Directional control of a processive molecular hopper

See allHide authors and affiliations

Science  31 Aug 2018:
Vol. 361, Issue 6405, pp. 908-912
DOI: 10.1126/science.aat3872

Tiny cargos ferried along a track

Control of molecules at the nanometer scale requires motors that convert potential energy into movement. Qing et al. designed a small molecule that could hop along a track of cysteine residues within a membrane-embedded protein pore. The direction of processive movement along the track was reversible, driven by an applied potential across the membrane. Cargos were attached to a carrier motor, and their position and chemical identity read out from changes in the current through the pore. These features enabled repeat observations of a single molecule as it moved back and forth on the track.

Science, this issue p. 908


Intrigued by the potential of nanoscale machines, scientists have long attempted to control molecular motion. We monitored the individual 0.7-nanometer steps of a single molecular hopper as it moved in an electric field along a track in a nanopore controlled by a chemical ratchet. The hopper demonstrated characteristics desired in a moving molecule: defined start and end points, processivity, no chemical fuel requirement, directional motion, and external control. The hopper was readily functionalized to carry cargos. For example, a DNA molecule could be ratcheted along the track in either direction, a prerequisite for nanopore sequencing.

Processivity lies at the heart of biological machines. A replicative DNA polymerase can incorporate thousands of nucleotides before dissociating from its template (1). Molecular motors, such as kinesin and dynein, travel directionally along microtubules over hundreds of steps without detaching from the track (24). For years, scientists have been trying to build moving molecules that resemble their biomolecular counterparts but use simpler components (5). The ultimate goals are to achieve true processivity, which can be defined as directional motion without leaving a track and the performance of useful work such as the transport of a cargo. Ideally, a synthetic system should exhibit the reversibility of stepping seen in various biological systems (6, 7) to enable the direction of motion to be switched through external control.

We report the design of a one-legged molecular hopper that is ratcheted by dynamic covalent chemistry along a protein track (Fig. 1A) with robust processivity (table S1). Further, the direction in which the hopper moves is subject to external control by an electrical potential. The track is built inside a protein nanopore, α-hemolysin (αHL), and consists of a series of cysteine footholds facing the lumen of the transmembrane β barrel (Fig. 1B). The cysteines are evenly spaced along a β strand with an average interfoothold distance of 6.8 Å (Cα-Cα) and an average vertical spacing of 5.6 Å. The hopper uses consecutive thiol-disulfide interchange reactions to move in the direction in which the DNA cargo has been oriented by an applied potential. To execute the SN2 reaction, the three participating sulfur atoms must align in a near-linear configuration (810). Under the applied potential, the DNA inside the barrel is pulled in the electric field with a force of ~10 pN (see supplementary materials). The force sets the overall direction of motion by flipping the DNA (see below) and helps to orient the disulfide for cleavage by the neighboring downstream cysteine thiolate, which moves the hopper one step forward, although other forces can contribute to the forward motion (Fig. 1C). Backstepping is disfavored and overstepping is impossible. Release of the hopper from the linear track was not observed, presumably because the track is too rigid to accommodate the resulting disulfide bridge between adjacent footholds on the same β strand. In short, each step is chemically directional because the hopper’s “foot” is positioned to favor the forward reaction. Further, the motion is autonomous, requiring no chemical fuel.

Fig. 1 A molecular hopper on a protein track.

(A) A hopper carrying a cargo (red flag) moves along a track by means of consecutive thiol-disulfide interchange reactions. The overall direction is set by the applied potential. (B) A six-foothold track comprising odd-numbered cysteine residues on a β strand inside the αHL protein nanopore. (C) The applied potential exerts a force on the DNA cargo, which helps to align the three sulfur atoms (yellow) participating in the interchange. The collinear geometry promotes hopping (1) (movie S1) but not the formation of an intrastrand disulfide, which would release the hopper from the track (2). Occasionally, backstepping is observed (3). Overstepping (4) does not occur. (D) The hopper enters the nanopore as a carrier-hopper disulfide conjugate.

Under +150 mV, the hopper was delivered to the track from the cis compartment as a hopper-carrier conjugate (Fig. 1D) capped with a single traptavidin, which was arrested at the pore entrance (Fig. 2A). The disulfide in the construct reacted strictly regioselectively with Cys115, releasing the carrier and placing the hopper-DNA cargo on the starting foothold (Fig. 2A and fig. S1). The location of the hopper was ascertained from the residual current passing through the nanopore, which reflected the length of the DNA located within the β barrel when the hopper was at a particular foothold (fig. S1). By monitoring current changes, we followed the stepwise hopping motion at the single-molecule level in real time.

Fig. 2 Monitoring individual hopper steps.

(A) Under +150 mV, a hopper-carrier conjugate capped with traptavidin was pulled from the cis compartment into an αHL nanopore containing cysteines at positions 113, 115, 117, 119, and 121 in one of the seven subunits. The resultant blockade reduced the ionic current from (i) to (ii). Reaction of the disulfide in the hopper-carrier with Cys115 covalently attached the hopper to the track, and the ionic current increased to (iii). (B) With a five-cysteine track, four hopping steps were observed at ±150 mV. Every forward step moved part of the DNA cargo outside the β barrel, producing an increase in conductance (movie S2). Alternation of the applied potential drove the hopper repeatedly up and down the track. (C) A hypothetical free energy diagram (not to scale) of the controlled hopping motion. (D) On an L-shaped track consisting of cysteines at positions 115, 117, 119, and 139, the hopper moved along the track from Cys115 to Cys119, where it was released by the side chain of Cys139. Subsequently, a second hopper became loaded at Cys115, but its motion was arrested at Cys117 because Cys119 was now engaged in an interstrand disulfide bond [(i), (ii), (iii) as in (A)]. Backstepping to Cys115 was also seen. Conditions: 2 M KCl, 20 mM HEPBS, 20 μM EDTA, pH 8.5, 20° ± 1°C.

The voltage-controlled hopping motion was directional and processive. On a track containing five cysteine footholds (at positions 113, 115, 117, 119, 121), a hopper carrying an oligoadenosine 40-mer (A40, hopper 1) moved cis to trans under +150 mV, and trans to cis under –150 mV (Fig. 2B and fig. S2). When the hopper reached a terminal foothold, the sign of the applied potential was reversed in order to reorient (flip) the DNA cargo, and hence the hopper. Alternation between positive and negative potentials repeatedly drove the hopper toward the trans or the cis end of the track. For the DNA to experience a force, at least one negatively charged phosphodiester bond must lie within the electric field, which drops along the length of the pore’s β barrel (11). Therefore, in the present nanopore construct, the length of the track was limited and voltage-controlled hopping could only be demonstrated with up to six footholds (fig. S3). The hopping direction could be changed by reversing the applied potential at any foothold; thus, our system offers complete control over directionality and the ability to move a hopper back to the initial foothold after an outing. Moreover, the applied potential provides an external energy source to produce directional motion (Fig. 2C; see supplementary materials). Limited by bilayer stability, the longest records of processive hopping were documented with hopper 1, which completed 249 forward steps in 93 min on a five-cysteine track (113 to 121) with a mean dwell time of ~22 s per foothold (fig. S2). Dissociation of the hopper from the track was never seen (n > 30 outings on different tracks), which implies that substantial improvements on step numbers would be achieved if the stability of the bilayer were improved. In comparison, previous synthetic small-molecule walkers moved directionally for less than 10 steps (5, 12). Wild-type kinesins typically exhibit a mean step number of 75 to 175 before dissociation (2, 3).

The hopping rates for each of the four steps on the five-cysteine track were derived for both the cis-to-trans and trans-to-cis directions at pH 8.5 and displayed differences of less than a factor of 40 (0.0081 to 0.30 s−1; tables S2 to S4). Because a thiolate is the reactive nucleophile in disulfide interchanges, the rate differences might arise from variations in the pKa values of the foothold thiols, which will be affected by neighboring residues. Previously, an arsenic(III) walker showed a factor of ≤50 difference in attachment rates with the footholds on the same five-cysteine track (113 to 121) at pH 8.0 (12). In the future, tracks of thiols might be engineered with optimized interfoothold distances and enhanced chemical reactivity to speed up the hopping process. Alternatively, the properties of the reactive sulfur atom in the hopper might be manipulated by flanking functional groups. With both the two-cysteine and three-cysteine tracks, the influence of voltage on hopping was examined at ±100 mV, ±150 mV, and ±180 mV (tables S5 and S6). The rates showed weak nonexponential voltage dependences, suggesting that the applied potential might not be the only source of propulsion (see supplementary materials). However, an electrical potential is essential (i) to flip the DNA over a large barrier to set the direction of motion, and (ii) to aid in the orientation of the three participating sulfur atoms to favor “forward” reactions over backsteps. With respect to the latter, the estimated “effective concentrations” of participating downstream thiols are not especially high (see supplementary materials) (13) and indeed need not be high to produce overall forward motion (see below).

Although disfavored, backstepping was occasionally detected, which we attributed to conformational lability of the hopper within the nanopore even under an applied potential. During a recording with hopper 1 on a five-cysteine track (113 to 121), there were 33 backward steps on the nonterminating footholds out of 282 steps in total (12%). Of the 33 backward steps, 29 occurred from 115 to 117 at –150 mV (table S3). Despite the large forward equilibrium constant for 117-to-115 stepping (K = k117–115/k115–117 = 22), 115-to-117 backstepping was observed because of the comparatively slow forward movement to the next foothold, 113 (at –150 mV, k115–117 = 0.0094 s−1, k115–113 = 0.0081 s−1; tables S2 and S3). Backstepping was observed when the hopper was left on a terminal foothold, as no forward footholds remained (Fig. 2B). These backsteps were quickly reversed by the hopper, which preferentially resided at the final station (K = k119–121/k121–119 = 5.2; table S4). The overall motion of the hopper is governed by the product of the K values for each step. A modest value (K > 1) at each step produces a considerable overall tendency toward forward movement.

Each subunit of the αHL pore offers two antiparallel β strands to the transmembrane β barrel with an interstrand distance of ~5 Å (Cα-Cα). Given that the formation of cross-strand disulfides has been reported (14), we reasoned that the addition of a cysteine on an adjacent strand would compel hopper release from the track at a designated foothold. Indeed, with an L-shaped track consisting of cysteines at positions 115, 117, 119, and 139, the hopper attached to the track at foothold 115 by regioselective disulfide formation and dissociated from the track when it reached foothold 119. The release was initiated by Cys139 through thiol-disulfide interchange to form a cross-strand disulfide bridge, which blocked the access of subsequent hoppers to foothold 119 (Fig. 2D). The preference for hopper release versus hopper transfer to the adjacent strand is attributed to an unfavorable collinear alignment of the three participating sulfur atoms necessary for transfer. In the future, the engineering of footholds on a surface will allow the construction of more complex hopping pathways where hoppers are transferred to new tracks at designed junctions and cargos are released at predesignated depots.

The ability to translocate a stretched DNA cargo while maintaining a covalent bond with the nanopore suggests a method for the chemical ratcheting of a nucleic acid during nanopore sequencing (15), which was explored in a proof-of-concept experiment. To provide a marker, we incorporated two adjacent abasic residues (1′,2′-dideoxyribose, dS) (16) into the cargo oligos carried by hoppers 2 and 3 and recorded current patterns during four-step hopping between Cys113 and Cys121 (Fig. 3A). By comparison with hopper 1, hoppers 2 and 3 showed different patterns of current modulation (Fig. 3, B and C, fig. S4, and table S7). The conductance patterns generated by the four-step hopping motion could be repeated with different molecules of hoppers 2 and 3 (n = 3 for each hopper), establishing the patterns as clear identifiers of each cargo sequence. The residual currents (Ires%, the remaining current as a percentage of the open pore current) for the three hoppers residing at each foothold were plotted for comparison (Fig. 3D). Hoppers 1 and 2 gave almost identical current blockades at each of footholds 115 and 113 under –150 mV, implying that the dSdS sequence had been transported well out of the sensing region by hopper 2. Moreover, hoppers 2 and 3 have a single nucleotide offset in the dSdS positions, and we observed a one-step offset between hoppers 2 and 3 in ΔIres%, the difference in Ires% between two successive steps [Fig. 3D; the vertical step size, 5.6 Å, is similar to the internucleotide distance in stretched single-stranded DNA, 6.9 Å (17)].

Fig. 3 Discrimination of different DNA cargos.

(A) Two sequential abasic nucleotides (dSdS) were substituted at positions 3 and 4 (hopper 2) or at positions 2 and 3 (hopper 3). The numbers of nucleotides (brown circles, dA; red circles, dS) placed inside the β barrel are based on PyMOL modeling. (B) With a five-cysteine track, four-step hopping was observed with hopper 2 at ±150 mV. The current decreases for hops from 115 to 117 and from 121 to 119 are marked (green arrows). (C) Four-step hopping with hopper 3 at ±150 mV. The current transitions for hops from 117 to 119 and from 119 to 117 are marked (red arrows). (D) Top: Overlaid current traces of hoppers 1, 2, and 3 [colors as in (A)] with step durations normalized. The current levels are given as the residual current with respect to the open pore level (Ires%). Bottom: Step sizes of hoppers 2 and 3 plotted as ΔIres%. Minima in the plots showing the single nucleotide offset are marked. Conditions: 2 M KCl, 20 mM HEPBS, 20 μM EDTA, pH 8.5, 20° ± 1°C.

These observations demonstrate that the hopper system reported here has the potential to discriminate bases for sequencing purposes (16). An advantage of a processive hopper, which might improve sequencing accuracy, is the ability to reverse the chemical ratcheting process and thereby obtain many-fold coverage of an individual DNA strand. Of course, the present system is limited by its short track. Although longer β-barrel pores exist (18, 19), a viable sequencing process will require protracted ratcheting of numerous DNA strands in parallel, perhaps by using footholds on an extended crystalline surface or internal thiophosphate feet to transport long replica strands over relatively short tracks.

Supplementary Materials

Materials and Methods

Figs. S1 to S12

Tables S1 to S7

Movies S1 and S2

References (2051)

References and Notes

Acknowledgments: We thank H. Tamagaki and J. Lee for their help with peptide design and synthesis; S. Lee for her help with data analysis using QuB; L. Taemaitree for her help with LC-MS characterization; and M. Howarth for providing a sample of traptavidin for preliminary tests. Funding: Supported by a European Research Council Advanced Grant; a China Scholarship Council–University of Oxford Scholarship (Y.Q.); and an Amelia Jackson Studentship, Exeter College, Oxford (S.A.I.). Author contributions: Y.Q. and H.B. designed the study; Y.Q. performed the experiments; S.A.I. repeated the single-channel recordings; G.S.P. contributed to the protein preparation; and Y.Q., S.A.I., and H.B. wrote the manuscript. Competing interests: H.B. is the founder of, a consultant for, and a shareholder of Oxford Nanopore Technologies, a company engaged in the development of nanopore sensing and sequencing technologies. Y.Q. and H.B. have filed patents describing the molecular hopper and applications thereof. Data and materials availability: All data are available in the main text or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article