Sensing Metal Ions with DNA Building Blocks: Fluorescent Pyridobenzimidazole Nucleosides

Abstract

We describe novel fluorescent N-deoxyribosides (1 and 2) having 2-pyrido-2-benzimidazole and 2-quino-2-benzimidazole as aglycones. The compounds were prepared from the previously unknown heterocyclic precursors and Hoffer’s chlorosugar, yielding alpha anomers as the chief products. X-ray crystal structures confirmed the geometry, and showed that the pyridine and benzimidazole ring systems deviated from coplanarity in the solid state by 154° and 140°, respectively. In methanol the compounds 1 and 2 had absorption maxima at 360 and 370 nm respectively, and emission maxima at 494 and 539 nm. Experiments revealed varied fluorescence responses of the nucleosides to a panel of seventeen monovalent, divalent and trivalent metal ions in methanol. One or both of the nucleosides showed significant changes with ten of the metal ions. The most pronounced spectral changes for ligand-nucleoside 1 included red shifts in fluorescence (Au⁺, Au³⁺), strong quenching (Cu²⁺, Ni²⁺, Pt²⁺), and in substantial enhancements in emission intensity coupled with redshifts (Ag⁺, Cd²⁺, Zn²⁺). The greatest spectral changes for ligand-nucleoside 2 included a redshift in fluorescence (Ag⁺), a blueshift (Cd²⁺), strong quenching (Pd²⁺, Pt²⁺), and in substantial enhancements in emission intensity coupled with a blueshift (Zn²⁺). The compounds could be readily incorporated into oligodeoxynucleotides, where an initial study revealed that they retained sensitivity to metal ions in aqueous solution, and demonstrated possible cooperative sensing behavior with several ions. The two free nucleosides alone can act as differential sensors for at multiple metal ions, and they are potentially useful monomers for contributing metal ion sensing capability to DNAs.

Introduction

Although in nature DNA acts primarily as a storehouse of genetic information, recent studies have been aimed at endowing new properties on this genetic molecule. Among these properties is the ability to act as a sensor of physical conditions or molecular species. Because DNA has the innate ability to recognize other nucleic acids, many studies have reported on the use of modified DNAs as sensors of RNA and DNA.1a–n However, DNA (especially in the single-stranded form) can fold into diverse structures that allow for recognition of varied types of molecules,2a–e and many changes can be made to add functional groups of DNA to increase chemical diversity further.3a–d As a result, an increasing number of studies have focused on the use of modified DNA in detection of other species, such as peptides,4 proteins,5a,b and small molecules,6a–c and sensing physical changes such as pH7 and light.8 A few experiments have been aimed at sensing metal ions with DNAs as well.9a–d The properties of DNA as a sensor offer a number of advantages in application over other organic scaffolds; among these are water solubility, ease of synthesis (including automated methods), and predictable assembly and recognition properties.

As part of a larger program aimed at development of water-soluble, diverse fluorescent species, we have adopted a strategy of replacing DNA bases with varied fluorescent groups.10a–d We considered the possibility that adding metal ligand functionality to fluorescent systems might render such species able to recognize and report on metal ions in solution. In this regard, a number of laboratories have recently described deoxyribosides with metal ligands as replacements for nucleobases,11a–h but these were designed for crosslinking nucleic acid strands, and use in ion detection was not reported.

Here we report on two new nucleosides in which the nucleobase is replaced by pyridobenzimidazoles that are designed to act as metal ligands. Their preparation, structure, and optical properties are described. The nucleosides are fluorescent, and were found to be sensitive to the presence of twelve different metal ions, displaying significant changes in their emission spectra, both as free nucleosides and in the context of DNA. The compounds are expected to be useful in lending sensing properties to DNA.

Results

Nucleoside design and synthesis

Compounds 1 and 2 are bifunctional, and possess both the structure of a DNA building block (i.e., deoxyribose) and structures designed to bind metals with conjugated systems (Chart 1). We chose these pyrido- (and related quino-) benzimidazole heterocycles (3 and 4), both of which were previously unknown, because simple pyridobenzimidazoles were previously known as ligands for metals,12 because they have a rotatable bond that alters conjugation, and because they were known to be fluorescent. We added a dimethylamino group to the benzimidazole portion, placing this donor in conjugation with the pyridine ring nitrogen. Thus the act of chelating a metal ion might be expected to alter conjugation by changing the dihedral angle of the single bond, which would alter the donation by the amino group. In addition, varied metals in the act of binding the ligand could also further alter the electronic states of the conjugated systems. Thus we anticipated the possibility that these somewhat generic nitrogen ligands might yield diverse signals for different metal ions as a result of these variable electronic and structural factors.

Chart 1.

Structures 1 and 2

We prepared 1 and 2 starting with commercial 5-chloro-2-nitroaniline (Scheme 1). The chloride was displaced by N,N-dimethylamine. Reduction of the nitro group to the amine, followed by subsequent oxidative condensation with pyridine 2-carboxaldehyde in nitrobenzene yielded the heterocycle 3, and with quinoline 2-carboxaldehyde13 yielded 4. These were reacted with Hoffer’s alpha-chlorosugar14 in the presence of sodium hydride in N,N-dimethylacetamide. The chief products were assigned as alpha-anomers by proton NOE measurements. This stereochemical outcome is atypical for this reaction; we hypothesize that the steric bulk of the 2-aryl-benzimidazoles was better accommodated from the alpha face of the bis-toluoyl ester intermediate, yielding the alpha isomer in an S_N1 mechanism. Although there are two possible benzimidazole isomers for each nucleoside, we only observed the isomers in which the dimethylamino group is closer to the sugar; the origin of this preference is not yet clear, but may be related to the electron donation from this amino group into the imidazole ring. The ester protecting groups were subsequently removed under basic conditions. We were able to obtain single crystals of the free nucleosides 1 and 2 by crystallization from chloroform/methanol/hexane (Figure 1). The x-ray structures confirmed the α-anomeric geometry, and showed that the pyridyl(quinolinyl) and benzimidazolyl rings deviated from syn coplanarity in the solid state by 154° and 140°, respectively. This apparent near-anti conformational preference may arise from avoidance of lone pair repulsions from the two nitrogens, and/or from possible steric repulsion of the pyridine protons away from the deoxyribose substituent.

Scheme 1.

Synthesis of deoxyribosides 1 and 2

Figure 1.

Structures of deoxynucleosides 1 (A) and 2 (B) in the solid state, determined by X-ray crystallography, and shown in ORTEP form. Details are given in Supporting Information.

Nucleoside optical properties

We examined the optical properties of the free nucleosides in solution. They were poorly soluble in water, and were dissolved in methanol and ethyl acetate for the measurements. In MeOH, both compounds showed small absorption maxima at 260 nm; 1 also had a major longer-wavelength absorption band at 360 nm, while 2 had an absorption at 370 nm (Table 1 and Supporting Information). Emission spectra were measured with excitation at the long-wavelength bands; they revealed fluorescence in the visible ranges for both compounds (Figure 2), with emission maxima at 494 nm and 539 nm for 1 and 2, respectively. Quantum yields for these emission bands were 0.052 and 0.0085, respectively. The emission maxima in methanol were redshifted relative to the corresponding maxima in lower-polarity ethyl acetate solution (464 and 520 nm, respectively), consistent with formation of a strong excited-state dipole due to resonance overlap of the dimethylamino group with the pyridine ring nitrogen.

Table 1.

Optical data for nucleosides 1 and 2

Compound	Solvent	absorbance maximuma (nm)	ε (M−1cm−1)	emission maximum (nm)	Φfl	τ (ns)b
1	EtOAc	360	nd	460	nd	nd
1	MeOH	360	1.37 × 104	494	0.050	0.45
2	EtOAc	370	nd	521	nd	nd
2	MeOH	384	1.36 × 104	543	0.0085	0.36

^aLong wavelength band is listed; shorter wavelength bands at 260nm were also present.

^bConditions: 10 μM nucleoside in methanol, 25°C, with excitation at 370 nm. Compound. 1 was measured at 450 nm with 430 nm U.V. filter; compound 2 was measured at 550 nm with 515 nm U.V. filter.

Figure 2.

Excitation and emission spectra of 1 (A) and 2 (B) (2 μM) in methanol (25°C). Excitation spectra are shown with dashed lines and were monitored at the emission maxima. Emission spectra are shown with solid lines, and were measured with excitation at the long-wavelength absorption maxima.

Nucleoside response to metal ions

We then investigated the effects of metal ions on the fluorescence of nucleosides 1 and 2 in methanolic solution. Nucleosides and metals were studied at 50 μM and 100 μM concentrations (respectively), and we screened the effects of seventeen metal species on the two nucleosides by monitoring emission spectra. Note that the ligands could in principle bind with varied stoichiometry with a given metal; some may bind as 1:1 ligand:metal complexes while other combinations may form complexes with other stoichiometry. Metals (as their chloride salts) were added from concentrated stock solutions in DMSO, giving a final concentration of 5% DMSO in methanol. Results with compound 1 showed (Figure 3, Table 2, and Supporting Information) that a few metal ions (K²⁺, Mg²⁺, Ca²⁺, Mn²⁺, La²⁺, Eu²⁺) had little or no effect on the emission from the pyridobenzimidazole. Co²⁺ and Pd²⁺ showed weak quenching of this fluorophore, while Pt²⁺, Cu²⁺, Ni²⁺ showed apparently strong quenching. In contrast to this, both Au⁺ and Au³⁺ gave marked redshifts in emission (shifts of 55 and 37 nm) with little effect on intensity. Finally, Ag⁺, Cd²⁺, and Zn²⁺ yielded both redshifts and marked increases in emission intensity; redshifts increased in the order Ag⁺ (26 nm shift with 3-fold increase in intensity) < Cd²⁺ (59 nm with 8-fold increase) < Zn²⁺ (79 nm with 3-fold increase). The effects are summarized in Table 2.

Figure 3.

Fluorescence response of nucleoside 1 to 17 metal salts in methanolic solution (25 °C). A. Photograph showing varied responses of 1 to the metals, using 365 nm light from a UV box as excitation source. B. Spectra of 1 (2 μM) in the presence of metal ions that change the emission intensity or wavelength of the nucleoside. C. Spectra of 1 with metal ions that yield little or no change in the emission properties. Metals are present at 100 μM as chloride salts except AgNO₃.

Table 2.

Fluorescence responses of nucleosides 1 and 2 to metal salts in methanolic solution.

Compound	Mn+	emission maximum (nm)	Intensity (rel)	Compound	Mn+	emission maximum (nm)	Intensity (rel)
1	none	494	1.0	2	none	543	1.0
1	K+	494	1.1	2	K+	544	1.1
1	Mg2+	494	1.2	2	Mg2+	543	1.1
1	Ca2+	494	1.1	2	Ca2+	541	1.0
1	Mn2+	494	1.0	2	Mn2+	540	1.0
1	Co2+	494	0.6	2	Co2+	539	0.9
1	Ni2+	496	0.1	2	Ni2+	542	0.9
1	Cu2+	486	0.1	2	Cu2+	548	0.6
1	Zn2+	573	2.9	2	Zn2+	476	2.0
1	Pd2+	503	0.7	2	Pd2+	581	0.2
1	Ag+	520	3.4	2	Ag+	579	1.0
1	Cd2+	553	8.3	2	Cd2+	510	1.2
1	Pt2+	508	0.3	2	Pt2+	563	0.3
1	Au+	549	1.1	2	Au+	577	0.8
1	Au3+	531	0.9	2	Au3+	552	0.8
1	La3+	498	1.0	2	La3+	552	1.0
1	Eu3+	496	1.0	2	Eu3+	548	1.0
1	Hg2+	498	0.9	2	Hg2+	540	0.8

^aConditions: 50 μM nucleoside, 100 μM metal ion, 95% methanol, 5% DMSO, 25 °C. All counterions were chloride except AgNO₃.

Metal ion effects on quinobenzimidazole 2 were also measured (Figure 4, Table 2, and Supporting Information), and revealed both similarities and differences with 1. With 2, eight metal ions (K⁺, Mg²⁺, Ca²⁺, Hg²⁺, Mn²⁺, La²⁺, Eu²⁺) had little or no effect on the emission from ligand 1. Co²⁺, and Au³⁺, Ni²⁺ showed slight quenching of this fluorophore, while Cu²⁺ showed moderate quenching in addition to a 5 nm redshift. Addition of Pd²⁺ and Pt²⁺ gave apparently strong quenching coupled with redshifts (shifts of 38 and 20 nm, respectively). In contrast to this, Ag⁺ gave a 36 nm redshift with little effect on intensity. Finally, unlike ligand 1, ligand 2 yielded blueshifts of the emission band in two cases. Cd²⁺ yielded a shift from the original 543 nm emission to dual emission at 460 and 510 nm with little change in intensity, while Zn²⁺ yielded a shift to 476 nm and a substantial (2-fold) increase in emission intensity.

Figure 4.

Fluorescence response of nucleoside 2 to 17 metal salts in methanolic solution (25 °C). A. Photograph showing varied responses of 2 to the metals, using 365 nm light from a UV box as excitation source. B. Spectra of 2 (2 μM) in the presence of metal ions that change the emission intensity or wavelength of the nucleoside. C. Spectra of 2 with metal ions that yield little or no change in the emission properties. Metals are present at 100 μM as chloride salts except AgNO₃.

A summary of the results for both fluorescent nucleoside ligands is outlined in Table 2, and points out the similarities and differences. As free nucleosides in methanol, neither compound showed significant changes with seven ions (K²⁺, Mg²⁺, Ca²⁺, Hg²⁺, Mn²⁺, La²⁺, Eu²⁺). Both compounds showed similar changes for Co²⁺ (slight quenching) and for Pt²⁺ (strong quenching with redshift). Interestingly, although they are similar ligands, the compounds showed different responses for the other eight metal ions. The most dramatic differences occurred with Ag²⁺ (>3-fold enhancement of fluorescence with 1 versus a redshift with 2), and especially with Zn²⁺ and Cd²⁺, which showed marked redshifts and fluorescence enhancements with 1 but strong blueshifts with 2.

Responses of ligand pairs in DNA

We then investigated the incorporation of fluorescent nucleosides 1 and 2 into DNA oligonucleotides. Both compounds were converted to the 5′-dimethoxytrityl, 3′ phosphoramidite derivatives using standard methods (Supporting Information). They were then incorporated into two 13mer complementary oligonucleotides using automated synthesizer coupling programs. The presence of the intact ligands in the DNAs was confirmed by MALDI-TOF mass spectrometry. We also prepared complementary DNAs in which A, C, G, or T was paired opposite the ligands; these showed little or no differential effect on fluorescence emission (data not shown).

To investigate whether the fluorescence and recognition properties of these ligands were retained in water, we performed a screen of fluorescence responses of 13mer DNA duplexes in which nucleosides 1 and 2 were paired opposite themselves or each other (Figure 5). Four pairing geometries were tested: 1-1, 2-2, and the similar (but not identical) 2-1 and 1-2 cases. In the DNA context, the nucleoside 1 (as a 1-1 pair) showed only a small change (a 5 nm redshift) compared to its behavior in methanol as a free nucleoside. The nucleoside 2, on the other hand, showed a marked change, with its major emission band appearing in the blue (440 nm) in DNA as a 2-2 pair, whereas alone in methanol, 2 emits in the yellow range (543 nm). For the mixed 1-2 and 2-1 pairings, the emission behavior in the absence of metals resembled that of the 2-2 pairing more than the 1-1 pairing (Supporting Information), suggesting that 2 may quench the emission of 1. However, in the presence of metals the responses were more similar to those of the 1-1 pairing (see below).

Figure 5.

Fluorescence responses of 1 and 2 to metal ions in aqueous solution, measured in the context of 13mer duplex DNAs having ligands 1 and/or 2 placed opposite each other, as pseudo-base-pairs. A. Emission spectra of DNA with nucleoside 1 paired opposite itself, in the presence of varied metal ions. B. Emission spectra of DNA with ligand 2 paired opposite itself. See Figure S2 for the cases with ligand 2 paired with 1. Solutions contained 3 μM DNA duplex and 100 μM metal ion at 25 °C. Buffers also contained 100 mM NaCl and 10 mM Na•PIPES (pH 7.0). Excitation was at 370 nm.

We then screened these four different ligand “base pairs” for their responses to the addition of metal ions (Figure 5). In some cases, we found roughly parallel responses to those seen for the free nucleosides in methanol; for example, the 1-1 pairing showed strong quenching by Pd²⁺ and Pt²⁺, and displayed enhancements of fluorescence and redshifts with Zn²⁺ and Cd²⁺, similar to what was observed for 1 alone in methanol. In another example, Cd²⁺ gave essentially the same results whether with 2 alone or with the 2-2 pairing in DNA. For the mixed 1-2 and 2-1 pairings, the emission behavior in the presence of metals resembled the results with the 1-1 pairing more than the 2-2 pairing.

However, there were some marked differences between the fluorescence emission behavior in the DNA setting and that of the ligand nucleosides alone. For ligand 1, Cd²⁺, Ag⁺ and Zn²⁺ yielded considerably greater enhancements of fluorescence with the 1-1 pairing in DNA, giving 20-, 15-, and 11-fold overall enhancements (respectively) in emission intensity. Au³⁺ yielded a redshift in methanol, whereas with 1-1 in DNA it quenched fluorescence. In another example, Pd²⁺ did not quench 1 alone, whereas it strongly quenched 1-1 in DNA. For ligand 2, Zn²⁺ yielded a new band in the red for the 2-2 pairing in DNA, whereas alone in methanol it gave enhancement plus a blueshift. The Ag⁺ ion yielded dual bands at 440 and 552 nm with 2-2 in DNA, whereas with 2 in methanol it yielded only a 36 nm redshift.

In a few other cases the ligand-nucleosides showed additional striking evidence of apparently cooperative behavior in a “paired” configuration (see Fig. S2 in Supporting Information). For example, the 1-1 pairing yielded enhanced signals in the presence of Eu³⁺, whereas 1 alone in methanol had shown little or no effect with this lanthanide. In a second example, La³⁺ yielded a strong change for the 2-2 pairing, with a new red band appearing at 541 nm, whereas 2 alone in methanol yielded no change with this ion. In a third example, the 2-1 pairing showed a non-quenched redshift with Ni²⁺, whereas this metal strongly quenched 1 alone in methanol and had no effect on 2. Finally, Hg²⁺, another ion to which neither 1 nor 2 responded alone, yielded strong quenching in DNA containing the 2-2 pairing.

Discussion

These early experiments show that combinations of one or both of these nucleoside ligands, either as free nucleosides or in DNA, could sense 12 of the 17 metal ions in this initial panel of salts, and allow one to distinguish between all 12 (except Pd²⁺ and Pt²⁺, which give similar quenching effects with both ligands). We expect that the diverse responses arise from the combination of the rotatable bond, the electron donating amino group in conjugation with the pyridine nitrogen, and variable electronic interactions of the ligands with the individual metals. Further experiments will be needed to understand the details of these mechanisms for the individual ligand-ion combinations.

With the current ligand design, which employs bipyridyl-like structure, it is not surprising that the fluorescence responses are weak or absent for the more oxophilic metals in this panel of analytes, such as K⁺, Mg²⁺, Ca²⁺, Hg²⁺, and Mn²⁺. These metals are expected to have weak affinity for the current nitrogen ligands; future nucleoside designs for sensing these species may require different liganding groups. However, our data for the 2-2 pairing in DNA show that this combination of two ligands can apparently act cooperatively to recognize and give a response signal for Hg²⁺. Beyond the current studies, we do expect that the current nucleosides may be able to respond to the presence of other cations not included in this initial panel, and future studies will address this. One limitation should be noted for these nucleosides; namely, that they are only weakly water soluble, and are thus limited in their aqueous applications unless organic cosolvent can be added. Future studies will address whether more polar derivatives can be prepared. Of course, in the DNA context (see below) water solubility is no longer problematic.

As a whole, the fluorescence responses of nucleosides 1 and 2 are remarkably diverse. Nucleoside 1 in methanol gives emission maxima over an 87 nm wavelength range, and a >80-fold intensity range, depending on the metal ion. Nucleoside 2 gives a similar 85 nm emission wavelength range, yielding a redshifted emission with Ag⁺ and a blueshifted emission with Cd²⁺. Overall, ligand 1 gives at least five qualitatively different responses (strong quenching, weak quenching, strong emission enhancement, redshift, multiple emission bands) for different ions, and ligand 2, differing from 1 only by benzo substitution, also gives at least five qualitatively different responses, in some cases markedly different than 1. Taken together, the data suggest that monomer 1 can act as a useful “turn on” sensor of Ag⁺, Cd²⁺, Zn²⁺, and Au⁺ ions, and the responses to these are different enough to allow separate identification of these metals in an unknown sample. Because the responses to Au⁺ and Au³⁺ are similar, 1 would not be useful in discriminating between these latter two. In the quenching mode, both Ni²⁺ and Cu²⁺ give very strong quenching of this nucleoside, allowing one to discriminate between this pair and the above fluorescence-enhancing ions. Monomer 2, like 1, can also be used as enhanced-signal reporters of Zn²⁺ and Cd²⁺, and it also gives a marked positive response for Ag⁺ and Au⁺, although it is difficult to distinguish between these last two. Unlike 1, 2 can distinguish between Au⁺ and Au³⁺, by a 25 nm difference in wavelength. In quenching applications, ligand 2 is not as strongly quenched by the metals in this panel as 1, and would not be as useful as the latter ligand.

In addition to fluorescence responses involving increase or decrease in intensity, these ligand nucleosides in a number of cases change their emission wavelengths markedly in the presence of some metal ions. This is especially useful for two reasons: first, it allows for ratiometric measurements of concentration, as the ratio of two wavelengths can easily be measured. Second, it allows for distinguishing between background signal and positive signal in applications where separating the reporter from the analyte is undesirable or impossible, such as in real-time solution-phase assays and in intracellular assays. Nucleoside 1 shows its most marked redshifts in wavelength for Cd²⁺ (59 nm), Zn²⁺ (79 nm), and Au⁺ (54 nm), and thus might in principle be used ratiometrically for these. Nucleoside 2 shows the greatest shifts for Zn²⁺ (67 nm) and Cd²⁺ (33 nm) (these are blueshifts rather than redshifts).

A comparison of the fluorescence responses of 1 and 2 with other recently reported sensors reveals some similarities and differences. There exist a substantial number of reports of sensors for Zn²⁺,15 including some that respond with a single wavelength, and a few that give a change in wavelength,15b,d,g which is useful for ratiometric measurements. Compounds 1 and 2 fall into the latter category; it is not yet clear whether they would be useful in detecting Zn²⁺ in cellular media, as their cell permeability and affinity for low biological concentrations of this ion are not yet known. Compounds 1 and 2 also detect Cd²⁺, Au⁺, and Ag⁺ with enhancements of emission and wavelength shifts; although Zn²⁺ sensors often respond to Cd²⁺ as well, compound 1 is unusual in that it can readily distinguish between these two closely related ions by wavelength. Finally, we have shown that base pairs involving these nucleosides can also respond to lanthanides with wavelength shifts. Thus although metal ion sensing (especially of Zn²⁺) is by now well established in the literature, the diversity of responses of these two nucleoside-ligands with a range of metal ions is rare.

Our data show that these nucleosides can be readily used as monomers in the synthesis of DNA, and that they retain metal ion recognition properties in aqueous buffers and in the context of the DNA double helix. We have shown that pairing of two nucleoside ligands opposite one another can in some cases yield new recognition and sensing properties that the monomers alone do not have. A number of laboratories have described the strategy of assembling base pairs in which a metal ion crosslinks two paired ligands as DNA base replacements.11a–h The current studies build on that concept by showing a fluorescence reporting response as a result of that apparent interaction. Examples of responses that may be specially useful in application are those of the 1-1 pairing, which gives strong fluorescence enhancements and redshifts with Cd²⁺, Ag⁺, and Zn²⁺; it is even possible to distinguish between these three by wavelength. The 2-2 pairing may also be especially useful in sensing of Cd²⁺ and Zn²⁺, although these cannot be distinguished from one another. More useful may be the application of 2-2 in sensing of Ag⁺ and La³⁺, where the original blue short-wavelength band is strongly quenched, and new long-wavelength yellow bands appear. These findings, coupled with those of the nucleosides alone, suggest that DNAs containing 1 and 2 might have useful ion sensing applications both in single-stranded and double-stranded forms. Future studies will investigate this in detail.

In summary, the present compounds show promise not only as sensors alone, but also as components for introducing metal ion sensing capability into DNA. There is now a broad number of literature reports describing fluorescence sensing of metal ions by ligands in solution,15,16 but few if any of these are directly adaptable to DNA.16h We anticipate that the use of DNA as a scaffold for this fluorescence sensing will be advantageous by adding aqueous solubility to these heterocyclic species, thus allowing possible applications in biological systems and in aqueous waste streams. In addition, the use of DNA double-helical or folded structure may allow for useful cooperative sensing between multiple ligands, as our preliminary studies suggest. Finally, the DNA portion of such molecules could be useful for self-assembly of sensors on surfaces and in arrays.

Experimental Section

Nucleoside synthesis

5-N,N-Dimethylamino-2-(2-pyridinyl)-1H-benzimidazole (3)

Nitrobenzene (200 mL) was added to 4-N,N-dimethylamino-phenylenediamine (19.2 mmol) to make a solution in a round-bottom flask. 2-Pyridinecarboxaldehyde (1.83 mL, 19.2 mmol) was added and then the solution was slowly heated to 100 °C. After 1 h stirring, the reaction mixture was further heated to 150 °C and stirred for 2 additional hours. This solution was cooled to room temperature and directly loaded on silica gel column for chromatographic purification (eluent CH₂Cl₂ / hexane (1:1) followed by CH₂Cl₂ / EtOAc (2:3)). The product was obtained as dark yellow powder (3.57 g, 15.0 mmol, 78% yield over 2 steps); ¹H-NMR (400 MHz, DMSO-d₆): δ = 8.64 (d, J = 5.2 Hz, 1H), 8.20 (d, J = 8.0 Hz, 1H), 7.92 (t, J = 8.0 Hz, 1H), 7.46-7.40 (m, 2H), 6.82-6.78 (m, 2H), 2.88 (s, 6H). ¹³C-NMR (100.7 MHz, DMSO-d₆): δ = 149.9, 149.8, 149.3, 148.8, 138.1, 124.8, 121.5, 112.1, 96.9, 41.9. HRMS (DEI): [M ]⁺ calcd for C₁₄H₁₄N₄, m/z = 238.1218; found, 238.1226.

5-N,N-Dimethylamino-2-(2-quinolinyl)-1H-benzimidazole (4)

Nitrobenzene (250 mL) was added to 4-N,N-dimethylamino-phenylenediamine (27.1 mmol) to make a solution in a round-bottom flask. 2-Quinolinecarboxaldehyde (4.14 g, 26.3 mmol) was added and then the solution was slowly heated to 80 °C. After 1 h stirring, the reaction mixture was further heated to 150 °C and stirred for 24 additional hours. This solution was cooled to room temperature and directly loaded onto a silica gel column for chromatographic purification (eluent CH₂Cl₂ / hexane (1:1) followed by CH₂Cl₂ / EtOAc (1:1)). The product was obtained as orange yellow powder (4.86 g, 16.9 mmol, 62% yield over 2 steps); ¹H-NMR (400 MHz, CDCl₃) δ = 8.56 (d, J = 8.8 Hz, 1H), 8.25 (d, J = 8.4 Hz, 1H), 8.09 (d, J = 8.4 Hz, 1H), 7.83 (d, J = 8.0 Hz, 1H), 7.71–7.65 (m, 2H), 7.51 (t, J₁ = 7.3 Hz, 1H), 6.81 (dd, J₁ = 9.2, J₂ = 2.4 Hz, 1H), 6.28 (s, 1H), 2.84 (s, 6H). ¹³C-NMR (100.7 MHz, CDCl₃): δ = 149.4, 149.2, 149.1, 147.8, 137.4, 137.3, 136.0, 129.1, 128.5, 128.1, 127.1, 120.6, 119.4, 111.6, 93.5, 41.5; HRMS (DEI): [M]⁺ calcd for C₁₈H₁₆N₄: 288.1375; found 288.1375.

1-(5-N,N-Dimethylamino-2-(2-pyridinyl)-benzimidazole)-1′-a-deoxyriboside (1)

5-N,N-Dimethylamino-2-(2-pyridinyl)-1H-benzimidazole (1.1 g, 4.54 mmol) and NaH (0.20 g, 4.99 mmol) were charged into 100 mL round-bottom flask. 15 mL of N,N-dimethylacetamide was added into the flask via syringe under an atmosphere of argon. The reaction mixture was stirred at room temperature for 30 min, then Hoffer’s chlorosugar (2.12 g, 5.39 mmol) was added in one portion and the reaction mixture was stirred for 2 h at room temperature. The solvent was evaporated under vacuum. The crude mixture was extracted with water and CH₂Cl₂. The organic layer was dried with Mg₂SO₄ and evaporated under reduced pressure. The crude product was dissolved in 30 mL CH₂Cl₂/MeOH (1:1), and then 4.5 mL of NaOMe (0.5 N in MeOH) was added into the flask via syringe under an atmosphere of argon. The reaction mixture was stirred at room temperature for 1 h. The solvent was evaporated under reduced pressure. The crude products were purified by silica gel chromatography (eluting with hexane/ethyl acetate 1:1 to 1:3) to give the desired product (562 mg, 1.59 mmol, 35% yield); ¹H-NMR (400 MHz, CDCl₃/CD₃OD (1:1)) δ = 8.65 (d, J = 4.8 Hz, 1H), 8.03 (d, J = 8.0 Hz, 1H), 7.90 (td, J₁ = 8.0, J₂ = 2.0 Hz, 1H), 7.56 (d, J = 8.4 Hz, 1H), 7.41 (dd, J₁ = 7.6, J₂ = 5.2 Hz, 1H), 7.34 (d, J = 2.4 Hz, 1H), 7.12 (t, J = 8.0 Hz, 1H), 6.94 (dd, J₁ = 9.2, J₂ = 2.4 Hz, 1H), 4.55-4.50 (m, 1H), 4.31-4.27 (m, 1H), 3.75 (dd, J₁ = 12.0, J₂ = 3.6 Hz, 1H), 3.65 (dd, J₁ = 12.0, J₂ = 4.8 Hz, 1H), 3.02 (s, 6H), 2.96-2.88 (m, 1H), 2.72-2.65 (m, 1H). ¹³C-NMR (100.7 MHz, CDCl₃/CD₃OD (1:1)): δ = 150.0, 149.1, 148.9, 148.8, 137.5, 135.2, 134.6, 125.1, 124.1, 119.5, 112.4, 96.9, 87.2, 86.8, 71.0, 62.2, 41.4, 39.2. MS (EI): m/z = 355.4 [M + H]⁺. HRMS (DCI): [M + H]⁺ calcd for C₁₉H₂₃N₄O₃, m/z = 355.1770; found, 355.1780.

1-(5-N,N-Dimethyl amino-2-(2-quinolinyl)-benzimidazole)-1′-a-deoxyriboside (2)

5-N,N-Dimethyl amino-2-(2-quinolinyl)-1H-benzimidazole (1.13 g, 3.9 mmol) and NaH (264 mg, 6.6 mmol) were charged into a 100 mL round-bottom flask. 20 mL of N,N-dimethylacetamide was added into the flask via syringe under an atmosphere of argon. The reaction mixture was stirred at room temperature for 30 min, then solvent was evaporated under vacuum. The crude mixture was extracted with water and CH₂Cl₂. The organic layer was dried with Mg₂SO₄. The crude products were purified by silica column chromatography (eluting with hexane/ethyl acetate/dichloromethane 2:1:2 to 1:1:1) to give the desired product (971.4 mg, 1.52 mmol, 39% yield). This bis-toluoyl-protected deoxyriboside (971.4 mg 1.52 mmol) was dissolved in 50 mL CH₂Cl₂ / MeOH (1:1), and then 10 mL NaOMe solution (0.5 N in MeOH) was added into the flask via syringe under an atmosphere of argon. The reaction mixture was stirred at room temperature for 1 h. The solvent was then evaporated under reduced pressure, and the crude products were purified by silica colum chromatography (eluting with hexane/ethyl acetate 1:1 to 1:3) to give the desired product (541 mg, 1.34 mmol, 88% yield); ¹H-NMR (400 MHz, DMSO-d₆) δ = 8.45 (d, J = 8.8 Hz, 1H), 8.30 (d, J = 8.8 Hz, 1H), 8.06 (d, J = 8.4 Hz, 1H), 7.99 (d, J = 7.6 Hz, 1H), 7.80 (t, J = 8.0 Hz, 1H), 7.66–7.55 (m, 3H), 7.33 (s, 1H), 6.88 (dd, J₁ = 8.8, J₂ = 2.4 Hz, 1H), 5.63 (br, 1H), 4.95 (t, J = 5.2 Hz, 1H), 4.45 (br, 1H), 3.58-3.47 (m, 2H), 3.04-2.93 (m, 7H), 2.49-2.43 (m, 1H). ¹³C NMR (100.7 MHz, DMSO-d₆): δ = 150.7, 148.7, 148.3, 147.1, 137.6, 135.9, 135.7, 131.0, 129.8, 128.6, 128.1, 127.7, 122.3, 120.6, 112.0, 96.8, 88.2, 87.3, 71.4, 62.5, 41.6, 39.2. MS (EI): m/z = 405.2 [M + H]⁺. HRMS (DCI): [M + H]⁺ calcd for C₂₃H₂₅N₄O₃, m/z = 405.1927; found, 405.1924.

Oligonucleotide synthesis

We prepared the 5′-O-DMTr protected 3′-2-cyanoethyl phosphoramidite building blocks of 1 and 2 following standard methods (see Supporting Information). We directly applied them using solid-phase oligodeoxynucleotide (ODN) synthesis protocols on an automated DNA synthesizer (Applied Biosystems 394 DNA/RNA synthesizer). All oligodeoxynucleotides were synthesized on 1.0 μmole scale using standard β-cyanoethyl phosphoramidite chemistry, but with extended (200 s) coupling times, double coupling and high concentration (0.13 mM) for modified nucleosides. Stepwise coupling yields for nonnatural nucleosides were all greater than 95% as determined by trityl monitoring. The synthesized oligonucleotides were cleaved from the solid support by treatment with 30% aqueous NH₄OH (1.0 mL) for 10 h at 55 °C. The crude products from the automated ODN synthesis were lyophilized and diluted with distilled water (2 mL). The ODNs were purified by HPLC (Zorbax ODS column, 9.4 × 250 mm). The fractions containing the purified ODN were pooled and lyophilized. Aqueous AcOH (80%) was added to the ODN for deprotection of the 5′-terminal DMT. After 30 min at ambient temperature, the AcOH was evaporated under reduced pressure. The residue was diluted with water (2 mL), and the solution was purified by HPLC again. Modified oligonucleotides in the study were characterized by matrix-assisted laser-desorption-ionization time-of-flight (MALDI-TOF) mass spectrometry (see Supporting Information).

Fluorescence measurements

Fluorescence measurements were carried out on a Spex Fluorolog 3 fluorescence spectrometer equipped with Lauda Brinkmann RM 6 temperature controller. Fluorescence quantum efficiency (ϕ_samp) was obtained by comparing the integrated fluorescence spectra of a sample (I_samp) with that of a reference sample (I_ref) having a known fluorescence quantum efficiency (ϕ_ref). We used 9, 10-diphenylanthracene (10 μM) dissolved in EtOH (ϕ_ref = 0.81, excitation at 366 nm) as a standard. Compounds 1 and 2 were prepared as 10 μM solutions in methanol. Quantum yields were calculated by the following equation:

$${\varphi }_{\text{samp}}={\varphi }_{\text{ref}}\times (A_{\text{ref}}/A_{\text{samp}})\times {({\eta }_{\text{samp}}/{\eta }_{\text{ref}})}^2\times (I_{\text{samp}}/I_{\text{ref}})$$

Absorption spectra were measured with a Cary I UV-vis spectrometer, and excitation and emission spectra were observed by Spex Fluorolog 3 fluorescence spectrometer. Fluorescence lifetimes were measured on a Photon Technologies, Inc. Easylife LS instrument; a 465 nm pulsed LED was utilized for excitation, and a 535 nm long-pass filter selected the desired emission wavelengths.

Screening of metal ions

Seventeen metal salts were tested for fluorescence responses with 1 and 2: KCl, MgCl₂, AgNO₃, NiCl₂, PdCl₂, PtCl₂, CuCl₂, ZnCl₂, CoCl₂, AuCl, LaCl₃, EuCl₃, AuCl₃, CdCl₂, HgCl₂, MnCl₂, CaCl₂. All metal salt solutions were prepared as stock solutions (5.0 mM) in DMSO. Note that AuCl had poor solubility and left some precipitate. Nucleosides 1 and 2 were separately prepared as 50 μM solutions in methanol, and each metal salt stock solution was added to give 2-fold excess metal concentration (100 μM). Photographs were taken over a UV light box (354 nm), and fluorescence spectra were measured as described above.

Crystal structure determination

Single crystals of free ligand-nucleosides 1 and 2 were grown by a bi-layer method. The compounds were dissolved in a 1:1 mixture of dry CHCl3 and MeOH. These solutions (ca. 1 mL) were divided to NMR tubes (5 × 200 mm) and hexane (ca. 2 mL) was added very slowly, making a second layer. At the junction between the sample layer (lower) and hexane (upper) layer we obtained crystals for X-ray diffraction. The X-ray data for 1 and 2 were collected on a Siemens SMART-APEX CCD single-crystal X-ray diffractometers. The structures were solved and refined by full-matrix least-squares techniques on F2 using SHELX-97. Additional details are in the Supporting Information.

Supplementary Material

1si20060301_12. Supporting Information Available.

Details of synthesis and characterization of phosphoramidite derivatives of 1 and 2; characterization of modified oligonucleotides; fluorescence data.