A mirror-image experiment: Sorting carbon nanotubes by L-DNA

Abstract

DNA has found increasing applications in molecular engineering, yet its chiral property has rarely been utilized. Here, we report a mirror-image experiment using naturally occurring D-DNA and its enantiomer L-DNA to sort a chiral mixture of single-wall carbon nanotubes (SWCNTs). We find that parity conservation leads to a robust experimental outcome: changing DNA chirality results in handedness inversion of the purified nanotube. This finding provides a straightforward solution to the challenging problem of nanotube enantiomer sorting and a materials foundation for applications in fields such as spintronics and chiral sensing. To illustrate the latter, we show that enantiomeric pairs of DNA–SWCNTs can serve as bilateral chiral gauges for quantifying the degree of molecular chirality.

Introduction

Naturally occurring DNA is made of D-deoxyribose and is denoted as D-DNA. L-DNA, the mirror image of D-DNA, can be obtained through standard solid-phase synthesis using commercially available L-deoxyphosphoramidite monomers. L-DNA has been used to produce positive (left-handed) nodes in the topological application of DNA nanotechnology (1) and nuclease-resistant DNA nanocarriers for drug delivery with higher serum stability (2–4). Most relevant to this work is a recent computational study exploring the handedness of L-DNA wrapping on single-wall carbon nanotubes (SWCNTs) (5). Inspired by this study, here we demonstrate that L-DNA can be used to purify SWCNTs, yielding tubes with their handedness opposite to those obtained by D-DNA. We show that an enantiomeric pair of L-DNA- and D-DNA-wrapped SWCNTs can be employed in a bilateral chiral sensing scheme. We propose that an array of enantiomeric pairs of DNA–SWCNTs can be used to address a long-standing issue in stereochemistry: quantifying molecular chirality (6).

Results

DNA chirality change leads to SWCNT handedness inversion

The structure of a SWCNT is fully defined by a chiral vector composed of two integers (n, m), from which one can determine the real space stereoconfiguration of the tube and its reciprocal space electronic structure. As will be shown later, an outcome of this work is a proposal to extend the notion of the chiral vector as a means to describe the stereoconfiguration of any molecule.

Synthetic SWCNTs are racemic mixtures of left- and right-handed nanotubes. They can be sorted by single-stranded DNA to yield DNA-wrapped SWCNTs with defined nanotube handedness and helicity. In this approach, the chiral nature of DNA is essential for its SWCNT sorting function (7). We note that even though enantiomers of certain SWCNT helicities (e.g. (6, 5)) can be purified by the same DNA sequence, the resulting DNA–SWCNTs are diastereomers with limited utility in chiral sensing.

Conservation of parity symmetry in electromagnetic interactions dictates that if a D-DNA sequence allows purification of a SWCNT of certain handedness, then the L-DNA version of the same sequence should yield an exact mirror-image SWCNT with opposite handedness. We validated this prediction using three sequences: ss65 (TTA TAT TAT ATT), ss83 (TTT CCC TTT CCC CCC), and (GT)₂₀ (GTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGT). These sequences are chosen for their robust sorting performance demonstrated by our previous studies (7, 8). For each sequence, we used its D- and L-DNA version to obtain D- and L-DNA-wrapped SWCNT dispersions, respectively. Single-step extractions of these dispersions using a PEG/NaK phosphate aqueous two-phase system (8) yielded Dss65-(−)(6,5), Lss65-(+)(6,5) for the ss65 sequence; Dss83-(−)(8,3), Lss83-(+)(8,3) for the ss83 sequence; and D(GT)₂₀-(−)(8,4), L(GT)₂₀-(+)(8,4) for the (GT)₂₀ sequence. Figure 1 shows UV–Vis–near-IR absorption spectra and circular dichroism spectra for these enantiomeric pairs. Since many single-chirality SWCNTs can be purified by D-DNA sequences (7), our findings establish a general and straightforward method for purifying SWCNT enantiomer pairs using specific D- and L-DNA sequences. A relevant work reported in the literature makes use of chiral polyfluorene (9).

Fig. 1.

Spectroscopic characterization of three purified enantiomeric pairs of DNA–SWCNTs. UV–Vis–near-IR absorption spectra of A) Dss65-(−)(6,5), Lss65-(+)(6,5); B) Dss83-(−)(8,3), Lss83-(+)(8,3); and C) D(GT)₂₀-(−)(8,4), L(GT)₂₀-(+)(8,4). Circular dichroism (CD) spectra (normalized by the E₂₂ absorbance) of D) Dss65-(−)(6,5) with CD_norm = −101.0 mdeg and Lss65-(+)(6,5) with CD_norm = + 92.5 mdeg at E₂₂; E) Dss83-(−)(8,3) with CD_norm = −47 mdeg, Lss83-(+)(8,3) with CD_norm = + 48 mdeg at E₂₂; and F) D(GT)₂₀-(−)(8,4) with CD_norm = −85.8 mdeg and L(GT)₂₀-(+)(8,4) CD_norm = + 68.9 mdeg at E₂₂. There are slight differences between the spectra of SWCNTs purified by L- and D-DNA. This is most likely due to the different purity levels of the L- and D-DNA used in the experiment.

Bilateral chiral gauge

DNA–SWCNTs have been used in sensing and imaging via monitoring near-IR fluorescence of semiconducting tubes for various chemical compounds, including neurotransmitters such as dopamine (10), and cancer fingerprints (11). In particular, chiral molecule sensing by DNA–SWCNTs and its mechanism have been explored before (12). DNA coating structure controls how an analyte interacts with the hybrid and thereby its fluorescence. Inspired by the bilateral symmetry of biological perception organs such as eyes, ears, and nasal cavities (Fig. 2A), we propose a device called bilateral chiral gauge (BCG). The device contains an enantiomeric pair of DNA–SWCNTs forming two fluorescence-based sensing channels, labeled D or L corresponding to the chiral configuration of DNA used for SWCNT wrapping (Fig. 2B).

Fig. 2.

BCG concept and use for sensing 17β-estradiol and dopamine. A) Bilateral symmetry of human perception organs; B) BCG design. C) Molecular structure of estradiol; D) fluorescence of Dss83-(−)(8,3); and E) fluorescence of Lss83-(+)(8,3) at different levels of estradiol. F) Molecular structure of dopamine; G) fluorescence of Dss83-(−)(8,3); and H) fluorescence of Lss83-(+)(8,3) at different levels of dopamine hydrochloride. The estradiol assays were carried out with samples in 50% dimethyl sulfoxide/50% water solution, and the dopamine assays were carried out with samples in 20 mM sodium phosphate buffer, pH = 7. In these experiments, we used purified Dss83-(−)(8,3) or Lss83-(+)(8,3) solution (volume = 120 μL, SWCNT concentration ∼0.5 μg/mL) to manually carry out analyte sensing in a sequential manner. In each of the spectral plots, the fluorescence intensity of each sample was normalized by that of the nontreated one at the E₁₁ emission peak position.

As a proof of concept, we constructed a BCG using Dss83-(−)(8,3) and Lss83-(+)(8,3) pair and measured its response to a chiral molecule estradiol (Fig. 2C). Figure 2D and E shows that fluorescence enhancement in Dss83-(−)(8,3) and Lss83-(+)(8,3) differs significantly at several concentration levels of estradiol. As a control, we measured dopamine (Fig. 2F), an achiral molecule, showing no significant difference in fluorescence enhancement between Dss83-(−)(8,3) and Lss83-(+)(8,3) given experimental error (Fig. 2G and H). These data demonstrate a defining feature of our BCG: its two channels respond equally to an achiral molecule and differently to a chiral molecule. The response difference of the two channels provides a measure of the degree of chirality with a guaranteed null result for achiral molecules. The above experiment is carried out by probing one channel at a time manually, making it prone to excitation power fluctuation, detector drift, and titration errors. We envision that future use of the system will rely on (i) alternating photoexcitation and emission detection of the two channels with lock-in amplification scheme and (ii) having two channels sharing a common analyte bath. These changes are expected to improve the sensitivity and robustness of the response difference measured from the two channels.

Discussion

One enantiomeric pair of DNA–SWCNTs provides only limited information about an analyte's stereoconfiguration, insufficient to fully characterize the molecule's structure. This shortcoming can be mended by employing a BCG array composed of N distinct DNA–SWCNT enantiomer pairs, each with a unique surface chemistry defined by the DNA sequence and SWCNT chirality to probe certain aspect of a given analyte. For the ith pair of DNA–SWCNTs, we denote its response to the analyte (fluorescence intensity change and/or peak shift) by a pair of numbers: $d_i$ (for the D channel) and $l_i$ (for the L channel). We define a 2N-dimensional vector ($d_1,l_1$, $d_2,l_2$, … $d_i,l_i$, … $d_N,l_N$) and call it the chiral vector. When N pairs of DNA–SWCNTs are sufficiently different and N is large enough, say 20, the vector should capture enough stereoconfiguration information to uniquely identify the analyte molecule. We define a quantity $C\equiv \frac{1}{N}\sum _i|d_i-l_i|$. A nonchiral molecule should have $C=0$. A nonzero value of C provides a measure of the degree of chirality of the molecule. We note that C is a function of both the degree of the molecule's chirality and the strength of its coupling with the BCG array. However, since the two cannot be decoupled in any experimental chirality measure, we may simply define C as a measure of the degree of molecular chirality, provided the BCG array covers broad enough modes of molecular interactions. A BCG array may also be used as a molecular perceptron (13) to fingerprint biofluids for disease diagnosis (11). By highlighting the chiral aspect of biomolecules, a BCG array may be more efficient than its D-DNA–SWCNT-only counterpart for disease diagnosis. It is conceivable that some analytes may not bind to DNA–SWCNT hybrids due to limited chemical functionalities conferred by naturally occurring DNA. This limitation can be addressed by enlisting chemically modified DNA (XNA), which has been used successfully in the aptamer technology to greatly expand the repertoire of analyte binding (14).

In addition to optical-based BCG, we also envision an electronic version of BCG where an enantiomeric pair of DNA–SWCNTs form two conducting channels. Here lies the opportunity to make use of the chirality-induced spin selection effect for signal detection (15). More broadly speaking, the availability of enantiomeric pairs of DNA–SWCNTs may allow spintronic device designs mimicking in spirit complementary metal-oxide semiconductor electronic devices based on p-type and its charge-symmetry counterpart n-type semiconducting materials.

Materials and methods

L-DNA strand preparation

L-DNA molecules were synthesized on an Applied Biosystems 394 automatic DNA synthesizer using beta-L-DNA phosphoramidites and Universal Support III PS (Glen Research). They were removed from the support using 2 M ammonia in methanol at room temperature for 60 min and deprotected by adding 1 volume of 30% ammonium hydroxide at room temperature for 24 h.

DNA–SWCNT dispersion and purification

The dispersion followed previously published procedures with slight modification (7). We use a polyethylene glycol 6 kDa/sodium–potassium phosphate two-phase system for sorting, following a previously described process (8).

Funding

NIST internal funding, N000141912596 from the Office of Naval Research, DE-SC0007991 from the Department of Energy, and CCF-2106790 from the NSF.

Author Contributions

M.Z. performed DNA–SWCNT sorting and sensing. R.S. performed L-DNA synthesis.

Data Availability

All data supporting this study are either included in this report or cited in the references below.