Optical Detection of DNA Conformational Polymorphism on Single-Walled Carbon Nanotubes

See allHide authors and affiliations

Science  27 Jan 2006:
Vol. 311, Issue 5760, pp. 508-511
DOI: 10.1126/science.1120792


The transition of DNA secondary structure from an analogous B to Z conformation modulates the dielectric environment of the single-walled carbon nanotube (SWNT) around which it is adsorbed. The SWNT band-gap fluorescence undergoes a red shift when an encapsulating 30-nucleotide oligomer is exposed to counter ions that screen the charged backbone. The transition is thermodynamically identical for DNA on and off the nanotube, except that the propagation length of the former is shorter by five-sixths. The magnitude of the energy shift is described by using an effective medium model and the DNA geometry on the nanotube sidewall. We demonstrate the detection of the B-Z change in whole blood, tissue, and from within living mammalian cells.

Single-walled carbon nanotubes (1) are rolled sheets of graphene with nanometer-sized diameters that possess remarkable photostablity (2). The semiconducting forms of SWNTs, when dispersed by surfactants in aqueous solution, can display distinctive near-infrared (IR) photoluminescence (3) arising from their electronic band gap. The band-gap energy is sensitive to the local dielectric environment around the SWNT, and this property can be exploited in chemical sensing, which was recently demonstrated for the detection of β-d-glucose (4).

Among the molecules that can bind to the surface of SWNTs is DNA, which adsorbs as a double-stranded (ds) complex (5). Certain DNA oligonucleotides will transition from the native, right-handed B form to the left-handed Z form as cations adsorb onto and screen the negatively charged backbone (69). We now show that an analogous B-to-Z transition for a 30-nucleotide dsDNA modulates the dielectric environment of SWNTs and decreases their near-IR emission energy up to 15 meV. We have used this fluorescence signal to detect divalent metal cations that bind to DNA and stabilize the Z form. The thermodynamics of the conformational change for DNA both on and off the SWNT are nearly identical. These near-IR ion sensors can operate in strongly scattering or absorbing media, which we demonstrate by detecting mercuric ions in whole blood, black ink, and living mammalian cells and tissues.

Near-IR spectrofluorometry was performed on colloidally stable complexes of DNA-encapsulated SWNTs (DNA-SWNTs) buffered at a pH of 7.4 and synthesized by the noncovalent binding to the nanotube sidewall (10) of a 30–base pair single-stranded DNA (ssDNA) oligonucleotide with a repeating G-T sequence. This ssDNA can hydrogen bond with itself to form dsDNA. Several types of semiconducting SWNTs are present, but as we show below, they can be identified by their characteristic band gaps. The shift in band gap is similar for each type of SWNT, although there is a diameter dependence. After the addition of divalent cations, we observed an energy shift in the SWNT emission with a relative ion sensitivity of Hg2+ > Co2+ > Ca2+ > Mg2+, which is identical for free DNA (Fig. 1A) (11). The shift can also be observed by monitoring SWNT photoabsorption bands (fig. S1). The fluorescence peak energy traces a monotonic, two-state equilibrium profile with increasing ionic strength for each case (12).

Fig. 1.

(A) Concentration-dependent fluorescence response of the DNA-encapsulated (6,5) nanotube to divalent chloride counterions. The inset shows the (6,5) fluorescence band at starting (blue) and final (pink) concentrations of Hg2+. (B) Fluorescence energy of DNA-SWNTs inside a dialysis membrane upon removal of Hg2+ during a period of 7 hours by dialysis. (C) Circular dichroism spectra of unbound d(GT)15 DNA at various concentrations of Hg2+. (D) DNA-SWNT emission energy plotted versus Hg2+ concentration (red curve) and the ellipticity of the 285-nm peak obtained via circular dichroism measurements upon addition of mercuric chloride to the same oligonucleotide (black curve). Arrows point to the axis used for the corresponding curve. (E) Illustration of DNA undergoing a conformational transition from the B form (top) to the Z form (bottom) on a carbon nanotube.

The removal of ions from the system via dialysis returns the emission energy to its initial value, which is indicative of a completely reversible thermodynamic transition (Fig. 1B and fig. S2) (10). Under identical conditions, circular dichroism (CD) spectroscopy confirmed that the unbound DNA undergoes a conformational change from the B to the Z form, and the inversion of the 285-nm peak indicates a reversal of helicity (Fig. 1C).

We compare the ellipticity of the 285-nm CD peak versus Hg2+ concentration with the fluorescent emission energy from the nanotube under identical conditions (Fig. 1D). Assuming identical transitions, the overlapping points of inflection indicate that the difference in the Gibbs free energy (ΔG) changes for the DNA on and off the nanotube is quite small [Δ(ΔG) ∼0.05 kBT per phosphate, where kB is Boltzmann's constant, T is temperature, and kBT is the thermal energy] (13, 14). Thus, the transitions for DNA in solution or adsorbed on the SWNT appear to be thermodynamically identical.

A critical difference is apparent between the slopes at the inflection. Pohl (15) describes the B-Z transition—which requires a double-stranded helix to separate, change helicity, and re-form—as a process of nucleation and propagation in series. The dsDNA strand initially separates with a ratio of rate constants βBZ, whereas propagation proceeds as a series of equilibrium steps proportional to the number of base pairs, N, as the dislocation proceeds down the chain (10). The expression for the fractional transition K (6, 15) contains a scaling factor C0, which is the ion concentration (C) where K is independent of N Math The slope at the inflection is related to the propagation length aN, where a is the ratio of binding sites to oligonucleotide length. Regression of the data in Fig. 1D reveals that DNA on the nanotube precedes through only Math the number of transitions, as in the case of the free strand. As expected, βBZ, which is associated with the initiation of the event, is similar for the cases on and off the nanotube (1.21 and 1.04, respectively). If we assume that the same helicity change occurs both on and off the nanotube, our interpretation is one of a transition that propagates in small steps and requires about Math radians of the strand to unravel for propagation down the nanotube (Fig. 1E and fig. S3).

Examining this phenomenon for SWNTs of different diameters allows us to probe the influence of the cylindrical geometry. Perebeinos and co-workers (16) used a numerical solution to the Bethe-Salpeter equation (17) to yield a scaling relationship for the exciton binding energy Math, where μ is the reduced effective mass (18), rt is the nanotube radius, and e is the dielectric constant around the nanotube. The constants A and n were determined by fitting nanotubes with diameters in the range of 1 to 2.5 nm and were found to be 24.1 eV nm3/5 and 1.4, respectively. With this scaling, the change in emission energy from the B to Z form for a DNA wrapped nanotube is then Math(1) Approximating the dielectric constant of the B or Z wrapped nanotube using an effective medium gives Math(2) Here, ϵDNA and ϵWater are the dielectric constants of DNA (4.0) and water (88.1), and αi is the ratio of surface area covered by DNA per total area, which increases in transitioning from the B to Z form (19). To relate αi to the geometry of the adsorbed phase, we consider a helical surface described by three parameters: radius r; pitch b; and width of the strand w Math(3) By describing the mechanics of DNA as a continuum helix (20) maintaining an equilibrium curvature (21), one can then describe the total energy in terms of its deflection. Adsorbing the DNA to a nanotube of radius rt perturbs it from the equilibrium radius, ro, and pitch, bo. The resulting pitch that minimizes the total energy is Math(4) and the surface area is Math(5)

For B DNA, ro and bo are 1 and 3.32 nm, respectively; for Z DNA, the values are 0.9 and 4.56 nm, respectively (19, 22).

We used fluorescence excitation profiles to experimentally examine the diameter dependence of the transition on the nanotubes. The emission from the B and Z forms (Fig. 2, A and B, with vertical lines comparing the peak centers) shows that the (6,5) nanotube (rt = 0.38 nm) undergoes a 15-meV decrease, whereas the (8,7) (rt = 0.51 nm) shifts only 5 meV. This inverse dependence for 11 of the strongest emitting SWNTs in the sample is shown in Fig. 2C. The curve in Fig. 2D is generated by using literature values for the geometric constants and by assuming 0.51 and 1.18 nm for the regressed widths of the bands for B and Z, respectively. Using constant dielectric values in Eq. 1 generates the horizontal line. Despite several limiting assumptions in this treatment (23), the model is able to predict the correct magnitude of the energy shift, the trend with radius, and the direction of the shift (red) using only geometric constants for the DNA adsorbed phase (tables S1 and S2) (10).

Fig. 2.

Fluorescence three-dimensional (3D) profile of excitation versus emission energy of a DNA-SWNT solution with (A) 0 μM HgCl2 and (B) 52.37 mM HgCl2. Vertical lines highlight the fluorescence red shift of individual SWNT species upon addition of Hg2+. (C) Peak centers of the nanotubes present in the 3D profile. (D) The energy shift of individual SWNT species modified by the Bethe-Salpeter equation to evince the effective dielectric constant differences caused by DNA geometry (orange points). The model curve (blue line) is based on the radial dependence of the DNA surface area coverage of the nanotube on changing from the B to Z form.

We have shown that the conformational rearrangement of a biomolecule can be transduced directly by the SWNT system. Given the recent discovery of a class of Z-DNA binding proteins and the association of Z-DNA to transcriptional activity and potential biological functions, it will be useful to have new probes to interrogate the conditions under which Z-DNA formation can occur (9, 24). Our previous work has shown that this type of DNA-SWNT complex is found to readily enter mammalian cells upon a 3-hour incubation and localize in the perinuclear region of the cell via endocytosis (25). We localized DNA-SWNTs within murine 3T3 fibroblasts (Fig. 3A) and perfused various concentrations of HgCl2 (Fig. 3, B and C) (10) in the extracellular buffer space for 5 min. The SWNT emission from the (6,5) nanotube, although shifted by 3 meV already upon uptake within the cell, red shifts additively with increasing Hg2+ concentration. After correcting for the initial shift caused by the new environment, the response of cell-bound DNA-SWNTs fits the model curve created by the same complexes in pure buffer (Fig. 3C). Control experiments produced no additional shift. The successful operation of the complex within living mammalian cells creates opportunities for new molecular probes that operate in the near IR and avoid natural autofluorescence of biological media.

Fig. 3.

(A) Area map of the (6,5) nanotube peak fluorescence intensity of DNA-SWNTs within murine 3T3 fibroblast cells overlayed on an optical micrograph of the same region. (B) Illustration of the experimental method used for ion-binding experiments conducted in mammalian cells. A cell containing endosome-bound DNA-SWNTs undergoes 785-nm excitation through a microscope objective. (C) The (6,5) nanotube fluorescence peak energy of DNA-SWNTs in 3T3 fibroblasts plotted versus Hg2+ concentration in the cell medium. The fluorescence energy of a population of 8 to 10 cells was averaged for each data point. Error bars indicate 1 SD. The red line shows the model curve from original Hg2+ binding experiment conducted in Tris buffer. The inset shows individual spectra at each concentration. (D) The (6,5) nanotube fluorescence energy of DNA-SWNTs in the following highly absorptive and scattering media: whole rooster blood (green triangles), black dye solution (black squares), and chicken tissue (blue circles) plotted on a model curve (red) from Hg2+ addition to SWNTs in buffer. The ΔE of all blood and tissue data points were corrected for an initial red shift due to the cellular environment.

Ion detection is also possible in media that already possess a strong ionic background. A 12,000–molecular weight cut off dialysis capillary was filled with DNA-SWNTs and inserted into whole blood and muscle tissue. The complex was added directly to a black dye solution (optical density > 4). The HgCl2 was still detected through these highly absorptive media (Fig. 3D). The near-IR fluorescence of DNA-SWNTs in the dye solution exhibited the same response as SWNTs in pure buffer. In whole blood and tissue, the presence of interfering absorbers of Hg2+ (free DNA, proteins, etc.) predictably shifts the observed sensitivity to larger values (C0 = 3500 μM in blood; 8000 μM in tissue); however, the DNA-SWNTs still provide a measure of the residual ions that are locally bound to the complex in these heterogeneous media (26).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article