Anal. Chem. 2005, 77, 7195-7203
Species-Specific Identification of Mycobacterial 16S rRNA PCR Amplicons Using Smart Probes Katharina Sto 1 hr,† Bernhard Ha 1 fner,† Oliver Nolte,‡ Ju 1 rgen Wolfrum,† Markus Sauer,*,§ and Dirk-Peter Herten*,†
Physikalisch-Chemisches Institut, Universita¨t Heidelberg, Im Neuenheimer Feld 253, D-69120 Heidelberg, Germany, Fachbereich Biologie, Abt. Mikrobiologie, Technische Universita¨t Kaiserslautern, Paul-Ehrlich-Strasse, Geba¨ude 23, 67663 Kaiserslautern, Germany, and Applied Laser Physics and Laser Spectroscopy, University of Bielefeld, Universitaetsstrasse 25, 33615 Bielefeld, Germany
Due to growing problems with new emerging pathogens, cost-effective and manageable methods for their accurate identification in routine diagnostics are urgently required. Of particular importance is the genus Mycobacterium with its more than 100 species. Identification of these species is hampered by their slow growth in the laboratory and by the obligate need for DNA sequence analysis. To provide a fast and reliable diagnostic tool, we developed a novel approach using fluorescently labeled DNA hairpin structures (smart probes) for selective and sensitive detection of mycobacterial 16S rDNA PCR amplicons in homogeneous and heterogeneous assays. Smart probes are singly labeled hairpin-shaped oligonucleotides bearing a fluorescent dye at the 5′-end, which is quenched by guanosine residues in the complementary stem. Upon hybridization to target sequences, a conformational change occurs reflected in an increase in fluorescence intensity. Using optimized parameters for hybridization experiments we established a reliable method for the specific detection of Mycobacterium tuberculosis (M. tuberculosis complex) and Mycobacterium xenopi (member of the atypical mycobacteria) with a detection sensitivity of ∼2 × 10-8 M in homogeneous solution. The specificity of the smart probes designed is demonstrated by discrimination of M. tuberculosis and M. xenopi against 15 of the most frequently isolated mycobacterial species in a single assay. In combination with a microsphere-based heterogeneous assay format, the technique opens new avenues for the detection of pathogen-specific DNA sequences with hitherto unsurpassed sensitivity. Although a plethora of molecular methods is available for diagnostic purposes, accurate identification of pathogenic microorganisms is still cumbersome. Traditional culture-based routine diagnostics in medical microbiology, introduced by Robert Koch at the end of the 19th century,1 is at the edge of dramatic changes. * To whom correspondence should be addressed. E-mail: sauer@ physik.uni-bielefeld.de. Fax: +49-521-106-2958. E-mail: dirk-peter.herten@ urz.uni-hd.de. Fax: +49-6221-544255. † Universita¨t Heidelberg. ‡ Technische Universita¨t Kaiserslautern. § University of Bielefeld. (1) Ligon, B. L. Semin. Pediatr. Infect. Dis. 2002, 13, 289-299. 10.1021/ac051447z CCC: $30.25 Published on Web 10/12/2005
© 2005 American Chemical Society
Novel technologies, using the availability of species-specific DNA sequences, are under development and will cause a technological revolution in the field of diagnostics. However, high cost and comparatively complex methodology still limit the use of molecular methods outside the developed countries. So-called nontuberculous mycobacteria (NTM, or mycobacteria other than tuberculosis) have long been in the shadow of the most prominent genus member, Mycobacterium tuberculosis. On the other hand, with the rise of the acquired immunodeficiency syndrome (AIDS) and the availability of molecular methods, NTM infections become more important, e.g., infections with Mycobacterium avium complex or Mycobacterium xenopi.2-4 Currently ∼115 species of NTM have been validly described,5 indicating that novel opportunistic mycobacterial species will continue to be identified in the following years.3 Unlike the M. tuberculosis complex NTM are not obligate pathogens, at least in the immunocompetent host. In the laboratory, many mycobacteria render isolation and identification of the organisms time-consuming because of their slow growth rate. Typically, several weeks are required to complete routine diagnostic service when mycobacteria are isolated. Common standard procedures differ from laboratory to laboratory slightly, but all include growth on solid and in liquid media. Grown mycobacteria are identified using a set of different methods. Bacteria are used for biochemical profiling (typically nitrate reduction and niacin synthesis), culture-based susceptibility testing using the four or five most important tuberculostatic drugs, and DNA extraction. DNA is used for amplification of M. tuberculosis-specific genes (typically gyrase (gyrB) gene fragments or insertion element 6110 (IS6110)). If the isolated species is not of the M. tuberculosis complex, further tests are required. Besides sequence analysis, further biochemical and culture-based tests may by necessary (e.g., proof for catalase activity, Tween hydrolysis, growth at lower or higher temperatures, pigmentation of the colonies under daylight and in the dark). (2) Park, H.; Jang, H.; Kim, B.; Chung, B.; Chang, C. L.; Park, S. K.; Song, S. J. Clin. Microbiol. 2000, 38, 4080-4085. (3) Primm, T. P.; Lucero, C. A.; Falkinham, J. O. Clin. Microbiol. Rev. 2004, 17, 98-106. (4) Martin-Casabona, N.; Bahrmand, A. R.; Bennedsen, J.; Østergaard Thomsen, V.; Curcio, M.; et al. Int. J. Tuberculosis Lung Dis. 2004, 8, 1186-1193. (5) Euzeby, J. P. List of bacterial names with standing in nomenclature (http://www.bacterio.cict.fr/m/mycobacterium.html).
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005 7195
To accomplish fast and accurate differentiation of mycobacteria, several molecular methods have been developed in the past 10 years.6 Molecular methods provide two primary advantages when compared to phenotypic, culture-based identification: a more rapid turnaround time and improved accuracy in identification.7 The most widely accepted gene used for bacterial identification is the 16S rRNA gene (16S rDNA, ∼1550 bp), which is universal in bacteria and shows a high degree of conservation due to its importance as a critical component of cell function.8,9 The 16S rRNA gene is ideally suited for mycobacterial identification as it contains both conserved sequence regions and flanking highly variable regions. For detection and quantification of specific DNA and RNA sequences based on fluorescence measurements, numerous methods have been developed.10-16 Tyagi and Kramer introduced hairpin-shaped oligonucleotides that exhibit a strong increase in fluorescence intensity in the presence of a specific target sequence.11 These so-called molecular beacons are doubly labeled single-stranded nucleic acid molecules that possess a stem-loop (hairpin) structure. When the hairpin probe encounters a target molecule, it forms a longer and more stable hybrid than that formed by the arm sequences. Consequently, the hairpin undergoes a conformational change accompanied by a strong increase in fluorescence intensity. Since molecular beacons are very effective at detecting single-base mismatches, they hold great promise for future genetic studies.11,17-19 However, molecular beacons have important limitations. First, site-specific labeling of both termini with different chromophores or a chromophore and quenching moiety is difficult and still relatively expensive. Second, since the two termini of the DNA hairpin are already occupied by the chromophores, any further modification, for example to attach it to a solid support via biotin/streptavidin binding, requires the incorporation of an additional modified nucleotide into the stem. Thus, the stability of the DNA hairpin structure might deteriorate. Third, if the oligonucleotide is only labeled with the donor chromophore due to ineffective coupling, highly sensitive assays are interfered with by a high background due to unquenched probe molecules. Recently, an alternative technique for specific detection of target sequences in homogeneous assays was introduced. It takes advantage of the fact that several chromophores are selectively (6) Soini, H.; Musser, J. M. Clin. Chem. 2001, 47, 809-814. (7) Harmsen, D.; Dostal, S.; Roth, A.; Niemann, S.; Rothga¨nger, J.; Sammeth, M.; et al. BMC Infect. Dis. 2003, 3, 26-36. (8) Clarridge, J. E., III. Clin. Microbiol. Rev. 2004, 17, 840-862. (9) Turenne, C. Y.; Tschetter, L.; Wolfe, J.; Kabani, A. J. Clin. Microbiol. 2001, 39, 3637-3648. (10) Millar, D. P. Curr. Opin. Struct. Biol. 1996, 6, 322-326. (11) Tyagi, S.; Kramer, F. R. Nat. Biotechnol. 1996, 14, 303-308. (12) Fang, X.; Liu, X.; Schuster, S.; Tan, W. J. Am. Chem. Soc. 1999, 121, 29212922. (13) Frutos, A. G.; Pal, S.; Quesada, M.; Lahiri, J. J. Am. Chem. Soc. 2002, 124, 2396-2397. (14) Knemeyer, J. P.; Marme´, N.; Sauer, M. Anal. Chem. 2000, 72, 3717-3724. (15) Marras, S. A. E.; Kramer, F. R.; Tyagi, S. Genet. Anal. 1999, 14, 151-156. (16) Darby, R. A. J.; Sollogoub, M.; McKeen, C.; Brown, L.; Risitano, A.; Brown, N.; Barton, C.; Brown, T.; Fox, K. R. Nucleic Acids Res. 2005, 33, e13. (17) Giesendorf, B. A. J.; Vet, J. A. M.; Tyagi, S.; Mensink, E. J. M. G.; Trijbels, F. J. M.; Blom, H. J. Clin. Chem. 1998, 44, 482-486. (18) Hodgson, D. R.; Foy, C. A.; Partridge, M.; Pateromichelakis, S.; Gibson, N. J. Mol. Med. 2002, 8, 227-237. (19) El-Hajj, H. H.; Marras, S. A. E.; Tyagi, A.; Kramer, F. R.; Alland, D. J. Clin. Microbiol. 2001, 39, 4131-4137.
7196 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
Figure 1. Schematic of operation of smart probes. The fluorophore is attached to the 5′-end of the oligonucleotide and quenched by guanosine residues in the complementary stem via photoinduced intramolecular electron transfer. Upon hybridization to the target sequence (complementary to the loop sequence), the fluorescence is restored because of a conformational reorganization that forces the stem apart.
quenched by the DNA base guanine.14,20,21 The method uses the differences in specific properties of naturally occurring nucleotides, in particular, the low oxidation potential of the DNA base guanosine.22 Thus, dependent on the reduction potential of the chromophore used to label the DNA hairpin, efficient fluorescence quenching via photoinduced electron transfer occurs in the excited state upon contact formation with guanosine. In contrast to molecular beacons, these so-called smart probes14 are relatively easy to synthesize (i.e., a single labeling step is required) and exhibit a free DNA terminus at the other end for further modification of the hairpin, e.g., for immobilization on solid supports, without chemical modification in the loop.21 With careful design of these conformational flexible DNA probes and the selection of appropriate chromophores, efficient and highly sensitive smart probes can be synthesized that increase fluorescence intensity like molecular beacons up to 20-fold upon specific hybridization to synthetic target sequence (Figure 1).11,20,21 Here we report on design and application of smart probes for fast and specific detection of M. xenopi and M. tuberculosis in homogeneous and heterogeneous assay formats. The red-absorbing oxazine derivative MR121 was covalently attached to the 5′end of amino-modified oligonucleotides and quenched by up to five guanosine residues in the complementary stem. Upon hybridization to the target sequence, fluorescence intensity increases, thus reporting the presence of specific target sequences. Our studies indicate that for analytical applications of DNA hairpin probes target sequences cannot be chosen arbitrarily. Both suboptimal secondary structures of the DNA hairpin and the targeted PCR amplicon have to be considered to obtain reliable results. That is, careful selection of suited hairpin and target sequences as well as the length of PCR amplicons is key for successful assay development. Our data show that in combination with theoretical methods (mfold23) and careful optimization of experimental parameters smart probes constitute a reliable method for the identification of 16S rRNA genes of different (20) Heinlein T.; Knemeyer, J. P.; Piestert, O.; Sauer, M. J. Phys. Chem. B 2003, 107, 7957-7964. (21) Piestert, O.; Barsch, H.; Buschmann, V.; Heinlein, T.; Knemeyer, J. P.; Weston, K. D.; Sauer, M. Nano Lett. 2003, 3, 979-982. (22) Seidel, C. A. M.; Schulz, A.; Sauer, M. J. Phys. Chem. 1996, 100, 55415553.
Table 1. Smart Probes for Identification of M. xenopi and M. tuberculosisa smart probeb
sequence (5′ f 3′)
∆G (kcal/mol)c
Φf (20 °C)
IO/IC (50 °C)
SPxenopi1 SPxenopi2 SPxenopi3 SPxenopi4 SPxenopi5 SPxenopi6 SPtuberculosis
CCCCCGTTTTCTCGTGGTGACGGTAGGGGGGGG CCCCCGTTTTCTCGTGGTGACGGTAGGGGGTTT CCCCGTTTTCTCGTGGTGACGGTAGCGGGGGGG CCCCCTAGGACCATTCTGCGCATGTGGGGGGGG CCCCTAGGACCATTCTGCGCATGTGGGGTTT CCCAGTAGGACCATTCTGCGCATGTGGGGGG CCCCCGTGGTGGAAAGCGCTTTAGGGGGGGG
-2.6 -1.5 -2.3 -1.5 -0.0 +0.5 -1.7
0.21 0.24 0.11 0.21 0.36 0.30 0.13
2.95 2.56 2.13 3.45 2.36 2.04 2.95
a Includes Gibbs free energy ∆G for the most stable secondary structure as predicted by mfold23 as well as the respective quantum yield Φ at f 20 °C and the relative increase in fluorescence intensity IO/IC upon addition of 10-7 M of a synthetic complementary oligonucleotide to 10-8 M b concentrations of smart probes at 50 °C in 10 mM Tris-HCl (pH 7.5) containing 300 mM NaCl and 1 mM EDTA. Two different target regions within the 16S rDNA of mycobacteria were chosen for SPxenopi: SPxenopi1-3, first hypervariable region; SPxenopi4-6, second hypervariable region. c As calculated by mfold version 3.1.2 using the following parameters: temperature 50 °C; 300 mM Na+; no Mg2+.
mycobacterial strains in homogeneous solution. As exemplified here for M. tuberculosis and M. xenopi, PCR amplicons of particular mycobacteria can be accurately identified even in the presence of other mycobacteria. In addition, we present a highly sensitive alternative approach for the detection of specific target sequences in a heterogeneous assay format. The method is based on the immobilization of biotinylated target sequences (using, for example, biotinylated primers for PCR reactions) on streptavidin-coated 5-µm silica microspheres and subsequent incubation with complementary smart probes. Due to the increase in fluorescence intensity upon hybridization and the resulting accumulation of fluorescent molecules on the surface of the microspheres, target concentration down to the picomolar level could be detected without a washing step. EXPERIMENTAL SECTION Mycobacterial Strains, Identification, and Sequencing. DNA of 15 of the most frequently isolated NTM as well as M. tuberculosis was used for the present study (see Supporting Information). All strains have been identified unambiguously to the species level by sequencing the entire 1500-bp 16S rRNA gene and a subgenic 441-bp fragment of the gene coding for the mycobacterial 65 000-Da heat shock protein (hsp65).24 Sequencing of respective PCR-amplified DNA was done on an automated sequencer (ABI310Prism, Applied Biosystems, Weiterstadt, Germany), following standard protocols.25 DNA sequences were analyzed using components of the GCG software package, which is provided within the HUSAR package by the Deutsches Krebsforschungszentrum (Heidelberg, Germany). Assignment of sequences to a given species was accomplished by online analysis using the BioInformatic Bacterial Identification (BIBI) sequence analysis tool.26 DNA Amplification, Product Detection, and Quantification. Genomic DNA from the strains was extracted using Qiagen (23) Zuker, M.; Mathews, D. H.; Turner, D. H. In RNA Biochemistry and Biotechnology; Barciszewski, J.; Clark, B. F. C., Eds.; NATO ASI Series; Kluwer Academic Publishers: Dordrecht, 1999; pp 11-43 (http://bioweb. pasteur.fr/seqanal/interfaces/mfold.html). (24) Haefner, B.; Haag, H.; Geiss, H. K.; Nolte, O. Mol. Cell Probes 2004. 18, 59-65. (25) Patel, J. B.; Leonard, D. G. B.; Pan, X.; Musser, J. M.; Berman, R. E.; Nachamkin, I. J. Clin. Microbiol. 2000, 38, 246-251. (26) Available at http://pbil.univ-lyon1.fr/bibi.
purification kits (Qiagen, Hilden, Germany). PCR amplicons of a part of the 16S rDNA were amplified using the primers listed in the Supporting Information. By standard procedures, a 240-bp amplicon from all mycobacteria was generated. Additionally, PCR fragments with lengths of 85, 120, 240, 360, 480, and 600 bp from M. xenopi and Mycobacterium fortuitum were used for optimization experiments. Considering the length of primers used for PCR and the length of the target sequence accessible for specific hybridization, PCR amplicons shorter than 85 bp were not generated. All PCR reactions were carried out in a 50-µL reaction mixture containing 10 mM Tris-HCl (pH 8.8), 50 mM KCl, 3 mM MgCl2, 320 µM (each) deoxynucleoside triphosphate, and 50 pmol of both the forward (16S-27f, 16S-120f, 16S-145f) and reverse primers (16S-230r, 16S-240r, 16S-360r, 16S-480r, 16S-600r; see Supporting Information). One unit of Thermus aquaticus DNA polymerase was used per reaction (all reagents were obtained from MBI Fermentas, St. Leon Rot, Germany). The thermal profile consisted of initial denaturation for 5 min at 94 °C, followed by 45 cycles of 94 °C for 45 s, 64 °C for 60 s, and 72 °C for 90 s, with a final extension at 72 °C for 10 min. PCR products were analyzed in 2% agarose electrophoresis gels. The concentration of the PCR products was determined spectrophotometrically by measuring the OD at 260 nm. Synthesis of Smart Probes. All experiments were performed with the oxazine dye MR121 (kindly provided by K. H. Drexhage, ATTO-TEC, Siegen, Germany).14 Similar results were obtained with ATTO 655 (ATTO-TEC). Oligonucleotides carrying a 5′-amino C6 modifier for covalent coupling of the dyes were customsynthesized by IBA GmbH (Go¨ttingen, Germany). The coupling reactions were carried out as follows. A 50-µL sample of the respective oligonucleotide (Table 1) at a concentration of 10-4 M was dissolved in 50 µL of NaHCO3 buffer (100 mM, pH 7.0), and 5 µL of N-succimidyl ester of the dye (2 mg/mL dimethylformamide) was added. The solution was incubated for 15-20 h at room temperature in the dark. Labeled oligonucleotides were purified by reversed-phase (RP18 column) HPLC (Agilent, Waldbronn, Germany) using a gradient from 0 to 100% acetonitrile in 0.1 M aqueous triethylammonium acetate. Purified oligonucleotides were dissolved in 50 µL of water. Fluorescence Spectroscopy. PCR products were denatured by heating to 99 °C for 10 min in a Thermocycler PTC-10 (MJ Research Inc.) and then chilled on ice. For hybridization experiAnalytical Chemistry, Vol. 77, No. 22, November 15, 2005
7197
Figure 2. mfold structures (modeling parameters: 300 mM Na+; 0 mM Mg2+) of the antisense strand of the 240-bp amplicon of M. xenopi at 45 and 50 °C. The binding site for the loop sequence of the synthesized smart probes is marked yellow.
ments, denaturated PCR product was added to an aqueous solution containing smart probes in different concentrations. To investigate absorption characteristics and to adjust oligonucleotide concentration a standard, absorption spectrometer (Cary 500 Scan; Varian, Darmstadt, Germany) was used. Fluorescence emission spectra were measured in standard quartz cuvettes in 10 mM Tris-HCl buffer (pH 7.5) containing 300 mM NaCl and 1 mM EDTA at 50 °C in a Cary Eclipse fluorescence spectrometer with temperature controller (Varian). In these experiments, defrosted PCR amplicons were added to smart probes and incubated at 50 °C for 60 min. Corrected fluorescence spectra were measured upon excitation at 635 nm. Relative fluorescence intensities and quantum yields were determined with respect to the optical density of the sample measured at 635 nm. To avoid reabsorption and reemission effects, the concentration was kept strictly below 1.0 µmol/L. Preparation of the Heterogeneous Assay. For application of smart probes in heterogeneous assay formats, streptavidincoated 5-µm silica microspheres (Bangs Laboratories Inc., Fishers, IN) were used. Prior to use, the silica microspheres (1 µL, 10 mg/µL) were washed in 10 mM PBS, pH 7.4, containing 0.02 g/mL BSA and 0.001 g/mL Pluronic 10300. Then the microspheres were incubated with single-stranded biotinylated target sequence in various concentrations (10-5, 10-7, 10-8, 10-10, and 10-11 M) for 30 min followed by a second washing step. Finally, loaded microspheres were transferred into reservoirs of a 96-well plate and diluted with 50 µL of smart probes in 10 mM PBS, pH 7.4. Concentration of smart probes was kept constant at 10-8 M in all experiments. Microsphere-based measurements were performed without washing at 37 °C after an incubation time of 30 min using a standard fluorescence microscope (IX-70, Nikon) after loading 96-well plates (Nunc GmbH & Co. KG, Wiesbaden, Germany) with 50 µL of a microsphere suspension. A mercury lamp (Nikon) was used as excitation light source. Excitation light was filtered by a band-pass filter (622 DRLP; Omega Optical Inc.). A dichroic mirror (629 DRLP; Omega Optical Inc.) was used to direct the light into the oil immersion objective (60×, NA 1.4; Olympus, Tokyo, Japan). Fluorescence emission collected by the same microscope objective was filtered by a band-pass filter (680 DF 25; Omega Optical Inc.) and imaged on a CCD camera (Imager 3L, La Vision, Go¨ttingen, Germany). 7198 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
Modeling of Secondary Structures. Modeling of secondary structures was performed using mfold version 3.1.2.23 The program uses nearest-neighbor energy rules (also called loop-dependent energy rules) to estimate the free energies of possible secondary structures of a given DNA sequence. The secondary structures of the 240-bp PCR amplicon of M. xenopi at 45 and 50 °C (Figure 2) represent the predicted secondary structures that exhibit minimal free energies. In addition, a superposition of all possible folding possibilities within a given energy range of the minimum folding energy can be computed by mfold. The multitude of optimal and suboptimal secondary structures is represented in so-called energy dot plots. A dot plot is a triangular plot that depicts base pairs as dots. A dot in column i and j of a triangular array represents the base pair between the ith and jth base of the given nucleic acid sequence. For every possible base pair (i, j), mfold computes the minimum free energy of any secondary structure that contains the (i, j) base pair. The minimum free energy of a base pair is represented by its color within the energy dot plots shown in Figure 3.23 RESULTS AND DISCUSSION At first glance, the design of smart probes seems to be fairly easy. One simply has to identify the desired target sequence, order the complementary amino-modified oligonucleotide (∼20 bp) prolonged by four to five G/C base pairs forming the stem of the hairpin, and attach a suited fluorophore to the cytosine side of the stem that is selectively quenched by G. However, preliminary hybridization experiments with PCR amplicons of mycobacteria showed only a small increase in fluorescence intensity. This can be attributed either to inefficient quenching within the smart probe or to a low molar fraction of molecules forming the doublestranded hybrid. Temperature-dependent and time-resolved fluorescence studies indicate that kinetics of DNA hairpin formation as well as formation of suboptimal secondary structures is mainly responsible for the observed inefficient performance (data not shown). To improve the design of smart probes for the specific identification of target DNA sequences, we combined theoretical and experimental approaches. Secondary structures of PCR amplicons and DNA hairpins were modeled using mfold.23 Experimentally we varied the target DNA sequences, the stem length, and the dangling bases of the smart probes as well as
Figure 3. Energy dot plots of smart probes. The probability of formation of secondary structures is displayed as function of thermodynamic stability. The axes of the dot plots correspond to the number of individual bases in the oligonucleotide starting at the 5′-end. An ideal smart probe should exhibit base pairing only in the diagonal region of the terminating stem. The stability of base pairing should be balanced (medium free energy between -1.5 and -2.5 kcal/mol at 50 °C) such that the DNA hairpin is closed in absence of target sequence but can easily be opened by hybridization to the complementary target sequence. Dot plots were computed by mfold for 300 mM Na+ at 50 °C.
hybridization conditions, i.e., salt concentration, temperature, and length of the PCR amplicons. As depicted in Figure 2, the target sequence (e.g., the DNA antisense strand of the 240-bp amplicon of M. xenopi) is not necessarily accessible for hybridization of a smart probe at a given temperature. In other words, hybridization is thermodynamically favored compared to the closed conformation only when the complementary target sequence is not folded into secondary structures. Therefore, we applied a temperature of 50 °C at a sodium ion concentration of 300 mM to ensure accessibility of the target sequence for modeling of smart probes as well as for hybridization experiments with complementary oligonucleotides. Selection of Target Sequences and Design of Smart Probes. A target sequence from the 5′-hypervariable region of the 16S rRNA gene from M. xenopi was chosen as the model system for the design of DNA hairpins and optimization of experimental conditions. Six different smart probes each consisting of a varying 20-nucleotide loop with a stem of four or five G/C base pairs and a single-stranded triple-T or triple-G overhang at the 3′-end were synthesized (Table 1). Recent findings indicate that dangling bases at the 3′-end increase the quenching efficiency in the closed hairpin structure.20 The NHS ester of the oxazine derivate MR121 was covalently attached to the 5′-end of the smart probes using C6-amino modifier. Parts of the guanosine-containing stem sequence were designed to hybridize to the target sequence to maximize the thermodynamic stability of the resulting DNA hybrid. In aqueous solutions, MR121 exhibits an absorption maximum at 661 nm and emits with a fluorescence lifetime of 1.85 ns at an emission maximum of 673 nm.20 Upon covalent coupling to oligonucleotides, the absorption and emission maximums exhibit bathochromic shifts of 2-5 nm and show a
substantial decrease in fluorescence intensity. The relative fluorescence quantum yields, Φf of 0.11-0.36 measured at 20 °C (Table 1) announce a 3-9-fold increase in fluorescence intensity upon hybridization to their respective target sequences. The relative increase in fluorescence intensities measured at 50 °C upon addition of a 10-fold excess of complementary oligonucleotide, IO/IC (Table 1) indicates that the hairpins SPxenopi1 and SPxenopi4 are most suitable for the detection of M. xenopi PCR amplicons at higher temperatures. The free energies ∆G of closed DNA hairpins were estimated using mfold for a sodium ion concentration of 300 mM at 50 °C and should therefore corroborate the extracted experimental results (Table 1). The relatively high free energies calculated for SPxenopi 5 and 6 indicate the formation of unstable secondary structures, which is reflected in inefficient quenching in the closed state and a corresponding smaller increase in fluorescence intensity upon hybridization to the target sequence. The correlation between free energies and experimental results is less pronounced for the other DNA hairpin probes (SPxenopi 1-4), which renders the design of efficiently quenched smart probes based on theoretical considerations more complicated. While the experimental data are governed by the ensemble of all possible secondary structures, the Gibbs free energy of a certain smart probe reflects the stability of the most stable secondary structure only. The multitude of possible secondary structures of the DNA oligonucleotides chosen as smart probes is represented by mfold in energy dot plots (Figure 3). Each base pair (i, j) within a secondary structure is represented as a dot in column i and row j. The minimum free energy of the base pair, i.e., the minimum free energy of any secondary structure containing this base pair, is reflected by its color. For better comparison, the energy scale Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7199
Figure 4. (A) Melting curve of a 10-8 M solution of SPxenopi4 in the absence (dashed line) and presence of a 10-fold excess of complementary target oligonucleotide (black line) at temperatures between 10 and 90 °C. (B) Relative increase in fluorescence intensity versus length of the target PCR amplicons of M. xenopi (black squares) and the mismatching M. fortuitum (open circles) upon hybridization with the smart probe SPxenopi4 measured at 50 °C. Measurements were performed 60 min after addition of 10-7 M PCR amplicon to 10-8 M solutions of SPxenopi4 (i.e., a 10-fold excess of target sequence). All measurements were carried out in 10 mM Tris-HCl (pH 7.5) containing 300 mM NaCl and 1 mM EDTA.
was normalized to cover the total energy range of the secondary structures of all seven smart probes. While the left lower triangle shows the most stable secondary structure only, the upper right triangle depicts the superposition of all possible secondary structures of the respective smart probe. Comparison of the different dot plots shown in Figure 3 allows a more comprehensive interpretation of the interrelation between modeled and experimentally data. As can be seen by the existence of multiple dots in the upper right corner of the plots, the two smart probes SPxenopi1 and SPxenopi4 can form different helices within the stem. Since the fluorescent dye attached to the 5′-terminal cytosine residue is in proximity to guanosine residues in the complementary stem, all of these structures can be expected to yield a quenched state. The other smart probes directed against M. xenopi display only few helices in the stem. While the Gibbs free energies calculated for SPxenopi2 and SPxenopi3 predict a relatively stable secondary structure, the probes SPxenopi5 and SPxenopi6 are expected to be rather unstable and thus most of the time in an unquenched conformation. The higher stability of the secondary structures of SPxenopi1 as well as comparison of the loop region of SPxenopi1 and SPxenopi4 gives a clue as to why the latter might yield a higher increase in fluorescence intensity upon addition of complementary oligonucleotides (Table 1). While SPxenopi4 displays only few base pairs in the loop region, SPxenopi1 is predicted to form more and partly even more stable secondary structures also in the loop region, which is important for specific hybridization to the target sequence. Therefore, SPxenopi1 shows a comparatively lower increase in fluorescence intensity upon addition of complementary oligonucleotides. The other probes against M. xenopi are predicted to show only a poor performance as is confirmed by experimental data (Table 1). The approach described above was also applied for the design of a smart probe (SPtuberculosis) directed against M. tuberculosis. The corresponding dot plot in Figure 3 predicts multiple secondary structures in the stem region and only few and relatively unstable secondary structures in the loop that might interfere with hybridization. Based on these results, SPxenopi4 and SPtuberculosis were selected for the development of species-specific assays. In the following, SPxenopi4 was used exemplary for optimization of experimental assay conditions. 7200 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
Optimization of Homogeneous Assay Conditions. To improve the performance of the developed technique, we investigated the influence of the length of the PCR product on hybridization efficiency of hairpin probes, i.e., the accessibility of the target sequence in PCR products at different temperatures. As the mfold model predicts that the antisense strand of the 240-bp PCR amplicon from M. xenopi is only freely accessible at temperatures of g50 °C, we first investigated the temperaturedependent fluorescence intensity, i.e., the thermodynamic stability, of SPxenopi4 (Figure 4A). As can be seen in Figure 4A, temperature-induced opening and closing of the hairpin probe SPxenopi4 occurs at a melting temperature of 74.8 °C. The melting curve demonstrates that SPxenopi4 is still efficiently quenched at 50 °C while the 240-bp target PCR amplicon of M. xenopi should be accessible for hybridization at the same temperature (Figure 2). Temperatureinduced opening of the DNA hairpin of SPxenopi4 that causes unspecific fluorescence background can be avoided when working at temperatures below ∼60 °C. Thus, SPxenopi4 is ideally suited for hybridization experiments at 50 °C. Since the number of possible secondary structures that might interfere with the hybridization of the smart probe is expected to increase with the length of the antisense PCR amplicon, the hybridization efficiency of SPxenopi4 to PCR amplicons of various lengths was determined. Figure 4B shows the relative fluorescence intensity of 10-8 M solutions of SPxenopi4 mixed with a 10-fold excess of PCR amplicons (10-7 M) of different length. Interestingly, hybridization occurs most efficiently at 50 °C with the 240-bp amplicon. While the decrease in fluorescence intensity for fragment lengths of >240 bp can be explained by the higher number of possible secondary structures, the decrease in fluorescence intensity measured for shorter PCR fragments is unexpected and might be of interest for further investigations. To summarize, the obtained results imply that hybridization of SPxenopi4 to the 240bp amplicon of M. xenopi at 50 °C results in the maximum increase in fluorescence intensity. The maximum increase in fluorescence intensity of SPxenopi4 upon hybridization to target sequence of ∼4.7 was measured upon addition of a 100-fold excess of the antisense 240-bp amplicon (10-6 M) to a 10-8 M solution of SPxenopi4 (Figure 5A). In addition, the absolute hybridization efficiency can be determined by
Figure 5. (A) Increase in fluorescence intensity with time upon addition of a 100-fold excess of the 240-bp amplicon of M. xenopi to a 10-8 M solution of SPxenopi4 at 50 °C. (B) Relative fluorescence intensity of a 10-8 M solution of SPxenopi4 at 50 °C in the presence of varying concentrations of complementary target sequence (10-6-10-9 M 240-bp PCR amplicon M. xenopi) using a conventional fluorescence spectrometer (λexc ) 635 nm, λem ) 680 nm). All measurements were performed in 10 mM Tris-HCl (pH 7.5) containing 300 mM NaCl and 1 mM EDTA.
calculating the fluorescence quantum yield of SPxenopi4 using eq 1. Here I is the intensity of the integrated fluorescence emission
Q)
IODRn2 IRODnR2
(1)
spectrum of SPxenopi4 and OD is the absorbance of SPxenopi4 at the absorption maximum of the dye MR121. The index R stands for the labeled oligonucleotide (5′-ACT ACC AAG ATC TGC C AC-3′) that has been used as reference. The relative fluorescence quantum yield Φf of SPxenopi4 of 0.21 indicates that upon complete hybridization to the target sequence a 4.8-fold increase in fluorescence intensity can be expected, which corresponds well to the experimental data shown in Figure 5A. A direct comparison of the increase in fluorescence intensity measured for molecular beacons and smart probes upon complete hybridization to their complementary target sequences (PCR amplicons) is complicated. In general, molecular beacons are used for real-time monitoring of PCR, measuring the fluorescence generated by hybridization of the molecular beacon to its target during PCR. Typically fluorescence intensities are measured at 50-60 °C and start to increase after ∼30-35 PCR cycles with maximum rises of up to 10-fold after 40-50 cycles.18,19 However, Figure 5B demonstrates that the performance of the smart probe SPxenopi4 can be used advantageously to detect PCR amplicons down to concentrations of ∼2 ×10-8 M using a conventional fluorescence spectrometer. As recently demonstrated, the detection sensitivity can be further improved by 2-3 orders of magnitude by application of more sensitive detection techniques such as single-molecule fluorescence spectroscopy (SMFS).14 To determine the specificity of SPxenopi4 hybridization, further experiments were carried out with DNA of a number of different mycobacterial species. PCR amplicons from the 16S rDNA of M. xenopi, M. tuberculosis, and 14 further NTM species, containing 4-12 mismatches (with regard to M. xenopi) within the loop sequence, were used as control (see Supporting Information) applying similar hybridization conditions. Only the sense strand of the M. xenopi amplicon shows a perfect sequence homology to the loop sequence of SPxenopi4 within the first hypervariable region of the 16S rDNA. As demonstrated in Figure 6, SPxenopi4 can be used advantageously for the unequivocal identification of
Figure 6. Relative fluorescence intensity of 10-8 M solutions of SPxenopi4 (black) and SPtuberculosis (white) measured 1 h after addition of a 10-fold excess of 16 different 240-bp PCR amplicons (10-7 M) from mycobacterial strains in 10 mM Tris-HCl (pH 7.5) containing 300 mM NaCl and 1 mM EDTA, at 50 °C: 1, Mycobacterium szulgai; 2, Mycobacterium kansasii; 3, Mycobacterium abscessus; 4, Mycobacterium gastri; 5, Mycobacterium gordonae; 6, Mycobacterium celatum; 7, M. tuberculosis; 8, M. fortuitum; 9, Mycobacterium malmoense; 10, Mycobacterium peregrinum; 11, Mycobacterium interjectum; 12, M. xenopi; 13, Mycobacterium chelonae; 14, M. avium; 15, Mycobacterium marinum; 16, Mycobacterium intracellulare.
M. xenopi reflected in a significant fluorescence increase only upon addition of a 10-fold excess of the perfect matching PCR amplicon. The fluorescence increase upon addition of PCR amplicons can therefore be regarded as being specific for the species M. xenopi. To demonstrate the general applicability of the method, we designed a second smart probe that is exactly complementary to the antisense strand of the M. tuberculosis amplicon (see Supporting Information). Though the fluorescence intensity also increased slightly upon addition of related PCR amplicons because of unspecific interactions, the increase never exceeds a value of ∼1.5-fold, while the species-specific signals increase ∼2-fold upon addition of a 10-fold excess of PCR amplicons (Figure 6). In addition, we investigated the kinetics of hybridization of both smart probes, SPxenopi4 and SPtuberculosis in the presence of a 10fold excess of different PCR amplicons (Figure 7). The slightly reduced increase in fluorescence intensity of ∼1.5- and ∼1.7-fold measured for specific hybridization of SPxenopi and SPtuberculosis, respectively, might be explained by the presence of four Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7201
Figure 7. Increase in fluorescence intensity versus time upon hybridization of SPxenopi4 (A) and SPtuberculosis (B) to their respective complementary PCR amplicon (black lines) in the presence of four unrelated PCR amplicons of M. avium, M. gastri, and M. fortuitum as well as M. tuberculosis and M. xenopi, respectively. Experiments performed in absence of specific target PCR amplicons show no significant increase in fluorescence intensity (gray lines). For each experiment, a 10 mM Tris-HCl buffered solution (pH 7.5) containing 300 mM NaCl, 1 mM EDTA, and 5 × 10-8 M PCR amplicons was added to a 5 × 10-9 M solution of smart probe at 50 °C.
Figure 8. Fluorescence intensity images of 5-µm microspheres modified with single-stranded target sequence (5′-AAAAAAAAAAAAAAACCCCTTGTGAGGAACTACT-3′-biotin) in different concentrations upon addition of a 10-8 M solution of smart probe (3′-GGGGAACACTCCTTGATGATTCCCC-5′-C6-MR121) in 10 mM PBS, pH 7.4, at 25 °C. Biotinylated target sequences were immobilized on streptavidin-coated microspheres using concentrations of 10-5 (A), 10-7 (B), 10-8 (C), 10-10 (D), and 10-11 M (E). (F) Fluorescence intensity image measured for microspheres modified with unspecific DNA sequence (5′-TGGTG GAAAG CGTTT GGTAG CGGTG TGGGA TGGGC CCGC-Biotin-3′, 10-5 M) in the presence of 10-8 M smart probe.
unrelated PCR amplicons and subsequent possible unspecific interactions. In these experiments, the smart probe has to hybridize to the correct target sequence in the presence of a 40-fold excess of unrelated PCR amplicons. Nevertheless, the results clearly demonstrate that properly designed smart probes can distinguish among different species of mycobacteria under optimized hybridization conditions even if multiple mycobacteria are simultaneously present in the sample. Application of Smart Probes in Microsphere-Based Heterogeneous Assays. Homogeneous assays are in general restricted with respect to detection sensitivity.28 Although more sensitive methods, i.e., SMFS, are about to break the current 7202 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
barriers of sensitivity, alternative approaches such as heterogeneous assays can be used to increase the sensitivity by use of less demanding techniques.29 The presented approach for microsphere-based assays combines the increase in fluorescence intensity of smart probes upon specific binding with the accumulation of fluorescence on the surface of appropriately modified microspheres. In first experiments, short biotinylated (27) Fang, X.; Li, J. L.; Perlette, J.; Tan, W.; Wang, K. Anal. Chem. 2000, 72, 747A-753A. (28) Steemers, F. J.; Ferguson, J. A.; Walt, D. R. Nat. Biotechnol. 2000, 18, 9194. (29) Horejsh, D.; Martini, F.; Poccia, F.; Ippolito, G.; Di Caro, A.; Capobianchi, M. R. Nucleic Acids Res. 2005, 33, e13.
oligonucleotides were immobilized on streptavidin-coated silica microspheres in different concentrations (10-5-10-11M). Figure 8 shows fluorescence intensity images of microspheres modified with single-stranded target DNA measured in the presence of 10-8 M complementary smart probe without a washing step. Please note that the fluorescence intensity scale is different for oligonucleotide concentrations below 10-10 M. Comparison of the signals obtained for microspheres modified with specific target sequence (10-5-10-11 M) and unspecific target sequence (10-5 M) demonstrates that target DNA sequences down to a concentration of 10-11 M can be easily identified. These results suggest that the smart probe technique is also well suited for the development of highly sensitive heterogeneous DNA assay formats using microspheres and biotinylated primers for PCR reactions to immobilize the resulting amplicons. CONCLUSIONS We have demonstrated a simple and reliable method for the sensitive species-specific identification of mycobacterial strains using singly labeled DNA hairpins, i.e., smart probes. Both theoretical considerations and experimental results showed that besides thermodynamic considerations the multitude of possible secondary structures of DNA hairpins and their dynamics also has to be taken into account for their efficient design. Comparison of the dot plots generated by mfold and experimentally determined fluorescence quantum yields suggests that suitable DNA hairpins should form no or only few base pair contacts within the loop. In addition, our data imply that the formation of multiple alternative helices within the stem of the DNA hairpin maximizes fluorescence quenching. Theoretical results predict that the free energy of hairpin secondary structures should be in the range of -1.5 to -2.5 kcal/mol at 50 °C to ensure sufficient stability but likewise to enable easy hybridization to the target. Based on these findings, two smart probes SPxenopi4 and SPtuberculosis were selected for the species-specific identification of M. xenopi and M. tuberculosis, respectively. The optimum temperature for hybridization experiments was determined to 50 °C where the number of
secondary structures in the PCR amplicon is sufficiently low to enable specific hybridization while formation of the DNA hairpin structure of the smart probe is still favored. Experiments with PCR amplicons of various lengths demonstrate that the 240-bp amplicon of M. xenopi shows the best results with respect to increase in fluorescence intensity under the selected experimental conditions. Consideration of all theoretical and experimental results allowed us to develop a reliable method for detection of M. tuberculosis and M. xenopi with a detection sensitivity of ∼2 × 10-8 M in homogeneous solution. Experiments with amplicons of different mycobacterial strains demonstrate species-specific discrimination of M. tuberculosis and M. xenopi against 15 of the most frequently isolated mycobacterial species in a single assay. Results obtained by application of a microsphere-based heterogeneous assay format allowed us to increase the detection sensitivity by 3-4 orders of magnitude toward the picomolar range. Thus, the smart probe principle provides a new avenue for the development of diagnostic tools for use in molecular and medical microbiology based on the specific recognition of genetic information. Combination of specificity with novel sensitive detection approaches permits the development of faster, less expensive, and more reliable diagnostic assays that might soon pave the way for genetic identification avoiding PCR. ACKNOWLEDGMENT We thank K. H. Drexhage for providing the oxazine derivative MR121. Research described in this article was supported by the BMBF and VDI (Grants FKZ 13N8348 and 13N8349). SUPPORTING INFORMATION AVAILABLE Mycobacterial strains, the PCR primers used, and the alignment of the strains M. xenopi and M. tuberculosis. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review August 11, 2005. Accepted September 8, 2005. AC051447Z
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7203
