Detection of MicroRNA by Fluorescence Amplification Based on

Oct 15, 2009 - Small RNA molecules are effective regulators of gene expression, and the expression signature of one subgroup of small RNA, the microRN...
0 downloads 0 Views 264KB Size
Download PDF
Anal. Chem. 2009, 81, 9723–9729

Detection of MicroRNA by Fluorescence Amplification Based on Cation-Exchange in Nanocrystals Jishan Li,†,‡ Samantha Schachermeyer,† Yan Wang,† Yadong Yin,† and Wenwan Zhong*,† Department of Chemistry, University of California, Riverside, California 92521, and College of Chemistry and Chemical Engineering, State Key Laboratory for Chemo/Biosensing and Chemometrics, Hunan University, Changsha, 410082, P.R. China Small RNA molecules are effective regulators of gene expression, and the expression signature of one subgroup of small RNA, the microRNA (miRNA), has been linked to disease development and progression. Therefore, detection of small RNA in biological samples will greatly improve the understanding of their functions and render effective tools to researchers for cellular process control and disease prevention. To solve the challenges in detecting the low-abundance and short strand-length of small RNA molecules, we designed a ligation-assisted binding assay and applied the cation exchange-based fluorescence amplification (CXFluoAmp) method developed in our group for detection. Nonfluorescent, ionic nanocrystals (NCs) of CdSe were conjugated to detection probes and immobilized onto the array surface via ligation with the target small RNA, miR21, which bound to the capture probe complimentarily. Each binding event induced by one target miR21 molecule was then amplified by the release of thousands of Cd2+ from one NC. The free Cd2+ immediately turned on the fluorescence of thousands of fluorogenic Rhod-5N molecules. With such a powerful signal amplification strategy, our assay achieved a limit of detection (LOD) of 35 fM and signals were detectible with analyte concentrations spanning over 7 orders of magnitude. We also identified the differential expression of miR21 in total RNA extracts from healthy breast tissue and diseased cells. Furthermore, our detection scheme demonstrated good specificity in small RNA detection, because significant signal intensity could be observed from small RNAs with one or two nucleotides difference in sequences. Thus, our assay has great application potential for disease diagnosis relying on miRNA biomarkers, or in small RNA expression profiling for new target discovery and functional study. In recent years, microRNAs (miRNAs), one subgroup of the small RNA family, have come under intense investigation for their possible applications in cancer detection and typing. These short

noncoding RNA molecules act as potent, highly specific regulators of gene expression,1-3 and play roles in cell fate commitment and cell proliferation.2 Because abnormal cell proliferation is a hallmark of human cancers, researchers believe miRNA expression patterns could denote the malignant state.4-7 Indeed, altered expression of a few miRNAs has been found in some tumor types.4-7 The function of miRNA in cancer development and the potential for miRNA to be useful diagnostic and prognostic markers need to be systematically explored, but large progress in this field is hindered by difficulties in miRNA detection. Mature small RNAs contain only 17-25 nucleotides, and this short-length nature imposes great challenges in studies using conventional techniques. Northern blotting with radioactive labels is the most routine practice in small RNA discovery.8,9 Microarrays and quantitative PCR on small RNAs require either upfront small RNA enrichment or lengthy and technically demanding reaction processes.10-15 Biological enzymes have been employed in miRNA detection with detection limits in the femtomole range.16,17 Surface plasmon resonance (SPR) with hybridized gold nanoparticles was able to achieve attomole detection limit, but sophisticated imaging technique was involved.18 Thus, sensitive, specific, quantitative, (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

(15) (16)

* To whom correspondence should be addressed: Fax: +1-951-827-4713. E-mail: [email protected]. † University of California. ‡ Hunan University. 10.1021/ac901983s CCC: $40.75  2009 American Chemical Society Published on Web 10/15/2009

(17) (18)

Buchan, J. R.; Parker, R. Science 2007, 318, 1877–1878. Hagen, J. W.; Lai, E. C. Cell Cycle 2008, 7, 2327–2332. Kato, M.; Slack, F. J. Biol. Cell 2008, 100, 71–81. Calin, G. A.; Croce, C. M. Nat. Rev. Cancer 2006, 6, 857–866. Cummins, J. M.; Velculescu, V. E. Oncogene 2006, 25, 6220–6227. Tricoli, J. V.; Jacobson, J. W. Cancer Res. 2007, 67, 4553–4555. Waldman, S. A.; Terzic, A. Clin. Chem. 2008, 54, 943–944. Hamilton, A. J.; Baulcombe, D. C. Science 1999, 286, 950–952. Pall, G. S.; Codony-Servat, C.; Byrne, J.; Ritchie, L.; Hamilton, A. Nucleic Acids Res. 2007, 35, e60/61–e60/69. Miska, E. A.; Alvarez-Saavedra, E.; Townsend, M.; Yoshii, A.; Sestan, N.; Rakic, P.; Constantine-Paton, M.; Horvitz, H. R. Genome Biol. 2004, 5, r68. Tang, X.; Gal, J.; Zhuang, X.; Wang, W.; Zhu, H.; Tang, G. RNA 2007, 13, 1803–1822. Wark, A. W.; Lee, H. J.; Corn, R. M. Angew. Chem., Int. Ed. 2008, 47, 644–652. Yin, J. Q.; Zhao, R. C.; Morris, K. V. Trends Biotechnol. 2008, 26, 70–76. Chen, C.; Ridzon, D. A.; Broomer, A. J.; Zhou, Z.; Lee, D. H.; Nguyen, J. T.; Barbisin, M.; Xu, N. L.; Mahuvakar, V. R.; Andersen, M. R.; Lao, K. Q.; Livak, K. J.; Guegler, K. J. Nucleic Acids Res. 2005, 33, e179/171–e179/ 179. Raymond, C. K.; Roberts, B. S.; Garrett-Engele, P.; Lim, L. P.; Johnson, J. M. RNA 2005, 11, 1737–1744. Cissell, K. A.; Rahimi, Y.; Shrestha, S.; Hunt, E. A.; Deo, S. K. Anal. Chem. 2008, 80, 2319–2325. Su, X.; Teh, H. F.; Lieu, X.; Gao, Z. Anal. Chem. 2007, 79, 7192–7197. Fang, S.; Lee, H. J.; Wark, A. W.; Corn, R. M. J. Am. Chem. Soc. 2006, 128, 14044–14046.

Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

9723

Table 1. Sequences of RNA and DNA Probes Used in Our Study. Mutated Nucleotides in siRNA Were Presented in Bold name

sequence

miR21 capture probe (CP) detection probe (DP) small RNA (nat-siRNAATGB2) M1-siRNA M2-siRNA M3-siRNA CP-siRNA

5′- UAG CUU AUC AGA CUG AUG UUG A -3′ 5′-/5AmMC6/ TTT TTT TTT TTT A CTG TCT TAG CTT TCA ACA TCAGTC TGA TAA GCT AA-3′ 5′p AAG CTA AGA CAG TTT TTT TTT TTT TTT /3AmM/-3′ 5′-rArCrUrCrArArGrGrUrGrCrArGrCrUrGrGrArGrGrArA-3′ 5′-rArCrUrCrArArGrGrUrGrCrUrGrCrUrGrGrArGrGrArA- 3′ 5′- rArCrUrCrArArGrGrUrGrCrArArCrUrGrCrArGrGrArA -3′ 5′- rArCrUrCrArArGrCrUrGrCrArGrArUrGrGrArGrGrArA -3′ 5′-/5AmMC6/ TTT TTT TTT TTT A CTG TCT TAG CTT CTC CAG CTG CAC CTT GAG T AAG CTA AG-3′

rapid, and cost-effective methods for measuring the expression levels of small RNAs are in demand and could considerably advance the field. Nanoparticles are playing an increasing role in biosensing nowadays. In particular, quantum dots (QDs) emerge as prominent fluorescent labels for sensitive detection of low abundant targets.19,20 By fine-tuning the particle size and size distribution during synthesis, a large group of CdSe nanocrystals with diverse optical properties can be generated. Despite their phenomenal optical property, QDs have not found robust use in quantitative in vitro measurements. The occurrence of QD-specific photobrightening and presence of permanently nonfluorescent QDs within the same preparation prevent high quantification accuracy.21 Furthermore, Cd2+ imposes significant biological and environmental hazards. The preparation of highly fluorescent QDs made of nontoxic materials has not been successful.21-23 Another challenge involved with the applications of QDs is that their luminescent intensity and decay behavior could be largely altered by the introduction of hydrophilic surface functional groups and bioconjugation. Surface modification generates dangling atomic or molecular orbitals on the QD surface that quench the luminescent energy transfer.24 Alternatively, encapsulation of fluorophores inside sophisticated nanostructures has been developed to achieve ultrahigh sensitivity in detection by enhancing the dye-to-reporter molecule labeling ratio, but such approaches come with problems like bulky particle sizes, complicated structure design, and/or rather complex and laborious steps engaged in method development and assay.25-28 Ionic nanocrystals with hydration diameters of only a few nanometers encapsulate huge amounts of ions. Kucur et al. reported that there were 2051 Cd2+ ions in each CdSe nanocrystals (NCs) with an average diameter of 5.15 nm and every 1 nm increase (19) Ruan, G.; Agrawal, A.; Smith, A. M.; Gao, X.; Nie, S. Rev. Fluoresc. 2006, 3, 181–193. (20) Chan, W. C. W.; Maxwell, D. J.; Gao, X.; Bailey, R. E.; Han, M.; Nie, S. Curr. Opin. Biotechnol. 2002, 13, 40–46. (21) Resch-Genger, U.; Grabolle, M.; Cavaliere-Jaricot, S.; Nitschke, R.; Nann, T. Nat. Methods 2008, 5, 763–775. (22) Reiss, P.; Quemard, G.; Carayon, S.; Bleuse, J.; Chandezon, F.; Pron, A. Mater. Chem. Phys. 2004, 84. (23) Reiss, P. New J. Chem. 2007, 31, 1843–1852. (24) Smith, A. M.; Duan, H.; Mohs, A. M.; Nie, S. Adv. Drug Delivery Rev. 2008, 60, 1226–1240. (25) Yao, G.; Wang, L.; Wu, Y.; Smith, J.; Xu, J.; Zhao, W.; Lee, E.; Tan, W. Anal. Bioanal. Chem. 2006, 385, 518–524. (26) Thomas, R. N.; Guo, C. Y. Fresenius´ J. Anal. Chem. 2001, 369, 477–482. (27) Steinkamp, T.; Karst, U. Anal. Bioanal. Chem. 2004, 380, 24–30. (28) Cai, W.; Gentle, I. R.; Lu, G. Q.; Zhu, J.-J.; Yu, A. Anal. Chem. 2008, 80, 5401–5406.

9724

Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

in NC diameter represents 2200 more Cd2+ ions enclosed.29 The large numbers of encapsulated ions can be released by a fast and gentle process, the cation exchange reaction. Cation exchange has been employed to create nanoparticles with different compositions.30-34 For instance, it has been found that Ag2Se nanoparticles can be created by exchanging the Cd2+ cation from CdSe NCs with Ag+.30,32-34 Ag2Se is more thermodynamically favored compared with CdSe and Ag+ ions will therefore readily exchange out Cd2+ ions.32,34 We demonstrated that by combining the cation exchange process with the metalresponsive dyes, thousands of cations could be released from each nanocrystal attached to one target molecule and trigger strong fluorescence to obtain hundreds folds enhancement in signal intensity.35 Such a cation exchange-based fluorescence amplification (CXFluoAmp) scheme is a general, simple, and more importantly, highly sensitive detection method for planar arrays, and should be of great value in small RNA detection. Herein, we reported the successful application of this signal amplification strategy to identify miRNA in biological samples. We designed a capture probe (CP) with a molecule beacon structure that would open up and expose the binding site for the detection strand (DP) upon hybridization of target miRNA. The CP was conjugated onto the magnetic beads and the DP to the nonfluorescent CdSe nanocrystals (NCs). The NCs would then be immobilized onto the magnetic beads in the presence of target miRNA (miR21). Afterwards Cd2+ ions are released to react with the Rhod-5N dye to produce intense fluorescent signals. We were able to clearly detect as low as 100 fM target miRNA in standard solution, and the 3-STD limit of detection (LOD) obtained from the calibration curve was 35 fM. The superior assay performance led to clear identification of the expression difference of miR21 in normal and cancer cell lines. EXPERIMENTAL SECTION Reagents and Materials. Chemicals used to prepare assay buffers, metal ion solutions, or conjugations were all from Fisher (29) Kucur, E.; Boldt, F. M.; Cavaliere-Jaricot, S.; Ziegler, J.; Nann, T. Anal. Chem. 2007, 79, 8987–8993. (30) Camargo, P. H. C.; Lee, Y. H.; Jeong, U.; Zou, Z.; Xia, Y. Langmuir 2007, 23, 2985–2992. (31) Dong, C.; van Veggel, F. C. J. M. ACS Nano 2009, 3, 123–130. (32) Son, D. H.; Hughes, S. M.; Yin, Y.; Paul Alivisatos, A. Science 2004, 306, 1009–1012. (33) Sadtler, B.; Demchenko, D. O.; Zheng, H.; Hughes, S. M.; Merkle, M. G.; Dahmen, U.; Wang, L.-W.; Alivisatos, A. P. J. Am. Chem. Soc. 2009, 131, 5285–5293. (34) Chan, E. M.; Marcus, M. A.; Fakra, S.; ElNaggar, M.; Mathies, R. A.; Alivisatos, A. P. J. Phys. Chem. A 2007, 111, 12210–12215. (35) Li, J.; Zhang, T.; Ge, J.; Yin, Y.; Zhong, W. Angew. Chem., Int. Ed. 2009, 48, 1588–1591.

Scientific. Small RNA and other oligonucleotides (Table 1) were obtained from Integrated DNA Technologies (Coralville, IA) with no further purification except standard desalting. Rhod-5N was purchased from Invitrogen (Carlsbad, CA). T4 DNA ligase was obtained from New England Biolabs (Ipswich, MA). Carboxyl magnetic particles (5% w/v) with a nominal diameter of 4.35 µm were acquired from Spherotech Inc. (Lake Forest, IL). First Choice Human Breast Adenocarcinoma (MCR-7) and First Choice Breast total RNA (BTR) were obtained from Ambion (Austin, TX). The water-soluble CdSe nanocrystals were produced through a high temperature ligand exchange procedure,36 and the original fluorescence of the CdSe particles disappeared during ligand exchange with poly(acrylic acid) (PAA) due to the creation of surface defect sites.35 CXFluoAmp Using Rhod-5N. Typically, the CXFluoAmp solution contained 500 µM AgNO3 and 10 µM Rhod-5N in the background buffer of 0.1 M potassium acetate (KAc)-0.05% Tween 20. This solution was mixed with the CdSe NCs suspension to trigger intense fluorescence. To compare the response of Rhod-5N to different metal ions, standard solutions of Ca2+, Mg2+, Zn2+, Cu2+, Ni2+, Pb2+, Hg2+, and Cd2+ were prepared at 10 µM, except Ag+, which was prepared at 500 µM. Upon the addition of 2 µM Rhod-5N to each 100 µL metal ion solution, fluorescence was measured in the Spex FluoroLog Tau-3 fluorescence spectrophotometer (HORIBA Jobin Yvon Inc., NJ), using the excitation wavelength of 530 nm. Emission spectrum from 570 to 650 nm was collected with a slit width of 2 nm. Conjugation of Magnetic Particles and CdSe NCs with Oligonucleotides. The capture probe carried an amine group at its 5′ end and was coupled to the carboxyl magnetic particle by 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide hydrochloride (EDC) and sulfo-N-hydroxysuccinimide (sulfo-NHS). We mixed 100 µL magnetic particle suspension (5% w/v), 5 nmol CP, 30 mg EDC, and 5 mg sulfo-NHS in 1 mL 0.1 M NaH2PO4-Na2HPO4 (pH 7.3) buffer, and vortexed the mixture at 1× speed for 4 h at room temperature. The same phosphate buffer was employed for both the postconjugation cleanup and storage of the CPcoupled magnetic particles (M-CPs). Coupling of the detection probe to the carboxyl group of the PAA-coated CdSe NCs were achieved using 100 µL CdSe suspension (approximately 1013 NCs) and 10 nmol DP. The amount of Cd in NCs was measured by inductively coupled plasma atomic emission spectrometer (ICP-AES) after dissolving the NCs in HNO3, and the molar concentration of NC suspension was estimated from the particle size observed under TEM (∼8 nm), taking into account the volume of an ideal sphere and the CdSe bulk density of 5.81 g/cm3. After conjugation, the free DP strands were removed by centrifuging the coupling solution in a Microcon YM-100 centrifugal filter (Millipore Corporate, Billerica, MA) at 3300g for 20 min. The NCs conjugated with DP (CdSe-DP) were washed once with 0.1 M KAc (pH 8.0) solution using the same filtration device before being transferred to the same KAc buffer for storage at 4 °C. The conjugation efficiency was calculated by measuring the DP solution concentrations before and after conjugation with UV (36) Zhang, T.; Ge, J.; Hu, Y.; Yin, Y. Nano Lett. 2007, 7, 3203–3207.

absorption at 254 nm. Each NC could be conjugated to averagely three DP probes. Detection of miRNA by CXFluoAmp. Detection of miR21 was carried out with a 3-step assay, including hybridization, ligation, and detection. In general, 5 µL M-CP was mixed with RNA samples (various amounts of miR21 or 2 µg total RNA extracts) in 100 µL 1× T4 ligase buffer containing 0.1 M NaCl. Hybridization between the RNA strand and CP was initiated by heating the mixture at 95 °C for 2 min, followed with 1 h incubation at 55 °C. Subsequently, we collected the particles with a magnet, discarded the supernatant, and suspended the particles with another aliquot of 100 µL 1× T4 ligase buffer. Then, 2 µL CdSe-DP (∼2 × 1011 NCs) as well as 20 U T4 DNA ligase was applied and the magnetic particle solution was incubated at 37 °C for 2 h to link the RNA to the DP via ligation. Next, particles were washed twice in the 1X SSC buffer (15 mM trisodium citrate (pH ∼7-8) with 150 mM NaCl and 0.1% SDS) and once in 0.1 M KAc-0.05% Tween 20 (pH 7.4) at 37 °C. During each wash, particles were vortexed gently for 5 min to enhance the washing stringency. Finally, 100 µL detection solution was added, containing 0.1 M KAc (pH 7.4), 0.05% Tween 20, 500 µM AgNO3, and 2 µM Rhod-5N. The fluorescence was measured after the removal of magnetic particles. Optimization of the hybridization and ligation duration was performed by holding all other experimental conditions constant as described above and varying the corresponding duration only. Similar assay procedure was applied in the specificity study using a different strand of small RNA, the nat-siRNAATGB2 discovered in plants,37 as the target. Detection of the wild-type strand and three arbitrary 1- or 2-nucleotide mutations were performed using the CP designed for the wild-type strand. To deal with the high GC-content (54%) of these strands, at the end of hybridization step we heated up the sample to 70 °C for 5 min and immediately put it on ice, aiming to achieve better specificity. Melting analysis was carried out in a Varian Cary 50 UV/vis spectrophotometer to determine the appropriate hybridization temperatures in our assay. Hybridization was prepared by mixing 500 nM miR21 or nat-siRNAATGB2 strands with the corresponding CP at 1:1 ratio, incubating the mixture at 95 °C for 2 min. Following 4-h hybridization at 37 °C, the absorbance at 260 nm of the RNA-DNA hybrid was measured at 10 or 5 °C intervals as the temperature rose from 25 - 80 °C. Temperature control was realized by a water bath (Isotemp Refrigerated Circulator model 900, Fisher Scientific). In order to avoid RNA degradation, extreme precaution was taken in dealing with the RNA samples. All RNA samples were split into small aliquots and stored in at -80 °C upon arrival. All hybridization and ligation reactions were performed in a Biological Safety Cabinet (Forma Class II, A2, Fisher). The surface inside the cabinet was cleaned up regularly with Eliminase (Fisher), a cleaning reagent to remove DNase, RNase, and DNA contaminants, and was wiped with 75% alcohol solution on a daily basis. All biological reaction buffers were prepared with the DNase, RNase free water from Sigma (filtrated with 0.1 µm filter). (37) Katiyar-Agarwal, S.; Morgan, R.; Dahlbeck, D.; Borsani, O.; Villegas, A., Jr.; Zhu, J.-K.; Staskawicz, B. J.; Jin, H. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 18002–18007.

Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

9725

Figure 1. Fluorescence spectra of Rhod-5N’s response to a variety of metals to show the specificity the dye has to Cd2+. All metals were tested at concentrations of 10 µM except Ag+ at 500 µM.

RESULTS AND DISCUSSION Response of Rhod-5N to Various Metals. One prominent feature of our CXFluoAmp design is its flexibility in material selection. Not only nanocrystals consisted of different metal elements can be selected to avoid the toxicity concern of Cd2+, but also a variety of metal-responsive fluorescent sensors with distinctive optical characteristics and metal affinities are available to provide possibilities of parallel labeling and improved signal-to-noise level.38 We mainly used Fluo-4 in our proof-ofprinciple study. Since Fluo-4 has a dissociation constant (Kd) for Ca2+ of 345 nM, its usage required tedious water treatment with Chelex beads before preparing the detection solutions in addition to the employment of a masking reagent, ethylene glycol tetraacetic acid (EGTA), in the buffer to shield the residue Ca2+. By replacing Fluo-4 with Rhod-5N, another calcium sensor,39,40 a 10-fold fluorescence enhancement was achieved with the CdSe NCs. The larger signal enhancement factor can be attributed to the low binding affinity (320 µM) of Rhod-5N to Ca2+ that substantially reduces the interference from Ca2+. Neither Chelex treatment nor EGTA was needed with Rhod-5N and the detection scheme was significantly simplified. Other than Ca2+, we may find other metal ions, such as Zn2+ and Mg2+, in solutions prepared with chemicals of analyticalgrade purity. To confirm the better selectivity of Rhod-5N to Cd2+ over possible interfering ions, the fluorescence response of Rhod-5N to various metal ions (Figure 1) was tested. All metal cations were added at 10 µM, except Ag+, which was prepared at 500 µM, the same [Ag+] as in the CXFluoAmp detection buffer. The fluorescence at 576 nm, the peak emission of Rhod5N, from 10 µM divalent metal cations were divided by that from 500 µM Ag+ to obtain the signal-to-“noise” (S/N) ratios. Cd2+ resulted in the highest S/N of 98, while Hg2+, Pb2+, and Zn2+ had S/Ns of 65, 29, and 5, respectively. Hg and Pb are very unlikely to be found in normal buffer solutions, and impose little concern in our applications. Ca2+, Mg2+, Ni2+, and Cu2+ (38) Domaille, D. W.; Que, E. L.; Chang, C. J. Nat. Chem. Biol. 2008, 4, 168– 175. (39) Ribou, A.-C.; Salmon, J.-M.; Vigo, J.; Goyet, C. Talanta 2007, 71, 437– 442. (40) McNamara, C. J.; Perry, T. D.; Bearce, K.; Hernandez-Duque, G.; Mitchell, R. J. Microbiol. Methods 2005, 61, 245–250.

9726

Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

all induced negligible fluorescence from Rhod-5N with an S/N value below 3. This result pointed out that Rhod-5N should be a superior sensor for the released Cd2+ and could result in high S/N in detection. Therefore, we employed Rhod-5N as the fluorophore in the present study. Assay Design. Figure 2 illustrates the design we employed to detect small RNA in samples. Because magnetic particles are easy to handle with a magnet, they were employed as the solid support for our assay. The molecular beacon structure of CP would open up upon the hybridization of target nucleic acid and expose a binding site for DP that is coupled to the CdSe NCs. However, the short length of small RNA is always a concern for hybridization-based detection assays, because the binding efficiency could be low and the double-stranded structure may not sustain harsh wash conditions that are applied to remove nonspecific bindings. Therefore, we introduced a ligation step to enhance the binding stability and enable washing at a relatively high temperature.41 The small RNA molecules contain a free 3′-OH group naturally. Each DP is then designed to contain an amine group at its 3′ end for conjugation purpose as well as a phosphate group at the 5′ end for ligation. The joint molecule would have a total length of 34 nt and should remain bound onto the CP even during stringent washing. After several washes, the detection solution would then be applied to the magnetic particles to release the high Cd2+ content inside NCs and trigger intense fluorescence. Each binding event causes the capture of one NC which could release thousands metal ions via the cation exchange reaction for signal amplification. Ligation provides several advantages to our assay. It stabilizes the binding of target small RNA to CP, and allows the application of elevated temperatures, i.e., 37 °C, during wash to eliminate the nonspecific binding from other short RNA or DNA strands. Moreover, the ligation could increase the specificity of our detection in two ways. One is from the high fidelity of DNA ligase to perfect duplex structures. Extremely low ligation yield is normally obtained at the presence of mismatched sequences at positions close to the ligation junction.42,43 The other is from the free 3′-OH that abolishes the wrongly detection of nucleic acids with similar sequences but modified or overhang 3′ terminus. Our experimental results also pointed out the necessity of ligation. By comparing the detection of 1 µM miR21 with and without ligation, we found out that only a small fluorescence increase from the blank signal (no miR21) was observed if no T4 DNA ligase was supplied, while the fluorescence intensity with ligation was very strong (Figure 3). Thus, because of its short length, the target small RNA could not induce stable structure change in CP and keep the DP-modified CdSe NCs on the array surface for detection if a linkage was not formed between the DP and small RNA molecule. Assay Optimization. Even though affinity capture of target nucleic acids by surface-immobilized probes is a routine assay format for DNA/RNA expression profiling, these assays suffer from numerous performance limitations, including sensitivity and (41) Xue, X.; Xu, W.; Wang, F.; Liu, X. J. Am. Chem. Soc. 2009, 131, 11668– 11669. (42) Liu, P.; Burdzy, A.; Sowers, L. C. Nucleic Acids Res. 2004, 32, 4503–4511. (43) Housby, J. N.; Southern, E. M. Nucleic Acids Res. 1998, 26, 4259–4266.

Figure 2. Schematic presentation of the small RNA detection assay using CXFluoAmp. MP: magnetic particles; CP: capture probe; DP: detection probe.

Figure 3. Fluorescence spectra showing the higher signal obtained with the addition of a ligation step. The blank sample contained no miR21 and 20 U T4 DNA ligase, the sample without ligation had 1 µM miR21 but no T4 DNA ligase, and the sample with ligation contained 1 µM miR21 and 20 U T4 DNA ligase.

rapidity.44-46 CXFluoAmp promises ultrahigh detection signals, but fundamentally the sensitivity and detection limits of such assays are still controlled by how well they can maintain fast binding kinetics and high binding efficiency while minimizing nonselective hybridization. Hybridization temperature and duration as well as washing stringency are main parameters tunable to obtain optimal detection performance. We first looked into the hybridization temperature. The universal criterion is to attain equilibrium reaction condition which can help to reduce the contribution of mismatch species.44,46 Performing hybridization at temperatures around the melting point of the duplex structure should allow reaching equilibrium rapidly. Thus, we analyzed the melting curve of the hybrid between miR21 and CP. Even with a 10 °C interval, we could clearly see that the (44) Blair, S.; Williams, L.; Bishop, J.; Chagovetz, A. Methods Mol. Biol. 2009, 529, 171–196. (45) Petersen, J.; Poulsen, L.; Petronis, S.; Birgens, H.; Dufva, M. Nucleic Acids Res. 2008, 36, e10/11–e10/10. (46) Erickson, D.; Li, D.; Krull, U. J. Anal. Biochem. 2003, 317, 186–200.

miR21-CP hybrid started to denature at 50 °C (Supporting Information (SI) Figure S1a). The fluorescent signals from 2 µM miR21 obtained under different hybridization temperatures but the same, 2-h duration also confirmed that the highest signal was achieved at 45 or 55 °C (SI Figure S1b). We adopted the higher temperature of 55 °C in the following studies so that equilibrium condition could be reached in a faster manner to ensure good selectivity. Then, we moved on to optimize the hybridization and ligation durations. It was found that the fluorescent signal increased substantially when the hybridization duration extended from 30 min to 1 h. The hybridization seemed to be established after one hour because longer hybridization time of 2 or 3 h did not show further signal enhancement (SI Figure S2a). One hour was chosen to be the optimal hybridization duration. Because ligation yield directly determines the amount of target small RNA and thus that of DP staying on the array surface for detection, the signal continued to increase with longer ligation time (SI Figure S2b). In order to keep the total assay time within a reasonable range, we settled the ligation duration at 2 h. Additionally, we performed washing at 37 °C after ligation to further reduce the nonspecific binding. The high stringency of thermal washing has been demonstrated in microarrays to simplify probe design and enhance the performance of the arrays.45 Detection Range and Limit. Next, we evaluated the detection capability of our assay. Because of the high amplification efficiency of CXFluoAmp, we envisioned that our assay could be superior to the existing methods in terms of wider detection range and lower limit of detection (LOD). A serial standard miR21 solution with concentrations of 0, 100 fM, 1 pM, 10 pM, 100 pM, 1 nM, 10 nM, 100 nM, 1 µM, and 5 µM were prepared and mixed with the CP-coupled magnetic particles. Following the 1 h hybridization and 2 h ligation steps, the beads were washed three times before the detection solution was applied, which contained the cation exchange reagent of Ag+ and the fluorogenic dye of Rhod-5N. We could see the fluorescence rose rapidly from the blank with only 100 fM miRNA target, exponentially increased over 7 orders of magnitude analyte concentration, and leveled off Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

9727

Figure 4. (a) Calibration curve of miR21 detection. Fluorescence intensity was measured at 576 nm upon the addition of 0-10 µM miR21. The LOD was determined to be 35 fM, the concentration corresponding to the fluorescence intensity equal to blank ( 3STD. (b) Fluorescence intensity at 576 nm obtained from samples containing 2 µg of either Breast Total RNA (BTR) or FirstChoice Human Breast Adenocarcinoma (MCF-7) Total RNA.

beyond 1 µM (Figure 4a and SI FS3). The 3-STD LOD read from the curve was 35 fM (3.5 attomole in 100 µL reaction volume), which is lower than that with quantum dots47 or biological enzymes16 as the signaling probes and comparable to that with PCR amplification15 or gold nanoparticle-amplified surface plasmon resonance.18 Even though the fluorescence signal is linearly proportional to the nanocrystal molar concentration in our amplification scheme,35 the calibration curve for miRNA detection was not linear. This is probably due to the not-at-equilibrium hybridization and ligation conditions. Further investigation is needed to produce linear response curve with our method. Still, we reached low detection limit and signals were detectible within a large analyte concentration range. Both could be attributed to the high signal intensity of CXFluoAmp, as well as the immunity of CXFluoAmp to fluorescence quenching which originates from triggering the fluorescence by coordination of the released cations with fluorogenic dyes in bulk solution instead of a packed space. Since miR21 is a potential cancer biomarker which has been identified with elevated expression levels in numerous tumor tissues, including breast, liver, ovarian, pancreatic, and brain, in comparison to their normal counterparts, we tested the effective(47) Liang, R.-Q.; Li, W.; Li, Y.; Tan, C.-y.; Li, J.-X.; Jin, Y.-X.; Ruan, K.-C. Nucleic Acids Res. 2005, 33, e17.

9728

Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

ness of our assay in analysis of real biological samples by measuring the levels of miR21 in total RNA extracts prepared from normal human tissue or carcinoma cell line (Figure 4b). Instead of standard miR21 solutions, 2 µg of the FirstChoice Human Breast Total RNA (BTR, extracted from normal breast tissue) or FirstChoice Human Breast Adenocarcinoma (MCF-7) Total RNA was mixed with the CP-magnetic particles and the assay was carried out following the exact same procedure as for the calibration curve. It turned out that the MCF-7 total RNA generated a fluorescent signal (17098 ± 2581) about 3.9 times higher than that obtained from the normal breast tissue (4382 ± 880), indicating higher expression level of miR21 in the MCF-7 cell line. The overexpression of miR21 in MCF-7 has been observed previously using complicated procedures like RT-PCR or enzymatic amplification.16,48 Here, with the assistance of CXFluoAmp, a simple affinity array can be used to measure the expression levels of miRNA in biological samples and has the potential for rapid disease diagnosis. Because the large amount of background RNA molecules had been removed by wash before applying the CXFluoAmp solution, no interference was observed from the complex background to the detection performance of CXFluoAmp. This result well demonstrates the high compatibility of CXFluoAmp with sample matrixes of high complexity. Specificity Study. Specificity is another essential factor to evaluate the effectiveness of an affinity assay in analyte detection. Most of the discovered small RNAs, especially the miRNAs, differ by four or more nucleotide bases.10 Even though the 55 °C hybridization temperature chosen in our assay was determined by the melting temperature of miR21, we believe it should bring in sufficient specificity in miRNAs profiling in biological samples in combination with ligation and thermal washing. On the other hand, few small RNA families having members with only one or two nucleotides difference in their sequences were discovered.49,50 For the purpose of new small RNA discovery and study of the relationship between functions and molecular structure, it is essential that the assays possess the capability of discriminating even single-nucleotide mutation. Therefore, we also studied the specificity of our assay. Since miR21 has no close mutations identified, we employed a small RNA strand that had been investigated before by our group in this study. This particular small RNA was discovered in Arabidopsis thaliana as a pathogeninducible small-interfering RNA, and may be involved with the antibacterial function of plants. It represents more challenging targets than miR21 for hybridization-based assays because it has a GC-content of 54%, much higher than the 36% of miR21, and its single-nucleotide mutants would have high melting temperatures and thus high binding affinity to the capture probes. We tested three arbitrary mutations containing up to two nucleotides replacements in the wild-type strand sequences (Table 1), and the melting temperatures were measured to be 72.8 °C (wild-type), 68.7 °C (M1), 58.0 °C (M2), and 59.6 °C (M3) (SI Figure S4). M1, the single-nucleotide mutant, has a melting temperature over 68 °C, which may impose difficulty to our assay to clearly distinct (48) Lu, J.; Getz, G.; Miska, E. A.; Alvarez-Saavedra, E.; Lamb, J.; Peck, D.; SweetCordero, A.; Ebert, B. L.; Mak, R. H.; Ferrando, A. A.; Downing, J. R.; Jacks, T.; Horvitz, H. R.; Golub, T. R. Nature 2005, 435, 834–838. (49) Roush, S.; Slack Frank, J. Trends Cell Biol. 2008, 18, 505–516. (50) Torrisani, J.; Parmentier, L.; Buscail, L.; Cordelier, P. Curr. Genomics 2007, 8, 229–233.

Figure 5. The fluorescence intensity at 576 nm was taken to show the specificity of our assay in discriminating small RNA strands with one or two nucleotide mutations. M1- siRNA contains 1 nucleotide mutation, while M3- and M2-siRNA contain two nucleotide mutations, all resulting in signals much lower than that from the target.

it from the target. Hence, we revised the assay design and added a posthybridization step before ligation in which the sample was heated at 70 °C for 5 min and immediately put on ice. Significant difference in signal intensity was observed among the strands (Figure 5). The average (n ) 3) fluorescence intensity of M1 was only 21% of that of the wild-type, and M2 and M3 only yielded signals of 3.0% and 7.0%, respectively, of that of the wild type, when 1 µM of each strand were mixed with the magnetic particles conjugated to the CP designed for the wild-type siRNA. This result indicated that the hybridization step, particularly the temperature, could be altered to enhance specificity of detection and distinguish one or few nucleotides mutation. Requiring great caution in probe design and hybridization condition selection is a common challenge in microarrays where thousands of analytes are targeted simultaneously. However, with the high sensitivity of CXFluoAmp, it should be less ambiguous to discriminate different mutations from slight difference in signal intensity and obtain high confidence in identification of small RNA sequence. CONCLUSIONS We successfully developed an affinity-based assay to detect low abundant small RNA in total RNA extracts using CXFluoAmp.

The fast and gentle cation exchange reaction released thousands of cations from each NC to illuminate thousands of metalresponsive dyes and increased the signal intensity dramatically. The bright signal of CXFluoAmp resulted in an LOD of 35 fM, which, along with its free-from-photoquenching characteristics, also delivered a detectible concentration range spanning over 7 orders of magnitude. A ligation reaction increased the binding stability of the short small RNA and thus the DP-coupled NCs onto the capture probe, utilizing the free 3′-OH of small RNA. In addition, the longer ligation product sustained thermal wash for the reduction of nonspecific binding. Besides ligation and elevated washing temperature, single nucleotide polymorphism in challenging targets, i.e., those with high GC contents, could be identified with short duration of additional posthybridization heating. Our assay should be valuable to cancer diagnostics and prognostics based on miRNA signatures or to discovery and functional study of new small RNA strands. Even though we employed CdSe in the present study, our amplification scheme is highly flexible in selection of NCs. We are currently investigating to replace CdSe with the safer material of ZnSe in amplification to improve the biocompatibility of our assay. Furthermore, stability in long-term storage and ion leakage from NCs should be tested in order to enhance the applicability of our amplification scheme. Since our detection technique does not rely on the intrinsic fluorescence property of the NCs, we have higher freedom to select the most effective surface coating materials for better stability and structure integrity, without concerns of surface defects generated from modification. ACKNOWLEDGMENT We are grateful to Department of Chemistry, UCR, for financial support. The work was also supported by the Institute for Integrative Genome Biology (IIGB) Research Award from UCR. SUPPORTING INFORMATION AVAILABLE Figures S1-S3. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review September 2, 2009. Accepted October 3, 2009. AC901983S

Analytical Chemistry, Vol. 81, No. 23, December 1, 2009

9729