Spectroscopic characterization of fluorescein- and tetramethylrhodamine-labeled oligonucleotides and their complexes with a DNA template

Spectrochimica Acta Part A 60 (2004) 2741–2750

Spectroscopic characterization of fluorescein- and tetramethylrhodamine-labeled oligonucleotides and their complexes with a DNA template L. Wang∗ , A.K. Gaigalas, J. Blasic, M.J. Holden Biotechnology Division, National Institute of Standards and Technology, 100 Bureau Drive, Stop 8312, Gaithersburg, MD 20899-8312, USA Received 15 December 2003; accepted 13 January 2004

Abstract We measured absorption and emission spectra, fluorescence quantum yield, anisotropy, fluorescence resonance energy transfer (FRET), and melting temperature to characterize fluorescein- and tetramethylrhodamine (TMR)-labeled oligonucleotides in solution and when hybridized to a common DNA template. Upon hybridization to the template, both the absorption and emission spectra of TMR-labeled duplexes exhibited a shift with respect to those of labeled oligonucleotides, depending on the location of the TMR on the oligonucleotide. Measurements of quantum yield, anisotropy, and melting temperature indicated that TMR interacted with nucleotides within the duplexes in the order (T1 > T5 > T11, T16) that the oligonucleotide with TMR labeled at the 5 end (T1) is stronger than that labeled at position 5 from the 5 end (T5), which is also stronger than those labeled at the positions, 11 and 16, from the 5 end (T11, T16). In the case of the duplex formed between T1 and the template, fluorescence quenching was observed, which is attributed to the interaction between the dye molecule and guanosines located at the single-stranded portion of the template. A two-state model was suggested to describe the conformational states of TMR in the duplex. The melting temperatures of the four FRET complexes show the same pattern as those of TMR-labeled duplexes. We infer that the interactions between TMR and guanosine persist in the FRET complexes. This interaction may bring the donor and the acceptor molecules closely together, which could cause interaction between the two dye molecules shown in absorbance measurements of the FRET complexes. © 2004 Elsevier B.V. All rights reserved. Keywords: Fluorescein; Tetramethylrhodamine; DNA complexes; Fluorescence quenching; Fluorescence resonance energy transfer; Melting temperature

1. Introduction The number of fluorescence-based assays has grown rapidly in the last twenty years, especially in the fields of biological research and clinical diagnosis. The growth is mostly driven by high sensitivity, selectivity, and advancement in fluorophore conjugation chemistry. For example, in the area of nucleic acid research various methods based on fluorescence have been developed for the detection and quantification of DNA and RNA [1–4]. DNA microarray technology is becoming a potent tool for detection of gene expression levels [5–7]. Therefore, the understanding of the underlying mechanisms of interactions between fluorophores and nucleic acids and factors that affect fluores∗ Corresponding author. Tel.: +1-301-975-2447; fax: +1-301-975-5449. E-mail address: [email protected] (L. Wang).

1386-1425/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.saa.2004.01.013

cence measurements will further enhance the growth and improve the accuracy of fluorescence-based assays. Until recently it has been reported that guanosine and guanine cause fluorescence quenching of many commonly used fluorescent dyes, such as fluorescein, Coumarin, BODIPY FL, TAMRA, JOE, HEX, TET, ROX, and some of Alexa dyes, to name a few [8,9]. The quenching mechanisms were suggested to be due to photo-induced electron transfer from guanine to the singlet-excited state of dye molecules [8,10]. Compared to other nucleobases, guanine serves as a better electron donor. Although the quenching mechanism by photoinduced electron transfer is generally accepted, the effect of fluorescence quenching in oligonucleotides with specific sequences is not fully understood and depends on details of the sequences and hybridization. In the report of Lee et al. [11], for instance, the authors observed two-fold fluorescence quenching upon hybridization of a 14 mer fluorescein 5-isothiocyanate-labeled oligonucleotide to its comple-

2742

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

mentary strand. When BamHI restriction endonuclease was added which recognizes the sequence that is five nucleic bases away from fluorophore labeling site, the fluorescence of fluorescein was totally recovered in spite of the existence of the same close-by sequence of the duplex. Another example reported by Nazarenko and coworkers is that the fluorescence of oligonucleotides containing fluorescein conjugated to the 5 -terminal dC or dG decreases by ∼40% upon duplex formation [9]. The authors explained the quenching phenomenon as the terminal dG–dC and dC–dG base pairs responsible for the fluorescence quenching. Although the mechanisms of fluorescence quenching of oligonucleotides and formed duplexes need further investigation, many biological assays have been developed that make use of the quenching effect by guanosine [12–14]. One frequently used tactic is to design a molecular beacon with the labeling site of a fluorophore in the hairpin region and opposite to several guanosine residues. This design results in highly quenched fluorescence. When the molecular beacon is hybridized to its complementary strand, the fluorescence from fluorophore is restored. Assay developments using fluorescence quenching and de-quenching effects would certainly add to the understanding of the mechanism of the interaction between fluorophores and nucleotides. In a previous publication [15] we reported the FRET efficiency in a set of DNA constructs in which fluorescein and TMR served as the energy donor and acceptor, respectively. The two fluorophores were covalently attached to two separate oligonucleotides and both labeled oligonucleotides were hybridized to adjacent sections of a common DNA template to form a three-component double-stranded duplex. A similar configuration is implemented for a quantitative real-time PCR with LightCycler technology, where a 1–5 base separation between a donor and an acceptor is recommended to optimize the efficiency of resonance energy transfer. The aim of the previous study was to evaluate fluorescence resonance energy transfer (FRET) as a quantitation method in the context of flexible linkers, the unpaired nucleic base between two fluorophore-labeled nucleotides, the separation distance between two fluorophores, and temperature. The study identified a critical parameter, the separation distance between the donor and acceptor, in quantifying the FRET response. We also found that when fluorescein and TMR were separated under six nucleic bases, the absorbance measurements indicated ground-state interactions between two fluorophores although the FRET efficiencies were above 80%. The present investigation is intended to reveal the possible interactions between fluorophore and nucleotide and between two fluorophores conjugated to a template nucleotide by available spectroscopic methods. We measured absorbance, fluorescence, quantum yield, anisotropy, fluorescence resonance energy transfer, and melting temperature to characterize fluorescein- and tetramethylrhodamine (TMR)-labeled oligonucleotides free in solution and when hybridized to the same DNA template. To develop an assay with a high sensitivity it is important to know the fluo-

rescence quantum efficiency of the fluorophore because of quenching and other microenvironment effects. Many studies reported only ratios of fluorescence intensities between formed duplex and the single stranded nucleotide. Fluorescence anisotropy is a useful technique to detect possible interactions between a fluorophore and the micorenvironment through a change in rotational orientation of the fluorphore. Moreover, the quenching phenomenon is most problematic in FRET assays in that quenching of the donor fluorescence could result from both resonance energy transfer and quenching by the microenvironment. This work and previous study [15] are in support of our effort to develop model DNA reference materials for standardizing real-time PCR instruments and protocols.

2. Materials and methods Carboxyfluorescein succinimidyl ester (mixture of the five and six isomers) and 5-carboxytetramethylrhodamine succinimidyl ester were purchased from Molecular Probes, Inc. (Eugene, OR).1 All fluorophore-labeled oligonucleotides (Table 1) were synthesized on a Perseptive Biosystems Expedite DNA synthesizer using standard phosphoramidite chemistry. 3 -Amino-Modifier C7 CPG was used for the attachment of fluorescein, and either Amino-Modifier C6 TFA or Amino-Modifier C6-dT was used for the attachment of rhodamine. These reagents were obtained from Glen Research (Sterling, VA). The attachment of fluorophores was carried out after the completion of DNA synthesis. The fluorophores were conjugated to the nucleotides through formation of a carboxamide bond. The final products were purified by HPLC on a C18 reverse-phase column with linear gradients of 0.2 M TEAA (Triethylamine Acetate), pH 7.0, and acetonitrile. The fluorescein-labeled oligonucleotide contains about 70% of the five isomer and 30% of the six isomer. The chemical structures of the fluorophore moieties and their linkages to the oligonucleotides have been given in the previous study [15]. The oligonucleotides without a label were obtained from Invitrogen Corporation (Carlsbad, CA). For the present study, the hybridization occurs in buffer A (20 mM Tris–HCl, 50 mM KCl, 2 mM MgCl2 , pH 8.4) using the following protocol: 25 ◦ C for 30 s, 95 ◦ C for 2 min, then decreasing the temperature to 25 ◦ C in the rate of 0.2 ◦ C s−1 , incubation for 8 min at 25 ◦ C. A molar ratio of 1 to 2.5 between a fluorophore-labeled oligonucleotide and the template was used to assure the formation of duplexes with a single fluorophore. For FRET to oc1 Certain commercial equipment, instruments, and materials are identified in this paper to specify adequately the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment are necessarily the best available for the purpose.

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

2743

Table 1 Oligonucleotide sequences Name

Labeled fluorophore

Sequence

F NF T1 T5 T11 T16 NT Temp

Fluorescein None TMR TMR TMR TMR None None

sTemp

None

5 -CAG CAC TGT CTC GTT GAC AGG CGT Ga -3 5 -CAG CAC TGT CTC GTT GAC AGG CGT G-3 5 -Ta CA ATC AGC CTG AAA TCC TCG ATC AGA GTG-3 5 -TCA ATa C AGC CTG AAA TCC TCG ATC AGA GTG-3 5 -TCA ATC AGC CTa G AAA TCC TCG ATC AGA GTG-3 5 -TCA ATC AGC CTG AAA Ta CC TCG ATC AGA GTG-3 5 -TCA ATC AGC CTG AAA TCC TCG ATC AGA GTG-3 5 -ACA CTC TGA TCG AGG ATT TCA GGC TGA TTG ACC ACG CCT GTC AAC GAG ACA GTG CTG A-3 5 -ACA CTC TGA TCG AGG ATT TCA GGC TGA TTG ACC AC-3

Refers to the fluorophore labeling site. TMR is labeled either through the phosphate group at the 5 end (T1) or through the T nucleic acid base at various positions from the 5 end (T5, T11, T16). a

cur, the fluorescein-labeled oligonucleotide (F) and the rhodamine-labeled strand hybridize to their DNA complementary strand (Temp) to form a three-component construct (Table 1). The rhodamine-labeled strand was synthesized in four different forms, with rhodamine at different positions, so that four different complexes with different donor–acceptor distances could be studied (T1, T5, T11, and T16, Table 1). The same complexes with a single fluorophore, either fluorescein or rhodamine, were also made for the spectroscopic investigations. The hybridized, three-component complexes were purified by electrophoresis in 12% polyacrylamide gel with TBE (0.1 M Tris, 0.09 M Boric Acid, and 0.001 M EDTA, pH 8.4) as the running buffer. A 10 bp DNA ladder from Invitrogen Corporation (Carlsbad, CA) was used to confirm the size of the complexes. The identified complexes were recovered from the gel using Ultrafree-MC centrifugal filter devices (Durapore 0.22 ␮m) and further concentrated using Microcon YM-30 centrifugal filter devices. Both filter devices are products of Millipore Corporation (Bedford, MA). The purified complexes were diluted in hybridization buffer A (see above) to maintain a volume of 120 ␮l for the fluorescence measurements. The absorbance measurements were performed using a Chem2000 fiber optic spectrophotometer (Ocean Optics Inc., Dunedin, FL), operated with a tungsten halogen light source. Each spectrum was acquired for 15 s. Samples (0.25 ml) were introduced into 100 cm optical path length waveguide capillary cell (model LWCC-2100, World Precision Instruments, Inc., Sarasota, FL) by means of a peristaltic pump. Hybridization buffer was used as the reference for the measurements. The steady-state fluorescence measurements were made with a SLM 8000 spectrofluorimeter from Jobin Yvon, Inc. (Edison, NJ). The fluorescence spectra were collected under “magic angle” condition with the excitation wavelength of 490 nm for fluorescein and of 550 nm for TMR. The slits for excitation and emission monochromators were both set at 4 nm. Fluorescein in 0.01 M NaOH with a fluorescence quantum yield of 0.93 [16] was used as the standard to obtain relative quantum yields of fluorescein-labeled com-

plexes. Rhodamine B in ethanol with a quantum yield of 0.69 [17–19] was used as the reference standard for the measurement of the quantum efficiencies of TMR-labeled complexes. The general equation for the determination of relative quantum yields (Q) is given in Eq. (1) [20]. Qu =

Qs As Fu λs η2u Fs Au λu η2s

(1)

where the subscripts s and u stand for the reference standard and the unknown, respectively. The parameter A is the absorbance at the excitation wavelength, λ, and F is the integrated area under the emission spectrum. The square of the refractive index, η, is employed to make the correction due to the difference in media of the standard and the unknown. The fluorescence anisotropy (r) of labeled complexes was determined based on (FVV − GFVH ) r= (2) (FVV + 2GFVH ) where G is the correction factor for the detection system given by the ratio FHV /FHH , where the subscripts V and H refer to fluorescence intensities collected with the incorporation of vertical and horizontal polarizers, respectively, in the order of excitation followed by emission. Melting curve measurements were performed using a LightCycler from Roche Diagnostics Corporation (Indianapolis, IN), which employs a 470 nm LED as the light source and detects emission at three different wavelengths, 530, 640, and 710 nm. For the current measurements, the fluorescence signal was detected by either the green fluorescence channel (530 nm) or the 640 nm channel during the temperature transition from 54 to 82 ◦ C at the rate of 0.1 ◦ C s−1 . The melting temperature was calculated using the melting curve analysis program within LightCycler.

3. Results The absorption spectra of fluorescein-labeled oligonucleotide (F) free in solution and the hybridization product (F/Temp) with the template match each other perfectly (re-

2744

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

Fig. 1. Absorption spectra of TMR-labeled oligonucleotides (solid line) and the duplexes formed with the template (dash): (a) T1 and T1/Temp; (b) T5 and T5/Temp; (c) T11 and T11/Temp; (d) T16 and T16/Temp. These spectra are normalized for easy comparison.

sults not shown). However, the absorption spectra of the TMR-labeled oligonucleotides and the duplexes formed with Temp (Fig. 1) are shifted relative to each other, depending upon the location of the dye molecule. The shift is the smallest in T5 and the duplex, T5/Temp. By comparison, the duplex formed by T1 and Temp shifts to the red and the duplexes, T5/Temp, T11/Temp, and T16/Temp show the blue shift, all relative to the labeled oligonucleotides free in solution. The emission spectra of TMR-labeled oligonucleotides and the duplexes, given in Fig. 2, show similar phenomena with respect to the absorbance and display image relationship with the absorption spectra. As expected the emission spectra of fluorescein-labeled oligonucleotide (F) and the formed duplex, F/Temp, show no shift relative to each other (results not shown). The detailed explanation on the spectral shifts will be given in Section 4. The fluorescence quantum yield and anisotropy are given in Table 2 for the nucleotides and the corresponding complexes. It’s worthy of mention that the anisotropies for nucleotides and duplexes are relatively constant within the

range of corresponding excitation wavelengths, meaning that one singlet-excited state exists in all cases. When the fluorescein-labeled oligonucleotide F hybridizes to the template (Temp), the quantum yield increases from 0.52 to 0.70 and the anisotropy also increases from 0.051 to 0.079. In the case of TMR-labeled nucleotides and duplexes, it appears that the higher the quantum efficiency the lower the anisotropy. T11 and T16 behave similarly in terms of both quantum yield and anisotropy, so do their duplexes. When hybridization occurs between T11 and Temp, for example, the fluorescence quantum yield increases from 0.55 to 0.60 and anisotropy decreases from 0.17 to 0.14. On the contrary, the efficiency for T1 and T5 drops and the anisotropy increases upon hybridization to Temp. The increase in anisotropy upon hybridization is more apparent for T1 than for T5. Moreover, we replaced Temp with a shorter template, sTemp, which is a 35 mer oligonucleotide with the same sequence as Temp, but is truncated before the guanosine at position 36 from the 5 end of Temp (Table 1). The fluorescence quantum yield of the duplex,

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

2745

Fig. 2. Normalized emission spectra of TMR-labeled oligonucleotides (solid line) and the duplexes formed with the template (dash): (a) T1 and T1/Temp; (b) T5 and T5/Temp; (c) T11 and T11/Temp; (d) T16 and T16/Temp. The excitation wavelength is 550 nm with the slits set for both excitation and emission at 4 nm.

T1/sTemp, is 0.77 that is even higher that of T1 (Table 2). When T1 is hybridized to sTemp, TMR experiences an even more hydrophobic microenvironment given the facts that its anisotropy increases from 0.13 to 0.20 and the emission maximum shifts to the red with respect to that of T1. The emission exhibits the same shift in terms of the wavelength as that of T1/Temp (data not shown). For FRET to occur, the fluorescein-labeled oligonucleotide and a TMR-labeled nucleotide will have to hybridize adjacently to the same template (Temp), and fluorescein, serving as the energy donor, is excited to the singlet excited-state through a 490 nm light source. Fig. 3 shows the emission spectra of four different three-component DNA complexes investigated here. The spectra are normalized to the same height at 520 nm for easy comparison. The fluorescence from TMR peaked at ∼580 nm increases when the fluorophore is placed closer to the donor fluorescein, implying that the amount of energy transfer increases. Although the emission is collected under a “magic angle” configuration in the present study, the normalized spectra are similar

to those shown in the earlier report in the absence of excitation and emission polarizers [15]. It has been found in the previous study that the FRET efficiencies of four DNA constructs are 0.97 for F/T1/Temp, 0.85 for F/T5/Temp, 0.64 for F/T11/Temp, and 0.43 for F/T16/Temp, respectively. We have also found that there may be ground-state interaction between the donor fluorescein and the acceptor TMR. Fig. 4 shows the absorption spectra of the donor-alone (dash line) and acceptor-alone (dot line) oligonucleotides together with their constructs (solid line) for the two cases of two different separation distances. The composed spectra of the two complexes, F/T1/Temp and F/T11/Temp, are shown in Fig. 4a and b, respectively, by the dash dot line. When the spectrum of F/Temp was added to the spectrum of T11/Temp or T16/Temp the composed spectrum matched the measured spectrum of the respective three-component complex (Fig. 4b). However, differences were observed for complexes F/T1/Temp and F/T5/Temp. As shown in Fig. 4a, the spectrum of F/T1/Temp shows red shift or broadening when compared with the sum of the spectra of F/Temp and

2746

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

Table 2 Fluorescence quantum yields (Q) and anisotropies (γ) of fluorophorelabeled oligonucleotides, duplexes, and three-component complexes Sample

Q

γ

F F/Temp F/NT/Temp

0.52 (a) 0.70 (a) 0.72 (a)

0.051 (c) 0.079 (c) 0.081 (c)

T1 T1/Temp NF/T1/Temp T1/sTemp

0.71 0.60 0.61 0.77

0.13 0.24 0.23 0.20

T5 T5/Temp NF/T5/Temp

0.63 (b) 0.54 (b) 0.57 (b)

0.16 (d) 0.18 (d) 0.18 (d)

T11 T11/Temp NF/T11/Temp

0.55 (b) 0.60 (b) 0.57 (b)

0.17 (d) 0.14 (d) 0.14 (d)

T16 T16/Temp NF/T16/Temp

0.53 (b) 0.61 (b) 0.59 (b)

0.18 (d) 0.13 (d) 0.14 (d)

F/T1/Temp F/T5/Temp F/T11/Temp F/T16/Temp

(b) (b) (b) (b)

(d) (d) (d) (d)

0.11 (c) 0.10 (c) 0.090 (c) 0.092 (c)

(a) Fluorescein in 0.01 M NaOH with fluorescence quantum yield of 0.93 was used as the standard to obtain relative fluorescence quantum yield. (b) Rhodamine B in ethanol with quantum yield of 0.69 was the reference standard for the measured fluorescence quantum yield. The anisotropy is given at 490 nm for fluorescein (c) and at 560 nm for rhodamine (d). The standard deviation for the quantum yield is less than 0.02. The standard deviations for the anisotropy are between 0.002 and 0.003 for fluorescein and ≤0.01 for tetramethyl-rhodamine, respectively.

Fig. 4. Spectral comparison of the absorption spectrum of the complexes (solid line), F/T1/Temp (a) and F/T11/Temp (b), and the sum (dash dot) of the absorption spectra from fluorescein-labeled duplex (F/Temp, dash) and TMR-labeled duplex ((a) T1/Temp or (b) T11/Temp, dot).

Fig. 3. Fluorescence spectra of the three-component FRET constructs, F/T1/Temp (solid line), F/T5/Temp (dash), F/T11/Temp (dot), and F/T16/Temp (dash dot dot). In each complex, the acceptor TMR is located at a different distance from the donor fluorescein, which results in the difference in rhodamine fluorescence (>550 nm). The spectra are normalized to the same height at 520 nm, and the excitation wavelength is 490 nm.

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

T1/Temp. The broadening is more apparent in the case of F/T1/Temp than in the case of F/T5/Temp. To further probe the possible interaction between fluorescein and TMR in the two FRET complexes, F/T1/Temp and F/T5/Temp, the fluorescence anisotropy of fluorescein was measured by setting the observation window at 520 nm for the four three-component complexes. Their anisotropies are given at the excitation wavelength of 490 nm in Table 2 and they are relatively constant over the range of excitation wavelengths from 440 to 500 nm. When FRET occurs between a donor and an acceptor, the fluorescence lifetime of the donor is shortened and the extent of rotational diffusion within the shortened lifetime is less than that within the lifetime in the absence of the acceptor. Hence, the anisotropy of the donor increases in the presence of FRET. The anisotropies of the four complexes are higher than those of F/Temp and F/NT/Temp. Moreover, the anisotropies of F/T1/Temp and F/T5/Temp are slightly higher than those of F/T11/Temp and F/T16/Temp. Considering the FRET efficiencies of the four constructs in the following sequence, F/T1/Temp > F/T5/Temp > F/T11/Temp > F/T16/Temp, the fluorescein fluorescence lifetimes (τ) in the presence of the acceptor will be in the order of τF/T1/Temp < τF/T5/Temp < τF/T11/Temp < τF/T16/Temp . The anisotropy data show, in general, the trend that the construct with higher FRET efficiency exhibits higher anisotropy. We further measured the melting temperatures of the three-component, FRET complexes using the LightCycler. The fluorescence signal was collected by the green fluorescence channel (530 nm), in which the signal resulted primarily from the fluorescein chromophore. Fig. 5a and b show the measured fluorescence intensities and −dF/dT as a function of the temperature. In general, fluorescence quantum yield decreases as the temperature increases. When FRET occurs, the fluorescence from the donor fluorescein further decreases. The closer the donor is located from the acceptor, the more the energy transfer and the less the observed fluorescence signal from the donor. During the melting process fluorescein is pulled away from TMR and hence, the fluorescein fluorescence is restored. The derivatives of the fluorescence signal over the temperature allow the estimations of the melting temperatures of duplexes and constructs. These temperatures are listed in Table 3 along with the melting temperatures of NF and NT, which were obtained using the nearest neighbor (n–n) model [21,22] on an assumption that the oligonucleotide was hybridized to its complementary strand in the same buffer solution used in the present study. The melting temperatures of F/T11/Temp and F/T16/Temp are very similar, and somewhat lower than that of F/T5/Temp, which is also lower than that of F/T1/Temp. However, all melting temperatures of the donor–acceptor constructs are lower than those of F/Temp and F/NT/Temp, which are very close together. Our initial hypothesis was that the higher melting temperature of F/T1/Temp might be indicative of possible molecular

2747

Fig. 5. Melting temperature measurements using a LightCycler: F (solid line), F/T1/Temp (dash), F/T5/Temp (dot), F/T11/Temp (dash dot), and F/T16/Temp (dash dot dot). (a) Fluorescence signal detected by the green fluorescence channel (530 nm), in which the signal mostly results from fluorescein fluorophore, as a function of temperature. (b) The negative derivative of fluorescence signal over the temperature as a function of the temperature.

interaction between donor and acceptor dye molecules. To verify the hypothesis, we further measured the melting temperatures of the duplexes formed between TMR-labeled nucleotides and Temp using the 640 nm detection channel. Although the 640 nm channel is not optimal for the detection of TMR fluorescence, the negative derivatives of TMR fluorescence over the temperature give clean peaks for the estimation of the melting temperatures of the duplexes. These temperatures are also given in Table 3. It is worthy of mention that the melting temperatures of T11/Temp and T16/Temp are consistent with the theoretically predicted temperature of NT. The melting temperature of T1/Temp is higher than that of T5/Temp, which is also higher than those of T11/Temp and T16/Temp. The sequence of the melting temperatures from high to low for TMR-only duplexes is similar to that for fluorescein-TMR constructs. We will give our interpretation of these results in discussion section.

2748

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

Table 3 Melting temperatures of non-fluorescent oligonucleotides obtained through theoretical calculation and DNA complexes carried out using a LightCycler Sample

Tm (◦ C, S.D.)

NF/complementary strand NT/complementary strand F/Temp F/NT/Temp F/T1/Temp F/T5/Temp F/T11/Temp F/T16/Temp T1/Temp T5/Temp T11/Temp T16/Temp

71.5a (model) 69.7a (model) 73.6 (0.3)b,c 73.2 (0.4)b,c 72.3 (0.4)b,c 70.8 (0.4)b,c 69.1 (0.4)b,c 69.3 (0.4)b,c 73.2 (0.2)b,d 71.3 (0.4)b,d 69.9 (0.2)b,d 70.0 (0.2)b,d

a The melting temperature was obtained using the nearest neighbor (n–n) model on an assumption that the oligonucleotide was hybridized to its complementary strand. b The temperature with standard deviation (S.D.) was obtained using the LightCycler melting curve analysis program. c Fluorescence signal was detected via the green fluorescence channel (530 nm). d Fluorescence signal was detected by the red fluorescence channel (640 nm).

4. Discussion The absorption spectra of TMR-labeled oligonucleotides (Fig. 1) display different shifts compared to their duplexes with the template. So do their emission spectra (Fig. 2). The absorbance and emission of both labeled oligonucleotides and formed duplexes display an image relationship. In another study, we reported that the maximum of the emission spectrum of fluorescein could be manipulated through changing the linker length between polymer microsphere and fluorescein chromophore with respect to that of fluorescein in aqueous solution [23]. The shorter the linker length is, the more the emission maximum shifts to the red, indicating that the fluorophore experiences a more hydrophobic microenvironment with a short tether. The result is predicted from a modified Lippert equation [24], which uses the dielectric properties of the solvent and polymer bead materials. While applying the same principle to TMR-labeled oligonucleotides and duplexes, one could estimate fluorophore position relative to the nucleotide. For instance, the emission spectrum of the duplex (T11/Temp) shows blue shift with respect to that of T11, suggesting that TMR fluorophore in the duplex is further away from DNA double helix. In the case of T1/Temp, on the other hand, the spectrum displays red shift regarding that of T1. This implies that upon duplex formation TMR tends to stay more closely to a more hydrophobic microenvironment, the nucleotide. Finally, the emission spectrum of T5/Temp shows a very minor blue shift relative to that of T5, which suggests a minor change in the position of TMR relative to the nucleotide. To our knowledge, the wide range of the spectral shifts of TMR-labeled nucleotides has not been reported and explained in detail so far.

Fluorescence anisotropy is a very useful technique in that it allows deducing possible interactions between a fluorophore and its micorenvironment through a change in rotational rigidity of the fluorophore. The higher the anisotropy of a fluorophore is, the more rigid the fluorophore, inferring a stronger association with the surroundings. The anisotropy data are given in Table 2 for TMR-labeled oligonucleotides and the duplexes formed with Temp. Upon hybridization the anisotropies of TMR decrease from 0.17 in T11 and 0.18 in T16 to 0.14 in T11/Temp and 0.13 in T16/Temp, respectively. The data imply that the fluorophore gains rotational freedom through hybridization and moves slightly away from the nucleotide into the aqueous solution. This conclusion is consistent with that from the absorbance and emission measurements. The anisotropies of TMR in both T5 and T5/Temp are very similar. The small change might reflect an alternation in their molecular weight. On the contrary, the anisotropy of TMR increases from 0.13 in T1 to 0.24 in T1/Temp upon hybridization. This suggests that a strong association occurs between TMR and the template and hence the fluorophore must stay more closely to the nucleotide in this case than in the case of T1 free in solution. TMR experiences a more hydrophobic microenvironment in T1/Temp, and therefore, displays a red shift in its emission spectrum. Again the anisotropy result is consistent with that from absorbance and emission measurements. In general the measurements of the melting temperatures of the hybridization products are used to determine size and stability of the PCR products. Initially, we measured the melting temperatures of the three-component FRET complexes in an attempt to reveal possible interactions between dye molecules, which were observed in absorbance measurements (Fig. 4). The melting temperatures of the four constructs (Table 3 and Fig. 5) did show differences. We further measured the melting temperatures of the corresponding duplexes formed between TMR-labeled oligonucleotides and the template and they displayed a similar trend. Interestingly the melting temperatures of T11/Temp and T16/Temp are very close to that of NT hybridized to the assumed complementary strand calculated using the nearest neighbor (n–n) model. It is reasonable to infer that TMR in T11/Temp and T16/Temp has no effect on the melting temperatures of the duplexes. The melting temperature of T1/Temp is higher than that of T5/Temp, which is also higher than those of T11/Temp and T16/Temp. The higher temperature might be indicative of interactions between TMR and the nucleotide that stabilize the duplexes. Since the melting temperatures of the FRET constructs from F/T1/Temp to F/T16/Temp display a similar changing pattern as those of duplexes from T1/Temp to T16/Temp, one may speculate that the interactions between TMR and nucleotide persist in the FRET constructs, such as F/T1/Temp and F/T5/Temp. There are a few reports concerning the possible interactions between TMR and nucleotides, which cause different conformational states of rhodamine molecules [25–28]. Although various techniques were used to probe the interac-

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

2749

Fig. 6. Model of two conformational states of the TMR-labeled duplex, T1/Temp. State A represents the case that rhodamine molecule experiences aqueous microenvironment and is removed from interaction with guanosines. State B stands for the case that rhodamine fluorophore interacts with guanosines, for instance, guanosine at the 36 position from 5 end of the template, which likely results in a lower quantum yield than that of state A. The graph only shows partial sequences of T1 and the template (Temp) with some of the positions given numerically from the 5 end of the strands to depict the model.

tions in those reports, they confirmed the existence of the interactions that resulted in various conformational states of TMR with distinct fluorescence lifetimes. However, there is a discrepancy in terms of the number of the conformational states and lifetimes associated with each state [26–28]. We measured the fluorescence lifetimes of rhodamine molecules in single-stranded oligonucleotides and the duplexes formed with the template. The long component is 3.6 ± 0.3 ns, and we are unable to resolve the short component (≤1 ns) due to the limitations in the instrument. Because of the presence of fluorescence quenching, a short lifetime component exists, which was confirmed by the reports [26–28]. For the present investigation, we suggest a two-state model using TMR-labeled oligonucleotide T1 as the example (Fig. 6) because of the large changes in quantum yield and anisotropy upon hybridization to the template. State A has a long lifetime and high quantum yield (3.6 ns, ≥0.7, respectively), and is more associated with the solution environment. State B has a short lifetime and a low quantum yield (≤1 ns). It appears that state A dominates in T1 and both states are present in T1/Temp and NF/T1/Temp. Since both quantum yield and anisotropy are very similar for T1/Temp and NF/T1/Temp, it is reasonable to deduce that the oligonucleotide NF does not affect the distribution of the two states. When we replaced Temp with a shorter template, sTemp, the fluorescence quantum yield of the duplex T1/sTemp is 0.77 that is even higher that of T1 (Table 2). TMR in T1/sTemp experiences more hydrophobic microenvironment given that the anisotropy increases from 0.13 to 0.20 and the emission exhibits the same red shift as that of T1/Temp with respect to T1. The present result also agrees with that of Crockett and Wittwer [13] given that the guanosine at position 30 from the 5 end of sTemp does not cause the fluorescence quenching. In their report, guanosine at position −1, defined as the position next to the fluorescein labeling site but closer to the 5 end on the unlabeled complementary strand, showed no quenching effect on fluorescein labeled at 5 end

of the other strand. As a result, we infer that the decrease in quantum yield upon hybridization of T1 to Temp is due to the interaction between TMR and guanosines located at the single-stranded portion of Temp. We have used a structural model of the duplexes and constructs, which was made using standard B-form DNA and visualized using Molscript [15], to estimate the distances between TMR in T1 and nearby guanosines. In T1/Temp, the distance between TMR and phosphate linkage site (16.3 Å) is about the same as that between the linkage site and guanosine at position 36 from the 5 end of Temp (17 Å). It is also likely that the unpaired portion of Temp is flexible and leads to even shorter distances between guanosine 36 and TMR. We have also used a mfold program [29], version 3.1, to check possible secondary structures on the single-stranded portion of the template in T1/Temp on the current buffer condition. One relatively stable secondary structure appears that involves 5 bp between bases at positions 38–42 and those at positions 48–52 from the 5 end of the Temp. This suggests that guanosines at position beyond 52 are not involved in any stable structures that could cause fluorescence quenching of TMR. The distance between the linkage site of the dye molecule and guanosine at position 26 of Temp (18 Å) is slightly larger than the linkage length (16.3 Å). Steric hindrance would prevent the direct interaction between rhodamine molecule and guanosine at position 30 of Temp. Considering the fact that the fluorescence quantum yields and anisotropies of T1/Temp and NF/T1/Temp are about the same, we infer that the fluorescence quenching is likely attributed to the interaction between TMR and guanosine at position 36 from the 5 end of Temp. FRET occurs when both fluorescein-labeled and TMR-labeled oligonucleotides hybridize adjacently to the same template (Temp) as shown in Fig. 3 The resonance energy transfer is also confirmed by the higher anisotropy value (Table 2) because of the shorter fluorescence lifetime. The melting temperatures of the four FRET constructs

2750

L. Wang et al. / Spectrochimica Acta Part A 60 (2004) 2741–2750

show a similar pattern as that of TMR-labeled duplexes (Table 3), suggesting that a similar force dominates in both cases. We have deduced, for example, that in the cases of T1/Temp and NF/T1/Temp the rhodamine molecule is likely to interact with guanosine at position 36 from 5 end of Temp. If the same interaction exists in F/T1/Temp, the interaction between fluorescein and TMR is likely. The unpaired base in the template may allow bending and twisting which will facilitate this type of interaction. In fact, the ground-state interaction was observed when fluorescein and rhodamine molecules were placed close together in the cases of F/T1/Temp and F/T5/Temp (Fig. 4).

improve the sensitivity and accuracy of fluorescence-based assays.

Acknowledgements The authors are indebted to Dr. D.T. Gallagher for both valuable suggestions and constructing the DNA structural model for the distance estimations, and Dr. V. Reipa for the absorbance measurements.

References 5. Conclusion Fluorophores are used extensively to label and detect nucleic acids because of the high sensitivity, selectivity, and advancement in fluorophore synthesis and conjugation chemistry. However, there are complications likely due to interactions between commonly used fluorophores and guanosine which cause fluorescence quenching. Therefore, to develop robust assays, it is important to understand the mechanisms of interactions between fluorophores and nucleic acids and the factors that affect fluorescence measurements. The present work is a follow-up study to the previous FRET report [15] and aimed at understanding the effect of neighboring nucleotides on the fluorescence of the label. Using the same labeled nucleotides and the template the present study shows that tetramethylrhodamine molecules exhibit spectral shifts when the labeled oligonucleotides are hybridized to the template, depending on the location of dye attachment on the oligonucleotide. To our knowledge, the wide range of the spectral shifts associated with TMR-labeled nucleotides and their hybridization products has not been reported and explained in detail so far. Since T1 shows most interesting features upon hybridization in measurement parameters, such as quantum yield, anisotropy, and melting temperature, we focused on the cause of fluorescence quenching upon hybridization of T1 to the template. We found that fluorescence quenching is attributed to the interaction with guanosines located at the single-stranded portion of the template. A two-state model was suggested to describe the conformational states of TMR in the duplex, T1/Temp. The measurements of melting temperature were used to deduce the likely interaction between TMR and nucleotide in three-component FRET complexes as well as in the TMR-labeled duplexes. The interaction may bring the donor and acceptor molecules closely together, which could cause interaction between dye molecules as shown in the absorbance measurements of the FRET complexes. The investigation of the properties of two-component duplexes facilitates our understanding of the three-component FRET complexes. Since fluorescence of the label depends upon the sequence, systematic studies are needed to generalize the dependence, which will ultimately assist assay design and

[1] V.L. Singer, L.J. Jones, S.T. Yue, R.P. Haugland, Anal. Biochem. 249 (1997) 228–238. [2] D. Personett, K. Sugaya, D. Hammond, M. Robbins, M. McKinney, Electrophoresis 18 (1997) 1750–1759. [3] R.A. Hoffman, Method Cell Biol. 63 (2001) 299–340. [4] S.J. Wu, D.C. Spink, B.C. Spink, L.S. Kaminsky, Anal. Biochem. 312 (2003) 162–166. [5] S.A. Armstrong, J.J. Hsieh, S.J. Korsmeyer, Curr. Opin. Hematol. 9 (2002) 339–344. [6] C.M. Roth, Curr. Issues Mol. Biol. 4 (2002) 93–100. [7] K.V. Chin, A.N. Kong, Pharm. Res. 19 (2002) 1773–1778. [8] M. Torimura, S. Kurata, K. Yamada, T. Yokomaku, Y. Kamagata, T. Kanagawa, R. Yurane, Anal. Sci. 17 (2001) 155–160. [9] I. Nazarenko, R. Pires, B. Lowe, M. Obaidy, A. Rashtchian, Nucl. Acids Res. 30 (2002) 2089–2095. [10] C.A.M. Seidel, A. Schulz, M.H.M. Sauer, J. Phys. Chem. 100 (1996) 5541–5553. [11] S.P. Lee, D. Porter, J.G. Chirikjian, J.R. Knutson, M.K. Han, Anal. Biochem. 220 (1994) 377–383. [12] J. Knemeyer, N. Marme, M. Sauer, Anal. Chem. 72 (2000) 3717– 3724. [13] A.O. Crockett, C.T. Wittwer, Anal. Biochem. 290 (2001) 89–97. [14] I. Nazarenko, B. Lowe, M. Darfler, P. Ikonomi, D. Schuster, A. Rashtchian, Nucl. Acids Res. 30 (2002) e37. [15] L. Wang, A.K. Gaigalas, J. Blasic, M.J. Holden, D.T. Gallagher, Biopolym. (Biospectrosc.) 72 (2003) 401–412. [16] N. Klonis, A.H.A. Clayton, E.W.J. Voss, W.H. Sawyer, Photochem. Photobiol. 67 (1998) 500–510. [17] C.A. Parker, W.T. Rees, Analyst 85 (1960) 587–600. [18] J.N. Demas, PhD Thesis, Department of Chemistry, University of New Mexico, Albuquerque, 1970. [19] R.A. Velapoldi, M.S. Epstein, Lumin. Appl. Bio. Chem. Environ. Hydrol. Sci. 383 (1989) 98–126. [20] G.K. Turner, Science 146 (1964) 183–189. [21] N. von Ahsen, E. Schutz, Rapid cycle real-time PCR, in: S. Meuer, C. Wittwer, K. Nakagawara (Eds.), Methods and Applications, Springer, 1999, pp. 43–56. [22] H.T. Allawi, J.J. SantaLucia, Biochemistry 36 (1997) 10581–10594. [23] L. Wang, A.K. Gaigalas, F. Abbasi, G.E. Marti, R.F. Vogt, A. Schwartz, J. Res. Natl. Inst. Stand. Technol. 107 (2002) 339–353. [24] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, second ed., Plenum Press, New York, 1999. [25] G. Vamosi, C. Gohlke, R.M. Clegg, Biophys. J. 71 (1996) 972–994. [26] L. Edman, U. Mets, R. Rigler, Proc. Natl. Acad. Sci. U.S.A. 93 (1996) 6710–6715. [27] S. Wennmalm, L. Edman, R. Rigler, Proc. Natl. Acad. Sci. U.S.A. 94 (1997) 10641–10646. [28] C. Eggeling, J.R. Fries, L. Brand, R. Gunther, C.A.M. Seidel, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 1556–1561. [29] M. Zuker, Nucl. Acids Res. 31 (2003) 3406–3415.