. 1999, 23, 387-388. (4) Scharf, S. J.; Horn, G. T.; Erlich, H. A. Science 1986, 233, 1076-1078. (5) Van Gelder, R. N.; von Zastrow, M. E.; Yool, A.; Dement, W. C.; Barchas, J. D.; Eberwine, J. H. Proc. Nat. Acad. Sci. U.S.A. 1990, 87, 1663-1667. (6) Spiess, A.-N.; Mueller, N.; Ivell, R. BMC Genomics 2003, 4, 44. (7) Ehrat, M.; Duveneck, G. L.; Kresbach, G. M.; Oroszlan, P.; Paulus, A. Chimia 1997, 51, 705-713. (8) Duveneck, G. L.; Abel, A. P.; Bopp, M. A.; Kresbach, G. M.; Ehrat, M. Anal. Chim. Acta 2002, 469, 49-61.

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005 5831

the vicinity of the surface. A full microarray can thus be imaged at a single point in time by collecting the isotropically emitted fluorescence with a CCD camera. Finally, the design of the optics is robust since the incoupling of the light and the excitation of the fluorescence are spatially separated. All of these features lead to an enhancement of the signal-to-noise ratio over conventional scanning systems by a factor of ∼60. However, a highly sensitive detection system alone is not enough to ensure reproducible and stable assays. An important factor that must not be neglected is the interface between the transducersin this case the optical waveguidesand the DNA. For DNA microarrays in which ex situ produced probes are to be spotted onto the surface, some sort of coating is typically applied to the transducer before deposition of the probes. [Throughout this paper, the nomenclature suggested in ref 9 is used where the words “probe” or “capture probe” refer to the nucleic acids (e.g., long oligonucleotides or cDNA) that are attached to the transducer and the word “target” refers to the pool of free fluorescently labeled genetic material (e.g., cDNA) that is to be monitored via hybridization reactions.] Such an interfacial coating must be homogeneous, reproducible, easy to apply, and chemically stable. In addition, it should be capable of quickly and stably attaching the spotted capture probes in an orientation that facilities the subsquent kinetics of target recognition. The coating must not show an affinity for any of the fluorescent dyes, additives, or analytes present during the hybridization step, and both the binding and hybridization performance of the microarray should be independent of the form of the DNA. Last, but not least, both the coating material itself and the production process ought to be cost-effective. To date, a variety of different techniques have been used to modify the surface of transducers,10,11 among them silanization12 (e.g., with amino-, epoxy-, or aldehyde-terminated silanes), selfassembly of alkanethiols,13 plasma polymerization, and hydrogel formation.14,15 These can then be combined with spotting, microfluidics, or photolithographic patterning to produce ordered arrays of capture probes. One of the earliest, and technically simplest, immobilization chemistries used for spotted DNA1,16 was an interfacial layer of the polyelectrolyte poly(L-lysine) (PLL). Since glass slides are negatively charged at neutral pH, PLL, which has primary amine groups with a pKa of ∼10.5, readily adsorbs to these surfaces, rendering them slightly positively charged and thus capable of attracting DNA or oligonucleotides, which, due to their phosphodiester backbones, are strongly negatively charged. However, it is necessary to passivate the surfaces afterwardsoften by reacting the remaining primary amines with aldehyde or anhydride groupssto neutralize or reverse the charge on the (9) Phimister, B. Nat. Gen. Suppl. 1999, 21, 1. (10) Angenendt, P.; Glo¨kler, J.; Murphy, D.; Lehrach, H.; Cahill, D. J. Anal. Biochem. 2002, 309, 253-260. (11) Tomizaki, K.-y.; Usui, K.; Mihara, H. ChemBioChem 2005, 6, 1-18. (12) Bhatia, S. K.; Teixeira, J. L.; Anderson, M.; Shriver-Lake, L. C.; Calvert, J. M.; Georger, J. H.; Hickman, J. J.; Dulcey, C. S.; Schoen, P. E.; Ligler, F. S. Anal. Biochem. 1993, 208, 197-205. (13) Bain, C. D.; Troughton, E. B.; Tao, Y. T.; Evall, J.; Whitesides, G. M.; Nuzzo, R. G. J. Am. Chem. Soc. 1989, 111, 321-335. (14) Gong, P.; Grainger, D. W. Surf. Sci. 2004, 570, 67-77. (15) Wischerhoff, E. In Protein Arrays, Biochips, and Proteomics: The Next Phase of Genomic Discovery; Albala, J. S., Humphrey-Smith, I., Eds.; Marcel Dekker: New York, 2003; pp 159-171. (16) http://cmgm.stanford.edu/pbrown/protocols/1_slides.html.

5832 Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

Figure 1. (a) Idealized chemical structure of the graft copolymer PLL-g-PEG. The grafting ratio is given by (m + n)/n (b) Schematic of polymer architecture as a function of grafting ratio. Low grafting ratios (far left) correspond to surfaces with high PEG densities, which would not be expected to bind significant amounts of oligonucleotides. PLL itself (far right) can be considered to have an infinite g value and is known to bind DNA. Between lie architectures that can potentially bind moderate amounts of negatively charged oligonucleotides.

surface so that the fluorescently labeled target DNA will not adsorb in a nonspecific manner during the hybridization process.1,17,18 Such passivation can potentially damage the spotted DNA, lower the quality of the microarray, or both. Recently, we have been studying19,20,21 a graft copolymer derivative of poly(L-lysine) called poly(L-lysine)-graft-poly(ethylene glycol) (PLL-g-PEG), which was originally developed as a coating for microcapsules.22 As in the case of PLL itself, the copolymer readily adsorbs to surfaces that are negatively charged at physiological pH, including, in addition to glass, metal oxides such as tantalum pentoxide, niobium pentoxide, silicon dioxide, titanium dioxide, and indium tin oxide, and tissue culture polystyrene. The architecture of the polymer can be described in terms of its grafting ratio, g, which is defined as the ratio of the total number of lysine units to the number of grafted poly(ethylene glycol) side chains (see Figure 1). Thus, a low g value corresponds to a polymer with many grafted PEG side chains, and a high g value indicates a polymer with few grafted chains. When the density of the PEG chains is sufficiently high (e.g., g ≈ 3.5 for 2-kDa PEG chains), PLL-g-PEG resists the adsorption of proteins from solution (17) Diehl, F.; Grahlmann, S.; Beier, M.; Hoheisel, J. D. Nucleic Acids Res. 2001, 29, e38. (18) Duggan, D. J.; Bittner, M.; Chen, Y.; Meltzer, P.; Trent, J. M. Nat. Gen. Suppl. 1999, 21, 10-14. (19) Kenausis, G. L.; Vo ¨ro ¨s, J.; Elbert, D. L.; Huang, N.-P.; Hofer, R.; Ruiz-Taylor, L.; Textor, M.; Hubbell, J. A.; Spencer, N. D. J. Phys. Chem. B 2000, 104, 3298-3309. (20) Huang, N.-P.; Michel, R.; Vo ¨ro ¨s, J.; Textor, M.; Hofer, R.; Rossi, A.; Elbert, D. L.; Hubbell, J. A.; Spencer, N. D. Langmuir 2001, 17, 489-498. (21) Pasche, S.; De Paul, S. M.; Vo ¨ro ¨s, J.; Spencer, N. D.; Textor, M. Langmuir 2003, 19, 9216-9225. (22) Sawhney, A. S.; Hubbell, J. A. Biomaterials 1992, 13, 863-870.

and is thus useful in immunoassay biosensing applications23 as well as for coating biomaterials.24-26 Among the advantages of PLLg-PEG are that is it easy to synthesize and characterize, that it can be applied to surfaces by simply dip-coating in aqueous buffered solutions, and that the adsorbed polymer film is stable over extended periods of time even when in contact with biological fluids such as fully serum-supplemented cell culture media.26 Although such PEG-dense coatings might not seem to be likely candidates for binding DNA to surfaces, it is possible to vary the architecture of PLL-g-PEG and thus obtain coatings with different properties.27 In this paper, we examine how optimization of the grafting ratio of PLL-g-PEG allows one to produce coatings for genomic microarrays that lead to a high DNA spot quality and that do not require a subsequent reactive blocking step. A surfacespecific mass spectrometry method is used to demonstrate our ability to tune the grafting ratio and thereby the density of the PEG chains on the surface. By using two optical planar waveguide techniques, optical waveguide light-mode spectroscopy (OWLS) and the ZeptoREADER, we determine the DNA binding capacity as a function of the grafting ratio of PLL-g-PEG and find an optimal range of g values for 2-kDa PEG chains. Finally, a SensiChip DNA microarray experiment that uses such an optimized interfacial coating shows that it is possible to accurately detect gene expression changes from medium- and low-expressed genes using as little as 1 µg of total RNA starting material with neither target nor signal amplification. EXPERIMENTAL SECTION Synthesis of PLL-g-PEG. Poly(L-lysine hydrobromide) (Catalog No. P-7890, nominal MW 15 000-30 000) was obtained from Sigma-Aldrich (Buchs, Switzerland), and the N-hydroxysuccinimidyl ester of methoxypoly(ethylene glycol) propionic acid (MeOPEG-SPA, MW 2000) was purchased from Shearwater Polymers, Inc. (now Nektar Therapeutics, Huntsville, AL). PLL-g-PEG synthesis was performed according to ref 23. All salts were obtained from Fluka (Buchs, Switzerland), and buffers were prepared using ultrapure water filtered to obtain a resistivity of at least 18.2 MΩ and an organic content of less than 5 ppb. The grafting ratio was varied by altering the relative amounts of the two reagents, and the grafting ratio of the final product was determined (to within an error of ∼0.5) from 1H solution-state NMR spectroscopy according to ref 23. For the time-of-flight secondary ion mass spectrometry (TOFSIMS) and grafting ratio optimization studies (cf. Figures 3 and 5), a series of 10 polymers with various grafting ratios was synthesized from the same batch of reagents on a single day. The batch of poly(L-lysine hydrobromide) that was used had a molecular weight of 11900 (despite the listed specifications) and a polydispersity (Mw/Mn) of 1.1 according to the low-angle laser (23) Huang, N.-P.; Vo¨ro ¨s, J.; De Paul, S. M.; Textor, M.; Spencer, N. D. Langmuir 2002, 18, 220-230. (24) VandeVondele, S.; Vo ¨ro ¨s, J.; Hubbell, J. A. Biotechnol. Bioeng. 2003, 82, 784-790. (25) Tosatti, S.; Schwartz, Z.; Campbell, C.; Cochran, D. L.; VandeVondele, S.; Hubbell, J. A.; Denzer, A.; Simpson, J.; Wieland, M.; Lohmann, C. H.; Textor, M.; Boyan, B. D. J. Biomed. Mater. Res. 2004, 68A, 458-472. (26) Michel, R.; Lussi, J. W.; Csu´cs, G.; Reviakine, I.; Danuser, G.; Ketterer, B.; Hubbell, J. A.; Textor, M.; Spencer, N. D. Langmuir 2002, 18, 3281-3287. (27) De Paul, S. M.; Pasche, S.; Textor, M.; Spencer, N. D.; Kauffmann, E.; Abel, A.; Ehrat, M. Presented at the Seventh World Congress on Biosensors, Kyoto, Japan, May 15-17, 2002.

light scattering results provided by the manufacturer. The MeOPEG-SPA had a molecular weight of 1953 and a polydispersity of 1.03 as determined by the manufacturer using gel permeation chromatography. The NMR-determined bulk grafting ratios of the 10 polymers are as follows: 3.6, 5.2, 6.0, 7.2, 8.8, 9.3, 10.5, 12.2, 14.2, and 18.0. The batch of ungrafted poly(L-lysine hydrobromide) used as a reference material had a molecular weight of 16 100 and a polydispersity of Mw/Mn)1.3. Other figures in this paper (e.g., Figure 4) represent the results of experiments that used different polymer batches but with similar PLL and PEG molecular weights. The grafting ratios of these polymers, also determined by NMR, are 3.7, 8.3, and 17.0. Throughout the remainder of this text, the grafting ratio of a given polymer will be listed in brackets behind the letter “g”, e.g., PLL-g[3.6]-PEG. Substrates. To facilitate the comparison of the results of different techniques, it is desirable to use substrates that are as similar as possible, both chemically and topographically. Although the nature of the underlying support is largely dictated by the experimental method itself (i.e., conductive support for TOF-SIMS, optically transparent supports for the two planar waveguide methods), the uppermost layer was chosen to be a sputter-coated tantalum pentoxide layer in all three cases. The substrates for the TOF-SIMS measurements consisted of silicon wafers (WaferNet GmbH, Eching, Germany) onto which a 12-nm-thick layer of Ta2O5 was sputter-coated by Unaxis Optics (Balzers, Liechtenstein). The layer was deliberately kept thin ( 3, and all of these genes are found to fall within a range of (1.5-fold differential expression. Thus, a gene expressed differentially by a factor of more than 1.5 can be reliably detected with such an array. (40) Sanchez-Cabo, F.; Cho, K.-H.; Hinds, J.; Wolkenhauer, O. Appl. Bioinformatics in press.

5838 Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

PLL-g-PEG surface coatings with grafting ratios that have been optimized for DNA microarray applications provide a strong signalto-background ratio in hybridization experiments and can be produced with robust and reproducible manufacturing procedures. Results from TOF-SIMS experiments show that the NMRdetermined bulk grafting ratios of PLL-g-PEG polymers correlate well with the PEG densities actually found at the surface. This correlation permits PLL-g-PEG to be synthesized and characterized ex situ, which is useful for quality control. The polymer can subsequently be applied to the substrate by a simple dip-coating process from an aqueous solution. Two different planar waveguide techniques demonstrate that for PLL-g-PEG with 2-kDa PEG chains, the optimum grafting ratios for DNA binding are found to be in the g ) 7-12 range. Different PEG chain lengths would presumably lead to systematically different optimum ranges although such experiments were outside the scope of this study. When such interfacial coatings are applied to DNA microarrays, neither covalent nor nonspecific blocking steps are required after the spotting, thus eliminating the risk of reactive degradation of the spotted DNA. Although the mechanism by which PEG reduces the background signal (so that reactive passivation is not required) is not yet understood, possible contributing factors include concentration and solvation effects as well as the stringency of the blanking and posthybridization washing steps. The resulting morphology of the hybridized cDNA spots is reproducible and columnar. By combining tuned PLL-g-PEG coatings with highly sensitive planar waveguide transducers, unamplified signal from medium- and low-expressed genes can be measured with a level of detectable differential expression of (1.5 when 1 µg pf total RNA equivalent is hybridized per microarray. ACKNOWLEDGMENT We thank Friederike Wilmer (Qiagen GmbH, Hilden, Germany) for providing the target mouse brain cDNA, Michael Horisberger (PSI, Villigen, Switzerland) for sputter-coating the OWLS optical waveguides, Paul Hug and Beat Keller (EMPA, Du¨bendorf, Switzerland) for providing support for the TOF-SIMS measurements, Janos Vo¨ro¨s (Laboratory for Surface Science and Technology, ETH Zu¨rich, Switzerland) for help with the refractometry measurements, Michael Pawlak (Zeptosens, Witterswil, Switzerland) and Nicholas Spencer (Laboratory for Surface Science and Technology, ETH Zu¨rich, Switzerland) for stimulating discussions, and the Swiss Commission for Technology and Innovation (KTI Project 5439.1 MTS) and the TOP NANO 21 Program (Project 4597.1) for financial support.

Received for review March 18, 2005. Accepted July 11, 2005. AC0504666