{"paper_id":"4a9ddf66-6b25-435b-b3ff-aa72e8eb3ab2","body_text":"Correlating fluorescence microscopy, optical and magnetic tweezers to study single chiral \nbiopolymers such as DNA \n \nJack W Shepherd1,2*, Sebastien Guilbaud1*, Zhaokun Zhou3*, Jamieson Howard1*, Matthew \nBurman1, Charley Schaefer1, Adam Kerrigan4, Clare Steele-King5, Agnes Noy1, and Mark C Leake1,2, ✝ \n1 School of Physics, Engineering and Technology, University of York, York, YO10 5DD \n2 Department of Biology, University of York, York, YO10 5DD \n3 Guangdong Provincial Key Lab of Robotics and Intelligent System, Shenzhen Institute of Advanced \nTechnology, Chinese Academy of Sciences, China \n4 The York-JEOL Nanocentre, University of York, York, YO10 5BR \n5 Bioscience Technology Facility, University of York, York, YO10 5DD \n✝ For correspondence. Email mark.leake@york.ac.uk \n* Authors contributed equally \n \nAbstract \nBiopolymer topology is critical for determining interactions inside cell environments, exemplified by \nDNA where its response to mechanical perturbation is as important as biochemical properties to its \ncellular roles. The dynamic structures of chiral biopolymers exhibit complex dependence with \nextension and torsion, however the physical mechanisms underpinning the emergence of structural \nmotifs upon physiological twisting and stretching are poorly understood due to technological \nlimitations in correlating force, torque and spatial localization information. We present COMBI-\nTweez (Combined Optical and Magnetic BIomolecule TWEEZers), a transformative tool that \novercomes these challenges by integrating optical trapping, time-resolved electromagnetic \ntweezers, and fluorescence microscopy, demonstrated on single DNA molecules, that can \ncontrollably form and visualise higher order structural motifs including plectonemes. This technology \ncombined with cutting-edge MD simulations provides quantitative insight into complex dynamic \nstructures relevant to DNA cellular processes and can be adapted to study a range of filamentous \nbiopolymers. \n \nKeywords \nOptical tweezers, Magnetic tweezers, Fluorescence microscopy, Plectonemes, DNA topology, Single -\nmolecule \n \nMultiple cell processes involve chiral biopolymers experiencing pN scale forces1 and torques of tens \nof pN.nm2, exemplified by molecular machines on DNA3 where torsion is a critical physical factor4. \nAlthough lacking torsional information, optical tweezers (OT)5 combined with fluorescence \nmicroscopy of dye-labelled DNA was used to image DNA extension in response to fluid drag6, and \nthough missing topological details from fluorescence visualisation, magnetic tweezers (MT) \nexperiments have demonstrated how DNA molecules respond to force combined with supercoiling7.  \nDevelopments in OT and MT have enabled molecular study of DNA over-stretching8, protein \nbinding9, dependence of mechanical properties to ionicity10, and DNA-protein bridge formation11. \nFluorescence microscopy combined with OT have enabled super-resolved measurement of extended \nDNA in the absence of torsional constraints12,13, with angular OT enabling torque control14. High-\nprecision MT has facilitated single-molecule DNA study of over- and undertwisting, including twist-\nstretch coupling15, dependence of torsion on temperature16 and salt17, and binding of filament-\nforming proteins such as RecA18. MT has enabled single-molecule torsion dependence \nmeasurements of non-canonical DNA structures including P-DNA19, left-handed DNA20 and higher-\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\norder motifs of the G-quadruplex21 and plectonemes22, applying vertical geometries to extend \nmolecules orthogonal to the focal plane. DNA plectonemes have been visualised using fluorescence \nmicroscopy combined with MT, by reorienting molecules almost parallel to the focal plane, limited \nto observations following but not during plectoneme formation20. Plectoneme formation has also \nbeen induced through intercalator binding to DNA combed onto surfaces then imaged with \nfluorescence microscopy to track plectonemes23.  \nOT combined with fluorescence have been used to study DNA over-stretching in torsionally \nunconstrained DNA24, revealing S-DNA structures by measuring polarized emissions from bound \nfluorophores25. Later developments using torsional constraints were implemented by stochastic \nstretch-unbind-rebinding applied to single optically trapped DNA molecules, with a caveat that \nsupercoiling cannot be controlled in advance, or reversed26.  \nSimulations have enabled insight into DNA structural transitions, coarse-grained approaches using \noxDNA27 on mechanically28–30 and thermally perturbed DNA15, and all-atom methods which predict \nsequence-dependence to torsionally-constrained stretching31 and explore the denaturing pathways \nof torsionally unconstrained stretching32. Simulations have also been used with single-molecule MT \nvia space-filling algorithms to predict P-DNA formation19 and molecular dynamics (MD) simulations \nto study stretch/twist coupling15. More recently, DNA minicircles comprising just a few hundred base \npairs (bp) have been used as computationally tractable systems to investigate supercoiling33 and \nusing AFM imaging34.  \nSingle-molecule experiments used to probe DNA mechanical dependence on structural \nconformations have advantages and limitations. OT enables high forces beyond the ~65 pN \noverstretching threshold up to ~1 nN 35. However, they cannot easily control torque of extended \nmolecules without non-trivial engineering of trapping beams and/or the trapped particle’s shape36, \nand establishing stable torque comparable to MT using readily available microbeads is not feasible. \nWhile MT can enable reversible supercoiling of single molecules at ~pN forces719, mechanical \nvibrations over ~seconds introduced during rotation of nearby permanent magnets places \nlimitations on structural transitions probed.  \nWe present Combined Optical and Magnetic Biomolecule TWEEZers (COMBI-Tweez), which \novercomes these challenges. COMBI-Tweez is a bespoke, correlative single-molecule force and \ntorsion transduction technology colocating low stiffness near infrared (NIR) OT with laser excitation \nfluorescence microscopy on DNA molecules in a ~mT B field, generated by pairs of Helmholtz coils, \nrotated in orthogonal planes independently under high-precision control (Supplementary Movie 1). \nOT and MT can be operated independently, while a trapped fluorescently-labelled DNA molecule, \nwhich we use as well-characterized test chiral biopolymer, can be extended parallel to the focal \nplane to enable fluorescence imaging in real time with sub-nm displacement detection via laser \ninterferometry. To our knowledge, this is the first report of stable optical trapping of magnetic beads \nat physiologically relevant temperatures. \nWe demonstrate this technology through time-resolved formation and relaxation of higher-order \nstructural motifs including plectonemes in DNA; these emergent features comprise several open \nbiological questions which relate to plectoneme size, position and mobility. We measure changes to \nthe buckling transition37 due to binding of the fluorescent dye SYBR Gold38, and controlled \nquantification of interactions between two “braided” DNA molecules. We discuss these findings in \nthe context of modelling structural motifs using MD simulations. The unique capability of COMBI-\nTweez is the correlative application of several single-molecule techniques on the same biopolymer \nmolecule, not attainable with a subset of techniques alone. OT enables high-bandwidth force \nmeasurement, while MT generates magnetic microbead rotation to induce controlled torque with \nno mechanical noise or lateral force. Control software enables precise tuning of supercoiling density \nσ, with bead rotation verified using fluorescence imaging of conjugated reporter nanobeads. We \ndemonstrate COMBI-Tweez using high-sensitivity fluorescence detection whose utility can be easily \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nextended to multiple fluorescence excitation and surface functionalization methods, with future \napplications to study several generalised biopolymers. \nResults \nDecoupling force and torque \nCOMBI-Tweez OT control of force and displacement, with MT control of torque (Fig. 1a,b, \nSupplementary Movie 1) is simple and robust. No modifications to NIR laser trapping beams are \nrequired, such as Laguerre-Gaussian profiles to impart angular momentum39, nor nanoscale \nengineering such as cylindrical birefringent particles40. Since COMBI-Tweez OT controls bead \ndisplacement, dynamic B field gradients that update continuously to reposition the bead, common in \nMT-only systems41, are not required. We use a uniform B field via two pairs of coils carrying \nsinusoidal currents to rotate an optically trapped magnetic bead (Fig. 1c, methods, Supplementary \nNote 1 and Supplementary Figs. 1-8). Decoupling force and torque allows COMBI-Tweez to exert \narbitrary combinations of sub-pN to approximately 10 of pN and of sub-pN·nm to tens of pN·nm, \nrelevant for cellular processes involving DNA such as plectoneme formation42. COMBI-Tweez also \nexploits rapid 50 kHz quantification of bead positions using back focal plane (bfp) detection of the \nNIR beam via a high-bandwidth quadrant photodiode (QPD) not limited by camera shot noise43 \n(methods). In MT-only setups, camera-based detection is often used to track beads with ~kHz \nsampling; our high-speed detection permits faster sampling to probe rapid structural dynamics. \n \nFigure 1 Principle of COMBI-Tweez. (a) An NIR OT is colocated with MT generated from a pair of orthogonal Helmholtz \ncoils enabling a single tethered DNA molecule to be visualised simultaneously in the focal plane using a variety of light \nmicroscopy modes, to observe DNA structural dynamics when the molecule is mechanically perturbed. (b) Cartoon of a \nplectoneme formed in COMBI-Tweez, using chemical conjugation to tether one end of a DNA molecule to a surface-\nimmobilised “anchor” bead with the other tethered to an optically trapped magnetic microbead. SYBR Gold fluorophores38 \nlabel the DNA via intercalation between adjacent base pairs. (c) Simulation of B field when vertical and horizontal coil pairs \nare separately activated, indicating a highly uniform field in the region of a trapped bead. Generalised magnetic field \nvectors can be generated by combining outputs from each coil; it is simple to generate stable B field rotation in a plane \nperpendicular to the microscope focal plane resulting in rotation of the trapped bead that enables torque to be \ncontrollably applied to a tethered DNA molecule. \n \nStretching/twisting label-free DNA \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nWe first assessed COMBI-Tweez to perform torsionally unconstrained stretch-release (MT module \noff) using a 15.6 kbp test double-stranded DNA (dsDNA) construct in the absence of fluorescent tags \n(methods, Supplementary Fig. 9), generated as a fragment from λ DNA digestion and functionalised \nwith concatemer repeats of biotin or digoxigenin “handles” on either end, since the results can be \ndirectly compared with several previous OT/MT molecular studies on similar constructs. Our handles \nwere 500 bp leaving a 14.6 kbp central region. We surface-immobilised 5 µm “anchor” beads with \nnon-specific electrostatic adsorption by flowing diluted anti-digoxigenin (DIG) beads into a flow cell \nbefore BSA passivation then introducing DNA in ligase buffer. Following incubation and washing we \nintroduced 3 µm Neutravidin-functionalised magnetic beads, sealed the flow cell and imaged \nimmediately. To form tethers, we hold an optically-trapped magnetic bead 500 nm from an anchor \nbead to allow a tether to form through in situ incubation, which we confirmed in preliminary \nexperiments across a range of buffering conditions and ionic strengths from approximately \n10-200 mM, prior to focusing just on physiological conditions using phosphate buffer saline \n(methods). Each tether’s mechanical properties could then be studied via stretch-release through \nnanostage displacement at a ~1 Hz frequency and ~5 µm amplitude (methods and Supplementary \nMovie 2).  \nSince it is well established that strong B fields induce above physiological temperature increases in \nmagnetic nanoparticles and also that previous attempts using optically trapped magnetic beads \nwere limited by temperature increases due NIR laser absorption by magnetite44 (and personal \ncommunication Nynke Dekker, University of Oxford), we characterized trapped bead temperature \nextensively. Performing transmission electron microscopy (TEM) on dehydrated resin-embedded \n70 nm thick sections of the magnetic beads (methods) indicated electron opaque nanoparticles \ndistributed in a ~100 nm thin surface just inside the outer functionalisation layer (Fig. 2a,b) while \nelectron diffraction was consistent with an iron (II/III) oxide magnetite mixture as expected (Fig. 2c). \nWe estimated temperature increase near trapped beads by measuring the distance dependence on \nmelting alkane waxes (methods). By positioning a trapped bead adjacent to a surface-immobilised \nwax particle, we observed the melting interface using brightfield microscopy (Fig. 2d-f), enabling \ndistance measurement to the bead surface (Fig. 2g). Modelling heat generation analytically \nsuggested ~1/x dependence for small x (distance from the bead surface) less than the bead radius \n(Supplementary Note 2). We used the data corresponding to small x to show that increasing the \ndistance of the magnetite from the bead centre significantly reduces the bead surface temperature \n(Supplementary Fig. 10), resulting in 45˚C in our case (Supplementary Note 3). Finite element \nmodelling of heat transfer outside the bead confirmed deviation of 1/x for larger x (Fig. 2h), fitting all \nwax data points within experimental error, indicating a range of 45-30˚C for distances from the bead \nsurface from zero through to the ~5 µm DNA contour length. The mean ~37-38˚C over this range is \nserendipitously perfectly aligned for physiological studies. This acceptable temperature increase \nlimits our trapping force to approximately 10 pN, well within the physiological range we want to \nexplore for DNA. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n  \nFigure 2 Absorption of NIR laser by magnetite results in physiological temperatures surrounding trapped beads. a) TEM \nof 70 nm thick section taken from one resin-embedded dehydrated magnetic bead (JEOL 2100+ TEM 200 kV). b) Zoom-in \non the bead surface showing red circled electron-dense nanoparticles produce a c) selected area diffraction pattern \nconsistent with iron (II,III) oxide. Over 50 beads were viewed across three sectioned samples with identical form by eye. \nSelect area diffraction was done in several locations, each showing an iron (II,III) oxide mix. d)-f) One representative of a \nbrightfield of a trapped magnetic bead positioned adjacent to solid ~10 µm surface immobilised alkane wax particle, shown \nfor waxes C19, C20 and C22, indicating formation of melting interface at 30 s timepoint. C22 often either did not visibly \nmelt or had interface less than a radius from bead edge g) Boxplot indicating mean (*) distance from bead edge to melting \ninterface for the three waxes (blue box interquartile range, the whiskers indicate the data range, with the numbers of wax \nparticles analysed nC19 = 6, nC20 = 7, nC22 = 12). h) Finite element model (blue, Supplementary Note 3) for heat transfer \npredicts that the temperature stays constant inside the bead at ~45˚C and decreases with distance away from the bead \n(experimental datapoints overlaid in red, mean and s.d. error bars shown, from same datasets as panel g). The \ntemperature dependence beyond bead surface becomes shallower than the ~1/x predicted from analytical modelling \n(Supplementary Note 2) due to heat transfer into the glass coverslip from the surrounding water. \n \nThe estimated DNA persistence length determined from wormlike chain (WLC)45 fits to force-\nextension data (methods) was comparable to previous studies46 with mean of approximately 50 nm \n(Fig. 3a) with no significant difference between stretch and release half cycles indicating minimal \nhysteresis due to stress-relaxation47 (Supplementary Fig. 13a).  Mean contour length was \napproximately 5 µm (Fig. 2a), comparable to   sequence expectations. We configured OT stiffness to \nbe low (10 pN/µm) compared to earlier DNA studies that probed overstretch, allowing instead stable \ntrapping up to a few pN relevant to physiological processes while avoiding detrimental heating.  \nWe tested COMBI-Tweez to twist torsionally constrained DNA, first keeping the OT stationary as a \ntrapped bead was continuously rotated at 1 Hz. Upon prolonged undertwist at a force of 1-2 pN \ntethers remains broadly constant consistent with melting of the two DNA strands, whereas \nperforming the same experiment using prolonged overtwist we found that the force increased as \nDNA is wound, until high enough to pull the bead from the trap, towards the anchor bead at which \npoint meaningful measurement ends (Supplementary Movie 3). As reported by others, we observed \nthat DNA force does not increase linearly with torsional stress upon overtwist but undergoes non-\nlinear buckling37 at which DNA no longer absorbs torsional stress without forming secondary \nstructures which shorten its end-to-end length (Fig. 3b). Utilising the 50 kHz QPD sampling, we \nmonitored rapid fluctuations in conformations between individual rotation cycles which would not \nbe detectable by the slower video rate sampling (Fig. 3b inset), with a ~2-fold increase in the \nfluctuation amplitude during buckling which may indicate rapid transitioning of ~100 nm metastable \nregions of DNA. For the example shown, there is an initial 1.7 pN force at 4.5 µm extension (bead \nimage inset, 36 s). Force increased non-linearly during rotation with corresponding decrease in \nextension to 4.3 µm (inset, 72 s) prior to buckling at 2.2 pN and exiting the trap (75 s). Performing \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\npower spectral analysis indicated an 85% increase in low frequency (<60 Hz) components during \nbuckling, qualitatively consistent with the emergence of cumulatively supercoiled DNA with higher \noverall frictional drag. Modelling twist for an isotropic rod suggests torque τ per complete DNA \ntwist is approximately C/L where C is the torsional modulus and L the DNA length. Using C = 410 \npN.nm obtained previously on DNA using MT single-molecule twist experiments2 indicated torque of \n0.08 pN.nm per bead rotation.  \nWe next sought to reproduce “hat curves” which map relations between DNA extension versus twist \nwhen positively and negatively supercoiled (methods). At forces less than a critical value Fc of 0.6-0.7 \npN 48, DNA supercoiling was approximately symmetric around σ = 0 49. At higher forces, negative \nsupercoiling no longer leads to plectonemes but instead to melting bubbles48. We selected forces \neither side of Fc, fixed in real time using bead position feedback to the nanostage. As anticipated, \nhigh-force hat curves (Fig. 3c left panel, 1.1 pN) indicate asymmetry with a high extension plateau \nfrom σ = 0 to σ < 0 (blue trace) and σ > 0 to σ = 0 (green trace), with decreasing extension as DNA is \novertwisted (red). At lower force (Fig. 3c right panel and Fig. 3d, 0.5 pN) negative supercoiling, \nindicated by green and red traces, comprised 200 rotations for each trace equivalent to maximum \nnegative torque of -1.6 pN.nM, was broadly symmetrical with the positive supercoiling pathway, \nindicated by blue and yellow traces, which comprised 200 rotations per trace equivalent to \nmaximum positive torque of +1.6 pN.nM. Lateral (15 nm/s) and axial (7 nm/s) drift was effectively \nnegligible for individual rotation cycles by allowing COMBI-Tweez to reach a stable temperature \nfollowing coil activation, however, over longer duration hat curve experiments comprising several \nhundred seconds we performed drift correction on bead positions (methods).  \n     \n \nFigure 3 COMBI-Tweez in brightfield quantifies key supercoiling features. a) Left: (Blue) representative DNA force vs \nextension showing no obvious stress-relaxation47, WLC fit (red) Lc = 5.3 µm and persistence length P = 52 nm obtained as \nmean from n = 44 consecutive stretch-release half cycles, s.d. 13%, error on individual fits <0.002% estimated from 1,000 \nbootstrapped fits from 1% of data per bootstrap. Right: violin plot of WLC fitted mean Lc and P for n = 27 separate tethers, \npresented as mean value +/- s.e.m. b) Extension measured from QPD of trapped bead (blue) plotted as function of time \nduring continuous overtwist for fixed trap, initially 1.7 pN. Inset: brightfield of bead pair at σ values of 0.02 and 0.04 taken \nat 36 s/72 s before magnetic bead lost from trap at 75 s. Zoom-in (blue) plus indicative running median overlaid (red, \n5,000-point window), showing same trend in decreasing extension as video analysis but revealing additional \nconformational heterogeneity between individual rotation cycles. Qualitatively similar non-linear buckling responses \nobserved from n = 5 overtwisted tethers pulled out of optical trap at a comparable force level to example shown here. c) \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nLeft: Hat curve for force clamp experiment, 1.1 pN (green “outward” trace is almost entirely overlaid onto blue “inward” \ntrace indicating negligible hysteresis); Right: hat curve, 0.5 pN. Both curves drift-corrected between different sections \ntaken at timepoints which span up to 25 min. Apparent hysteresis between blue/yellow sections at low force due to 1-2 s \nresponse time of force clamp, while that between green/red sections is due both to this effect and bead pair transiently \nadhering for a few seconds following formation of tether extension reduction upon negative supercoiling. Arrows indicate \npath taken during the experiment (blue outward and yellow inward overtwist, followed by green outward and red inward \nundertwist). d) Brightfield from low force clamp experiment of c) at indicated timepoints, with nanostage moving anchor \nbead (green) relative to optically trapped bead (magenta) to maintain constant force (see also Supplementary Movie 4).  \n \nDNA dynamics during stretch/twist visualized by correlative fluorescence imaging \n     \n \nFigure 4 Correlative fluorescence imaging and torsional manipulation of DNA enables visualisation of tether dynamics. a) \nOne representative example of an optically trapped bead (magenta) next to surface-immobilised anchor bead (green). \nMultiple DNA molecules clearly visible on anchor bead as distinct foci since entropic compaction of untethered DNA results \nin ~500 nm end-to end length (also see Supplementary Fig. 11). b) One representative example of a tether which is formed \nbetween the two beads in a), indicated by white arrow. c) One representative example of a tether showing image frames \nextracted from continuous stretch-release experiment performed in fluorescence. d) Representative force-extension data \ncompiled from n = 8 consecutive stretch-release half cycles obtained from same tether by WLC fits (red). Mean persistence \nlength determined from n = 9 separate tethers was 53 ± 15 nm (± s.e.m.), comparable with that obtained for unlabelled \nDNA construct while e) mean contour length larger by >30% consistent with reports of SYBR Gold interactions38 presented \nas the mean value +/- s.e.m. from n = 9 independent tethers. f) One representative example of two DNA molecules formed \nbetween anchor and trapped bead in which the tethers are coaxial. After one tether breaks (right), it retracts along the \nsecond leaving a bright still-braided region and a dimmer single-tether region. g) One representative example of two DNA \nmolecules which have formed a Y shape with “braided” section and two single tethers meeting anchor bead at different \npoints. After rotation the braided section grows (see also Supplementary Fig. 14). h) One representative example of a \ntether snapping and retracting to anchor bead, indicated by white arrow. All experiments were independently repeated at \nleast 9 times. \nWe tested the fluorescence capability using several illumination modes including widefield \nepifluorescence, Slimfield50 (a narrowfield microscopy enabling millisecond imaging of single \nfluorescent molecules), and an oblique-angle variant of Slimfield that combines narrowfield with \nhighly inclined and laminated optical sheet microscopy (HILO)51, using 488 nm wavelength laser \nexcitation (methods). To visualize tether formation, we pre-incubated DNA-coated anchor beads \nwith intercalating dye SYBR Gold38 at one molecule per 1-2 µm2 (Fig. 4a, methods and \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nSupplementary Movie 5), enabling inter-bead tethers to be visualized, using oxygen scavengers to \nmitigate photodamage (Fig. 4b). Low power widefield epifluorescence allowed single tethers to be \nobserved at video rate for ~100 s before tether breakage due to photodamage (Fig. 4c). Repeating \nthe same torsional unconstrained stretch-release protocol as for unlabelled DNA (Fig. 4d and \nSupplementary Movie 6) indicated a mean persistence length comparable to unlabelled of \napproximately 50 nm, whereas the contour length increased over 30% (Fig. 4e), consistent with \nearlier reports at comparable stoichiometry values of SYBR Gold:DNA38 if at least half available SYBR \nGold intercalation sites are occupied.  \nExperimenting with DNA concentration also demonstrated potential for studying interactions \nbetween multiple molecules, e.g. “braided” double tethers that can be controllably unwound by \nbead rotation to reform separated single tethers (Fig. 4f, Supplementary Fig. 14, methods and \nSupplementary Movie 7), including studying two fully braided coaxial DNA molecules (Fig. 4g and \nSupplementary Movie 8) in some instances resulting in visible entropic retraction when one of these \nsnaps (Fig. 4h and Supplementary Movie 9). These unique features of COMBI-Tweez allowed high-\nprecision control of DNA-DNA interactions as a function of supercoiling.  \nTime-resolved visualization of DNA structural motifs \nTo explore the capability to study complex structural dynamics, we investigated DNA plectonemes, \nhigher-order structural motifs that emerge in response to torque46. Applying constant force to a \nSYBR Gold labelled tether using the same protocol as for unlabelled DNA of 1-2 pN, above the critical \nFc of 0.6-0.7 pN 48, and rotation to control σ , resulted in a similar asymmetrical hat curve (methods) \n– in the example shown in Fig. 5a starting from σ = 0 and overtwisting (blue trace) to σ  > 0 with \nassociated drop in fractional extension, then undertwisting through to σ  = 0 (yellow trace) then to \nσ  < 0 with characteristic plateau in fractional extension (green/red traces), high-frequency QPD \nsampling enabling detection of rapid structural fluctuations over ~100 nm in this region (Fig. 5a \ninset). We measured a small difference to σ at which extension begins to decrease during \ncontinuous overtwisting; for SYBR Gold labelled DNA this occurred at σ close to zero, whereas for \nunlabelled DNA this was between 0.05-0.1, likely caused by DNA mechanical changes due to SYBR \nGold intercalation38. \nTo measure longer timescale structural dynamics, we redesigned data acquisition using stroboscopic \nimaging to expose just 10 consecutive images in fluorescence between 50 continuous bead rotations \nto minimise photodamage, allowing 1 min equilibration between 50 rotation segments – the \nexample of Fig. 5b obtained below Fc using this discontinuous imaging protocol showed the same \nqualitative features of an approximately symmetrical hat curve at low force for unlabelled DNA, with \nthe caveat of negligible hysteresis following equilibration between rotation segments.  \nUsing the stroboscopic illumination protocol, high forces for relaxed or negatively supercoiled DNA \ncaused no differences in terms of tether shortening from the fluorescence images and no evidence \nof the formation of higher-order structures (Fig. 5c left and centre panels), though negatively \nsupercoiled DNA exhibited image blur in localized tether regions, absent from relaxed DNA, manifest \nin a higher average full width at half maximum (FWHM) intensity line profile taken perpendicular to \nthe tether axis (Fig. 5d and methods). Conversely, overtwist resulted in visible tether shortening \nalong with the appearance of a higher intensity fluorescent puncta along the tether itself (Fig. 5c \nright panel and inset, indicated as a plectoneme, see also Supplementary Movie 10. Significant \novertwisting leads eventually to sufficient DNA shortening which pulls the magnetic bead out of the \noptical trap, shown in a different tether in Supplementary Movie 11).  \nAt low force, fluorescence images of positively supercoiled DNA indicated similar tether shortening \nand formation of a fluorescent punctum (Fig. 5e and inset, σ = 0.11 and 0.14). Conversely, although \nrelaxed DNA showed no evidence of higher-order localized structures, negatively supercoiled DNA \nindicated both tether shortening and the appearance of a fluorescent punctum (Fig. 5e and inset, \nσ = -0.14), albeit dimmer than puncta observed for positively supercoiled DNA. Notably, we find that \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\ntethers formed by undertwisting appeared at different locations from those created through positive \nsupercoiling. \nSimilar experiments on full length λ DNA (48.5 kbp) resulted in more than one plectoneme \n(methods), the example of Fig. 5f following positive bead rotations (Supplementary Movie 12) having \nthree puncta per tether. Using Gaussian fitting to track puncta to 20 nm precision and summing \nbackground-corrected pixel intensities enabled estimation of bp content, indicating 2.6-3.1 kbp per \nplectoneme whose rate of diffusion decreased with increasing bp content (Fig. 5g). In this example, \nthere was a step-like increase in DNA extension of ~400 nm at approximately 860 ms, consistent \nwith a single-strand nick. Following this, each plectoneme then disappeared in sequence between \napproximately 900-1,000 ms (Fig. 5h) indicative of a torsional relaxation wave diffusing from the \nnick52. \nEstimating the proportion of bp associated with puncta in the shorter 15 kbp construct for positively \nsupercoiled DNA indicated up to ~50% of bp for the brightest puncta (σ = 0.14) and ~40% for the \nσ = 0.11 case are associated with plectonemes, while for negatively supercoiled DNA the brightest \npuncta contained closer to ~10% of the number of bp in the total tether length, broadly comparable \nwith the range of bp per plectoneme in λ DNA. Puncta tracking indicated displacements of \napproximately 100 nm per frame equivalent to an apparent 2D Brownian diffusion coefficient in the \nfocal plane of 310-1,900 nm2/s (Table 1 and Fig. 6a). Note, that these 2D diffusion coefficient \nestimates give indications of the local mobility of plectonemes in the lateral focal plane of a DNA \ntether, though the physical processes of plectoneme mobility parallel (possible sliding movements) \nand perpendicular (likely to be lateral tether fluctuations in addition to plectoneme mobility relative \nto tether) to a tether are likely to be different, which we do not explore here. We found that puncta \nfor positively supercoiled DNA have higher mean diffusion coefficients than for negatively \nsupercoiled by a factor of ~6.  \nWe next predicted plectoneme location using a model based on localized DNA curvature53 and the \nstress-induced destabilization (SIDD) during undertwist algorithm54 to model the likelihood of \nforming melting bubbles and plectonemes at each different bp position along the DNA (methods), \nusing the same sequence of the 15 kbp construct and so presenting an excellent opportunity for \ncross-validation between experiment and simulation. This analysis showed agreement for bubble \nformation to the location of puncta observed during negative supercoiling at low force off-centre at \n~4.5 kbp (within 3.6% of our predictions, Fig. 6b). SIDD focuses on locating denaturation bubbles and \nhas limitations in assuming torsional stress is partitioned into twist, therefore over-predicting bubble \nprevalence in systems in which plectonemes are present, and failing to account for influences of \nDNA curvature on plectonemes, and by extension, bubble formation. The presence of a predicted \npeak in the centre 7 kb region was likely an artifact of over-prediction, borne out by the integrated \narea under the suite of ~6 peaks pooled around ~4.5 kbp being over five times greater than the \n7 kbp peak; it is likely this secondary bubble might have effectively been “replaced” by the \nplectoneme. Considering the high energy cost associated with initialisation of a run of strand \nseparation of 10.84 kcal/mol55, which is independent of bubble size, the formation of a single bubble \nis favourable in this regime.  \nConversely, plectonemes under the same conditions are predicted to form at the DIG-functionalised \nend of the DNA, opposite to that observed experimentally (Fig. 6c), although the plectoneme \nprediction algorithm was developed for exclusively positively supercoiled DNA and we are not aware \nof studies using it for negative supercoiling. The analysis predicted that plectoneme formation \nduring positive supercoiling was most likely located at the midpoint of the DNA tether, consistent \nwith experimental observations of puncta (Fig. 6c).  \nFormation of a localized bubble is a precursor of dsDNA melting, leaving two regions of single-\nstranded DNA (ssDNA) with high local flexibility which can form the tip of a plectoneme, a scenario \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nwhich is energetically and entropically favourable compared to forming a second bubble. We \ntherefore hypothesise that bubbles generated through low negative supercoiling act as subsequent \nnucleation points for plectonemes as negative supercoiling increases. SYBR Gold is capable of \nbinding to ssDNA in addition to dsDNA, with the result that during the sampling time used \nexperimentally, rapidly fluctuating ssDNA conformations will result in localized image blur, which is \nwhat we observed experimentally (Fig. 5d). However, fluorescence imaging of cy3b-SSB (methods), \nwhich indicated binding to ssDNA oligos, showed no binding at comparable levels in tethers, which \nmay mean that the typical bubble size is smaller than the probe ~50 nt footprint56; although SIDD \nmodelling indicated bubble sizes of ~60 nt these are limited by not considering plectoneme \ndependence, which may result in smaller effective bubble sizes. For positive supercoiling, \nconsidering only the intrinsic curvature and sequence accurately predicted the plectoneme location, \nas described previously. \nTo better understand bubble/plectoneme coupling, we performed MD simulations of DNA \nfragments held at constant force with varying σ (Fig. 6d,e, methods and Supplementary Movies 13-\n16). We found that plectoneme formation was always accompanied by bubbles forming at or near \nplectoneme tips, caused by significant bending in this region, while relaxed DNA showed no \nevidence for bubble or plectoneme formation.  However, the nature of these bubbles varied \nbetween over- and undertwisted fragments. For overtwisted, bubbles remained relatively close to a \ncanonical conformation, with bases on opposite strands still pointing towards each other. For \nundertwisted, the structural damage was catastrophic; base pairs were melted and bases face away \nfrom the helical axis, leaving large ssDNA regions. In addition, bubbles in undertwisted DNA tended \nto be longer (up to 12 bp) and more stable (lasting for up to 2,200 ns in our simulations, though \nsince simulations were performed in implicit solvent there is no direct mapping to experimental \ntimescales) than the ones formed in overtwisted DNA (maximum of 5 bp and 100 ns) resulting in a \nsignificantly higher prevalence of bubbles for negatively supercoiled vs positively supercoiled DNA \n(Fig. 6d and Supplementary Fig. 15). This agrees with previous experimental studies:  while the \npresence of defects has been widely reported in negatively supercoiled DNA, they have only been \ndetected recently in strongly positive supercoiling with σ > 0.06 57. We hypothesise that the differing \nnature, size and frequency of these bubbles accounts for the relative mobilities of plectonemes we \nobserve; for the less disturbed positively supercoiled plectonemes, motions away from bubbles have \na lower energy barrier than for undertwisted plectonemes which would need to form a large highly \ndisrupted site during motion.   \n \nσ rmsd‖ (nm) rmsd⊥ (nm) rmsdtot (nm) Dtot \n(nm2s-1) \n0.14 108 ± 37 96 ± 58 144 ± 47 1,900 ± 1,200 \n0.11 76 ± 52 72 ± 50 105 ± 41 990 ± 770 \n-0.14 49 ± 35 37 ± 14 62 ± 28 310 ± 280 \nTable 1 Plectoneme mobility is supercoiling-dependent. 1-dimensional root mean square \ndisplacements parallel (rmsd‖) and perpendicular to tether (rmsd⊥), and combined total 2-\ndimensional rmds (rmstot) and diffusion coefficient D for the puncta shown in Fig. 5e. The \nperpendicular thermal motion of the same tether without applied supercoiling was found to be \n66 ± 46 nm, s.d. errors.  \n \n \n \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n \nFigure 5 Plectoneme dynamics in single molecules. a) 1.0 pN continuous asymmetric hat curve, SYBR Gold added (arrows \nindicate red-green-blue overtwist followed by yellow undertwist, drift-corrected). Inset: Zoom-in (red) plus black running \n1,500-point median filter. b) 0.5 pN discontinuous hat curve showing extension reduction during positive/negative \nsupercoiling (optical resolution error bars +/- 0.03 fractional extension), n=1. c) Fluorescence microscopy of 1 pN \nplectoneme formation at σ = -0.28 (no plectoneme), 0 (no plectoneme) and 0.14 (inset plectoneme), one representative \nexample. d) Line profiles for tethers of c) normalized to σ = 0 spanning blur region of σ = -0.28 tether using same size region \nfor σ = 0 and σ = 0.14 tethers but excluding plectonemes, indicating FWHM of 251, 214, 205 nm for σ = -0.28, 0, 0.14 \nrespectively (s.d. error 0.2%, methods). e) One representative tether from fluorescence microscopy, low force plectoneme \nformation, anchor/magnetic bead indicated with green/magenta circles: when DNA overtwisted a plectoneme \nforms/grows (inset, with altered display settings to match local background). Removing positive supercoiling allows DNA to \nrelax, extend to original length, and removes plectoneme. After negative supercoiling applied, extension decreases and \nplectoneme is formed (inset, Supplementary Movie 6). f) One representative example of longer 48.5 kbp λDNA tether, \n~70% fractional extension (~0.3 pN), more plectonemes form upon torsional manipulation, three indicated (small circles); \nsingle-strand nick forms at ~860 ms (relative to start of fluorescence acquisition) causing torsional relaxation wave which \ndiffuses along tether to remove plectoneme positive supercoiling which disappear sequentially. g) Rate of plectoneme \ndiffusion here decreases with increasing bp content. Diffusion coefficient presented as linear regression of mean square \ndisplacement +/- a 95% confidence interval, plectoneme size as mean values +/- s.e.m. over 41 imaging frames, from one \nDNA tether. h) Integrated plectoneme intensity remain stable before nick (no net bp turnover), though qualitative evidence \nof anticorrelation between cyan/blues traces whose plectonemes separated by a few microns. g) and h) same colour code \nas circled plectonemes in f). Experiments in panels c, e and f each performed once. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n \n \nFigure 6 Plectonemes are mobile but form at preferred locations.  a) Distribution of inter-frame displacements for puncta \nmobility as assessed by 2D Gaussian tracking and calculating frame-to-frame displacement for tether with σ = -0.14, 0.11, \nand 0.14 respectively. Inset: scatter plots of frame-to-frame displacement in parallel to (y) and perpendicular to (x) the \ntether in nm, kymographs taken from a profile along the centre of the tether axis. b) Bubble forming probability density \nfunction as predicted by Twist-DNA58, σ = -0.14. Plotted in blue on same axes is the normalised, background-corrected \nintensity of the plectoneme imaged from a single tether in fluorescence microscopy (Fig. 4c). Red triangles indicate the \nclosest agreement between experimental and predicted peaks (within 3.6%). c) Plectoneme formation probability as \npredicted by the probabilistic model of Kim et al53 with modified average plectoneme sizes. Solid line: -200 turns \n(σ = -0.14), dashed line: +150 turns (σ = 0.11), dotted line: +200 turns (σ = 0.14). Dotted and dashed line overlap so they \ncannot be distinguished in the plot. d) Violin jitter plot indicating the percentage of each simulation over which bubbles are \nseen, determined using n = 200 separate bootstraps for each dataset (methods), median (white circle), interquartile range \n(thick line) and 1.5x interquartile range (thin line) indicated. Lack of error in σ=-0.1, F=0.7 pN case indicates presence of at \nleast one bubble in every frame. e) Representative structures of our simulations highlighting plectonemes in blue and \ndenaturation bubbles in red.       \n \nDiscussion \nThree existing approaches of molecular manipulation correlated with fluorescence visualisation \nenable DNA supercoiling to be studied: 1) utilising stochastic biotin-avidin unbinding26 between an \noptically trapped bead and DNA molecule, followed by transient torsional relaxation and religation \nwhich introduces negative supercoils. Fluorescence microscopy is applied to visualize DNA. The \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\napproach is valuable in not requiring engineering of precise magnetic tweezers as with COMBI-\nTweez, however, the introduction of supercoils is stochastic and limited to negative superhelical \ndensity, whilst COMBI-Tweez enables more scope to explore positive and negative supercoiling \ndependence on DNA topology with higher reproducibility; 2) permanent magnetic tweezers22 in a \nvertical geometry to twist a magnetic bead conjugated to the free end of surface-tethered dye-\nlabelled DNA to form plectonemes through positive supercoiling, followed by orthogonal rotation of \nmagnets to create an obliquely oriented tether which is almost but not entirely parallel to the \nmicroscope focal plane due to imprecision with knowing the exact position of the surface relative to \nthe bead, but which can enable plectoneme dynamics visualisation via epifluorescence. This method \nreports visualisation of mechanically induced plectonemes though not visualization of mechanically \ncontrolled bubbles nor of mechanically induced plectonemes directly correlated in real time to the \nmechanical driving torque; it does not require accurate engineering of coils nor optical tweezers, \nhowever, there is no high-precision force measurement and control compared to COMBI-Tweez, \nthere are potential issues with surface interference and uncertainty in the angle between tether and \ncoverslip leading to imprecision concerning plectoneme mobility and, importantly, it does not enable \nobservation of plectoneme formation at the same time as twisting DNA, so it is not possible to study \nearly onset plectoneme dynamics but only several seconds following mechanical perturbations; 3) \nintercalation of Sytox Orange dye23 by surface-tethered DNA induces torsional stress which is \nresolved by plectoneme formation through positive supercoils, but under certain concentration \nregimes can also enable negative supercoils to be added. This approach is relatively easy to \nconfigure, allows for simultaneous visualization of tethers and plectonemes, and can be performed \nwith higher throughput than COMBI-Tweez. However, a key advantage of COMBI-Tweez is that it can \ncontrol the state of DNA supercoiling directly, reversibly and consistently without relying on \nstochastic intercalating dye binding/unbinding which, unavoidably, varies from tether to tether. \nCOMBI-Tweez is also able to precisely exert biologically relevant force regimes to DNA whilst \nobserving plectoneme formation and dynamics. The independent control of torque and force over a \nphysiological range with high-speed data acquisition coupled to high-precision laser interferometry \nis also a unique capability which can enable biological insights in real time structural dynamics of \nDNA and processes that affect its topology over a more rapid timescale than is possible with existing \napproaches, such as DNA replication, repair and transcription. Also, COMBI-Tweez can monitor how \ntorsional dynamics affects plectonemes in real time, and can quantify torsional interactions between \ntwo DNA molecules to high-precision, which has not to our knowledge been reported with these \nearlier technologies. Being based around a standard inverted microscope, it is cost-effective to \nimplement offering access to interchangeable widefield and narrowfield illumination modes. These \nmodes can be multiplexed with trivial engineering adaptations to enable multicolour excitation or \ncombined with techniques such as polarization microscopy25,59,60 to extract multidimensional data \nconcerning biopolymer molecular conformations.  \nAlthough a previous report indicated the single-molecule fluorescence imaging of plectonemes in \nDNA using intercalators to change σ with both DNA ends attached to a coverslip23, we report here to \nour knowledge the first transverse single-molecule fluorescence images of mechanically controlled \nsupercoiling-induced higher-order DNA structural motifs including both plectonemes and bubbles. In \nthe same earlier report it was indicated that plectonemes had high mobility equivalent to a diffusion \ncoefficient of ~0.13 µm2s-1, which contrasts our observation using COMBI-Tweez of a lower mobility \nequivalent at its highest to ~0.002 µm2s-1 for the shorter 15 kbp construct, but is comparable to that \nobserved for full length λ DNA. Aside from these length differences, there are also differences in \nintercalating dye used, in that the previous study used Sytox Orange dyes, whereas we focus on \nSYBR Gold. We speculate that the different observed diffusion response may also result from the \ninterplay between applied supercoiling, axial force on differently sized plectonemes. The \nplectonemes reported previously were relatively small, equivalent to ~0.15-4 kbp, and were \ngenerated by relatively small supercoiling changes of σ ~-0.025 at forces typically <0.5 pN. \nConversely, in our study we used a higher σ and forces closer to the physiological scale, resulting in \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nlarger plectonemes which we find does affect their mobility, since it is known that DNA in general \nresponds to external mechanical stresses by preserving stretches of B-DNA while other regions are \nhighly distorted19,59. Also, as our previous MD simulations of the emergence of structural motifs in \nDNA in response to stretch and twist indicates31, we find that sequence differences make significant \nimpacts to DNA topological responses to mechanical stress and damage.  \nThe model we applied involves the SIDD algorithm54 which is limited in its ability to accurately \npredict where bubbles form since 1) it assumes torsional stress is partitioned into twist, effectively \nthat the DNA is fully extended, therefore over-predicting bubble prevalence if plectonemes are \npresent; 2) it does not consider DNA curvature on plectoneme or bubble formation. With these \ncaveats, our calculations perform well in the context of single-molecule experiments and previous \nmodelling studies in which we saw excellent agreement to AFM data34,61 for bending angles and radii of \ngyration in addition to validation against bulk elastic properties of DNA demonstrated by the \nSerraNA algorithm we reported previously62. \nThe agreement between predictions of bubbles and plectonemes with the observed positions of \npuncta for negatively supercoiled DNA raises several questions. Firstly, do bubbles nucleate \nplectonemes? This would be entropically and energetically favourable compared to forming two \nbubbles – one at the melting site and one at the plectoneme tip. Previous studies indicate a balance \nbetween denatured and B-DNA that depends on torsional stress19,63, and utilising a pre-existing \nbubble to nucleate plectonemes drives this balance towards more B-DNA, that may be \nphysiologically valuable in maintaining overall structural integrity. Secondly, is the smaller bp \ncontent measured for puncta during undertwisting compared to overtwisting a consequence of the \nseeding of plectoneme nucleation by a bubble? Future studies using dyes sensitive to both ssDNA \nand dsDNA such as Acridine Orange64 may delineate the kinetics of bubble from plectoneme \nformation, since if initial bubble formation takes a comparable time to form compared to \nsubsequent plectoneme nucleation this may explain the time-resolved differences observed in \nplectoneme content between under- and overtwisted DNA. Thirdly, can a bubble act as a “staple”, \nproviding an entropic and energetic barrier to plectoneme diffusion, holding it in place at the \nplectoneme tip – as mismatched/unpaired regions have been shown to do in experiment23 and \nsimulation30? If so, since predictions indicate sequence dependence on bubble formation, could \nplectoneme formation serve a role to regulate DNA topology at “programmed” sites in the genome? \nGiven that in E. coli the genome is negatively supercoiled with a mean σ ~-0.05 could sequence \nrepeats with enhanced likelihoods for bubble formation serve as a hub for plectonemes that prevent \nmechanical signal propagation along DNA, in effect defining genetic endpoints of “topological \ndomains” that have roles in protein binding and gene expression57? \nTo our knowledge, this is the first time that size, position, and mobility of higher-order structural \nmotifs in DNA under mechanical control have been directly measured in real time simultaneously \nwith the driving mechanical perturbation. The dependence of DNA shape and mechanics on its \ninteractions with its local environment represents a divergence from traditional views of DNA seen \nthrough the lens of the Central Dogma of Molecular Biology; COMBI-Tweez has clear potential to \nfacilitate mechanistic studies of DNA topology dependence on the interactions with binding partners \nsuch as enzymes, transcription factors and other nucleic acid strands. And, with trivial adaption, the \ntechnology can be implemented to study other filamentous chiral biopolymers; RNA is an obvious \ncandidate, but also modular proteins such as silk show evidence for interesting torsional properties65 \nwhich have yet to be explored at the single-molecule level in regards to their response to twist. It \nmay also be valuable to use COMBI-Tweez to explore specific mechanistic questions relating to \nemergent features of DNA topology, including the dependence on force and ionic strength on DNA \nbuckling, to further probe the source of very rapid fluctuations we observe in DNA supercoiling using \nbfp detection; the ~2-fold increase in fluctuation amplitude with increasing superhelical density \nduring overtwist may signify transitions between metastable structures of relatively small DNA \nsegments which simulations previously indicated emerge during extension of torsionally-constrained \nDNA in a sequence-dependent manner31. Similarly, our observation of an increase in the lower \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nfrequency power spectral components of buckling DNA merits further investigation to study \nwhether this effect is influenced by sequence, salt and plectoneme formation. Also, there may be \nvalue in using this instrumentation to investigate long-range rapid plectoneme hopping mobility \nreported previously22, as well as probing the role of DNA topology in crucial cell processes involving \ninteraction with DNA binding proteins, such as DNA repair mechanisms66. The collation of single-\nmolecule tools in COMBI-Tweez around a single optical microscope also presents valuable future \nopportunities to integrate even more single-molecule biophysics tools, many of which are developed \naround optical microscopy67–69. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nInstrumentation \nOptical tweezers (Supplementary Fig. 1) were built around an inverted microscope (Nikon Eclipse Ti-\nS, Nikon Instruments Inc.) with NIR trapping laser (opus 1064, Laser Quantum), to overfill objective \naperture (100x, NA 1.45, oil immersion, MRD01095, Nikon Instruments Inc.). Oil immersion \ncondenser (NA 1.4, Nikon Instruments Inc.) recollimated beam, back focal plane imaged onto a \nQPD43 (QP50-6-18u-SD2, First Sensor) and digitized (NI 9222 and NI cDAQ-9174). For most \nfluorescence microscopy, a 488nm wavelength laser provided excitation, detection via a -80°C air-\ncooled EMCCD (Prime 95B, Photometrics, or iXon Ultra 897, Andor Technology Ltd, 55 and 69 \nnm/pixel respectively). Magnetic tweezers were generated using Helmholtz coils (Supplementary \nFig. 2) on an aluminium platform. B field measurements indicated uniformity over several mm \n(maximum DC force 4.2 x 10-5pN, Supplementary Note 1). SWG20 copper wires (05-0240, Rapid \nElectronics Ltd.) were wound onto 3D printed spools (Autodesk Inventor, Object30, material: \nVeroWhitePlus RGD835), 95 turns for two smaller spools, 100 turns for two larger spools70. LabVIEW \nsignals were generated (NI 9263 and NI cDAQ-9174); two bipolar 4-quadrant linear \noperational amplifiers (BOP 20-5M, Kepco, Inc.) received voltages to convert to coil currents. COMBI-\nTweez was optical table mounted (PTQ51504, Thorlabs Inc.) in aluminium walls/card lids, ±0.1 °C \nclimate controlled with air conditioning (MFZ-KA50VA, Mitsubishi Electric). Custom LABVIEW \nsoftware (Supplementary Fig. 3) enabled instrument control/data acquisition.  Phased/modulated \ncurrents were sent to generate a 2D rotating B field; QPD (2 axes), current (2 coils), nanostage (3 \naxes) and camera fire signal were sampled at 50kHz. \nCharacterising tweezers. 40nm beads were incubated with 3µm magnetic beads to impart \nmorphological asymmetry detectable on the QPD as a magnetic bead rotated. A 0.54A sinusoidal 1-\n8Hz current was sent to the larger coil pair, 1.5A to the smaller pair (Supplementary Fig. 5a). We \nevaluated phase shifts between coil input and bead rotation by taking QPD voltage correlated with \nits time-reversed signal to reduce noise for peak detection. Supplementary Fig. 5b plots phase shift \nas peak vs rotation frequency. Data were fitted using linear regression to yield \ngradient -0.429 rad Hz-1, angular stiffness 𝑘𝜃 = 8𝜋𝜂𝑅3/|gradient| = 1.1𝑥103pN ⋅ nm ⋅ rad−1. To \nconfirm no OT/MT interdependence we intermittently switched a rotating 1 Hz B field on/off for 10s \nintervals and monitored QPD signals in the absence of beads, confirming no 1Hz peak (Supplementary \nFig. 5c). We could entirely remove any driving 1Hz signal from QPD responses of rotating trapped \nbeads using narrow bandpass filtering, indicating no intrinsic influence of bead rotation on QPD \nsignals. \nWe measured trapping force as displacement of a magnetic bead from the trap centre (estimated \nfrom mean-filtered QPD signals42, using prior 2D raster scanning of surface-immobilised beads as \ncalibration) multiplied by trap stiffness (determined from corner frequency of a Lorentzian fitted to \nthe trapped bead power spectrum), Supplementary Fig. 5.  \nForce clamping. A Proportional Integral Derivative (PID) feedback loop was enabled between QPD \nand nanostage to keep tether force constant parallel to its axis (defined as x) by dynamically \nrepositioning the nanostage by the difference between set and measured force divided by trap \nstiffness (10pN/µm), using PID optimisation to prevent overshooting/discontinuities. The clamp \nresponse time was ~1s (Supplementary Fig. 6) with set force constant ±0.1pN throughout entire \nover-/undertwisting experiments (Supplementary Fig. 7). Small forces detected parallel to y- and z-\naxes due to angular constraints in trapped beads were minimised by manual nanostage adjustment.  \nDNA preparation. A ~15kbp DNA construct was synthesised by ligating functionalized handles, \ncontaining biotin-16-dUTP or digoxigenin-11-dUTP (Roche, 1093070910/11277065910 respectively), \nto a λ DNA fragment (Supplementary Fig. 8). Two PCR reactions were performed using primers \nNgoMIV forward/NheI reverse, amplifying a 498bp region of plasmid pBS(KS+) generating a 515bp \nhandle. For biotinylation, the ratio of dTTP to biotin-16-dUTP was 1:1 generating ~120 handle \nnucleotides. For digoxigenin-labelling, the ratio of dTTP to digoxigenin-11-dUTP was 6.5:1 generating \n~32 digoxigenin-11-dUTPs. PCR reactions were monitored by gel electrophoresis and handles \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\npurified using a Qiaquick PCR kit before digestion with NgoMIV (biotinylation) or NheI-HF \n(digoxigenin-labelling). The remaining tether comprises a 14.6kb fragment, purified using the \nMonarch gel extraction kit from a 0.5% agarose TBE gel at 2V/cm for 36h. Handles were mixed with \ncut λ in 5:1 λ:handle ratio (two compatible sticky ends for each handle), ligated using T4 DNA ligase \n(NEB). 48.5kbp λ DNA was also created using a previously reported protocol26, removing \nphosphorylation steps using T4 polynucleotide kinase as already supplier-phosphorylated (IDT \nN3011S). \nBead functionalisation. To suppress bead brightness in fluorescence (Supplementary Fig. 9), beads \nwere functionalised; 250µL of 50mg/mL solution of Micromer or Micromer-m (Micromod 01-02-\n503/08-55-303 respectively), were incubated with 62.5µL of 5x MES (Alfa Aesar J61587), 2mg of EDC \nHydrochloride (Fluorochem 024810-25G), and 4mg N-Hydroxysuccinimide (Sigma 130672-5G) at \nroom temperature for 45min while vortexing. Beads were pelleted by centrifugation at 12,000xg and \nresuspended in 200µL of 200µg/mL anti-digoxigenin (Merck Life Sciences 11333089001) or 200µL of \n200µg/mL NeutrAvidin (Thermo Scientific 31000), then further incubated 3h. Beads were pelleted \nand resuspended in 100µL PBS and 25mM glycine quench for 30min. Beads were \ncentrifuged/resuspended 3x in 500µL PBS before resuspending in 250µL PBS/ 0.02% azide, \ngenerating anti-digoxigenin Micromer beads and NeutrAvidin Micromer-M beads, or the converse at \n50mg/mL. \nSample preparation/tethering. A flow cell was prepared using truncated slides by scoring/snapping \na 50x26mm glass slide (631-0114, VWR), forming a double-sided tape plus 22x22mm coverslip \n“tunnel” as described previously71. Flow cells were nominally passivated using polyethylene glycol \n(PEG)72 variants MeO-PEG-NHS and Biotin-PEG-NHS (Iris biotech PEG1165 and PEG1057 \nrespectively). Anchor and trapping beads were commercially carboxylated then functionalized with \nanti-digoxigenin or NeutrAvidin as above. 5μm anti-digoxigenin anchor beads were diluted to \n5mg/mL in PBS/0.02% sodium azide and 2mM MgCl2 and vortexed to disaggregate. 10μL was \nintroduced, inverted and incubated in a humidified chamber 15min room temperature for surface \nimmobilisation; for PEG passivated flow cells 5μm NeutrAvidin anchor beads were used. 20μL \n2mg/mL BSA in PBS was introduced and incubated 5min. Nominally, 10μL DNA 0.12ng/µL in T4 DNA \nligase buffer and with 1μL T4 DNA ligase was introduced and incubated 1h, washed 200μL PBS, 20μL \nimaging buffer introduced, and flow cell was sealed with nail varnish. \nFor stretch-release experiments we used ~1Hz nanostage triangular waves, amplitude ~5µm, \nacquiring for ≥2 consecutive cycles. For fluorescence, the imaging buffer comprised PBS, 1mg/mL \n3μm NeutrAvidin-functionalised Micromer-m beads, 6% glucose, 1mM Trolox, 833ng/mL glucose \noxidase, 166ng/mL catalase, 25nM SYBR Gold. For brightfield-only experiments, the imaging buffer \ncomprised 1mg/mL 3µm NeutrAvidin functionalised Micromer-m beads in PBS. Fluorescence \nexperiments used 40ms exposure time, maximum camera gain. Oblique-angle Slimfield73 \n0.11kW/cm2 was the default mode, but we confirmed compatibility with epifluorescence, TIRF and \nHILO. We first imaged DNA for a few seconds at 0.1mW to enable focusing but avoiding \nphotobleaching/photodamage, then imaged at 1mW. For continuous imaging, we acquired data for \n200 rotations. For discontinuous imaging, we recorded 50 bead rotations using no fluorescence, \npaused 1 min, then10 frames in fluorescence, acquiring ~10 supercoiling states per molecule. \nBrightfield-only experiments were 40ms exposure time, zero camera gain, no camera cooling. σ was \ncalculated as number of bead rotations divided by number of turns in relaxed DNA (B-DNA twist per \nbase pair multiplied by number of bp in the DNA construct). \nAn anchor bead was selected and nearby free magnetic bead optically trapped. Beads were brought \nwithin 500nm and incubated 2min to facilitate tether formation (optimised using fluorescence by \nacquiring 10 frames to visualise DNA and reposition beads). Beads were separated ~5µm and \nvisualised with fluorescence if appropriate to test if a tether had formed (Supplementary Movies 2, \n6). A force-extension curve was generated by oscillating the trapped bead and fitted by a wormlike \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nchain to generate persistence and contour lengths. Throughput, dependent on DNA concentration, \nwas nominally one in ~20 attempts (~5min) using 0.24ng/μL, analysis indicating binding fraction for >1 \ntether was 5% 74. Using 0.48ng/μL enabled controllable formation of two tethers between bead pairs \nto explore the effects of braided DNA. Surface drift was quantified from brightfield video-tracked \nintensity centroid displacements every 30s up to 1h of surface-immobilised beads in the absence of \ntethering, indicating mean of ~15nm/s and ~7nm/s for lateral and axial drift respectively. Trapping \ndrift was assessed using QPD signals from a trapped untethered magnetic bead but aside from \nexpected rms fluctuations of a few tens of nm no directed drift was detected. \nImaging SSB/hRPA. Surface-immobilized 100nt ssDNA-cy5 and ssDNA binding proteins (cy3b-\nSSB75and hRPA-eGFP76) were imaged on PEG passivated slides, incubated 5min at room temperature \nwith 200μg/mL Neutravidin (Thermofisher Scientific, 31000) in PBS followed by 200μL wash. 10μL \n500pM ssDNA-cy5 in PBS was introduced and incubated 10min to allow binding of 5ˈ biotin on \nssDNA-cy5 to Neutravidin on flow cell surface, excess ssDNA-cy5 washed 2x 100μL PBS. 100μL \n100nM cy3b-SSB/hRPA-eGFP in PBS was introduced before the flow cell was sealed. A bespoke \nsingle-molecule TIRF microscope with ~100nm penetration depth77 imaged surface-immobilised \nbinding complexes, excitation was by an Obis LS 50 mW 488nm wavelength laser (hRPA-eGFP); Obis \nLS 50mW 561nm wavelength laser (cy3b-SSB); Obis LX 50mW 640nm wavelength laser (ssDNA-cy5), \n10mW at 14µW/µm2. Both probes showed colocalization when incubated with surface-immobilised \nssDNA-cy5 oligo. Since cy3b-SSB was more photostable and less impaired by steric hindrance than \nhRPA-eGFP we focused on cy3b-SSB to probe DNA tethers. Due to PEG slide passivation, anchor \nbeads were functionalised with NeutrAvidin to bind to PEG-biotin coverslips, so Micomer-M beads \nwere functionalised with anti-digoxigenin. Tether preparation was as before with addition of 10nM \ncy3b-SSB to the imaging buffer, and alternating 488nm/561nm wavelength laser excitation78, \n1mW/10mW respectively. \nWormlike chain fitting. Force-extension data were fitted by a wormlike chain (WLC) using Python3 \nand numpy’s curve_fit to minimise least squares, error estimated from bootstrapping by taking 1% \nof randomly selected data, WLC fitting, iterating 1,000 times, indicating ~0.2% s.d.  \nPuncta analysis. To assess percentage of tether bp in a plectoneme we used ImageJ to integrate \nbackground-corrected pixel intensities associated with puncta normalised to the integrated \nbackground-corrected intensity for the whole tether. To measure two-dimensional rms puncta \ndisplacements, we used single-particle tracking software PySTACHIO79 with mean square \ndisplacement <r2>= 4D.δt where D is the two-dimensional diffusion coefficient, and δt the inter-\nframe time interval 40ms, to 20nm localization precision. \nPlectoneme/bubble calculations. Predictions for plectoneme loci were made using a model53 \nconsidering DNA curvature as determining factor, modified to increase the largest possible \nplectoneme to deal with the ~15kbp construct and higher σ levels. Cutoff was increased from 1kbp, \nas set in the original model, to 8.3kbp for σ=0.14 (or +200 DNA turns), to 6.7kbp for σ=0.11 (+150 \nDNA turns) and to 1.9kbp for σ=-0.14 (or -200 DNA turns); limits were differently established \naccording to experimental estimations of number of plectoneme bp. Predictions for bubble loci were \nmade using the Stress-Induced DNA Destabilization (SIDD) algorithm54 implemented on the Twist-\nDNA program to deal with long sequences at genomic scales58. Calculations were done at σ=-0.14, \n0.1M salt, T=310K. For comparison of experimental fluorescence data to bubble predictions, the \nplectoneme tether line profile was normalised and fitted with a cubic spline, regularised and plotted \non the central 14.6kbp. \nControl of tension and torsion in silico. DNA was modelled under restraints to control tension and \ntorsion mimicking experimental conditions (Fig. 5) using positional and ‘NMR’ restraints of AMBER61. \nOne end of the duplex was fixed by restraining coordinates of O3’ and O5’ atoms of the final bp \n(‘fixed end’). The other (‘mobile end’) was kept at force 0.3 or 0.7pN by applying a linear distance \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\nrestraint to each strand, between O3’ or O5’ of the final bp and two fixed dummy atoms62 used as \nreference points (A and B, Supplementary Fig. 12). Angular restraints were applied for confining \n‘mobile end’ motion to the tether axis (Supplementary Fig. 12, denoted z). To prevent y movement, \nangles θ1 and θ2 were constrained. Another pair of reference points and restraints were \nimplemented to prevent x movement. Supercoiling was relaxed by passing DNA over either end of \nthe molecule. To prevent ‘untying’, we applied one angular restraint per phosphorus atom of the \nbulk of the chain and per end (Supplementary Fig. 12b). Specifically, ψ angles (defined by a \nphosphorus atom, the last mobile O3’ or O5’ on that strand and the corresponding reference point) \nwere forced >90°, creating an excluded volume resembling a bead.  To ensure excluded volume \nrestraints were triggered as infrequently as possible, we added 60 GC bp to the ‘mobile end’ as a \nbuffering molecular stretch which were prevented from bending by dihedral restraints applied to \neach complementary pair of phosphorus atoms (Supplementary Fig. 12c). DNA torsional stress was \nmaintained by ensuring O3’ and O5’ of the first relevant bp of the ‘mobile end’ (61st bp) were co-\nplanar with respective reference points A and B, achieved through a dihedral restraint that reduced \nthe angle between ABF and BAE planes to zero. The ‘fixed end’ was torsionally constrained using an \nequivalent dihedral angle defined by O3’ and O5’ of the next-to-last bp and reference points I and J. \nDNA in silico. The structure of a linear 300bp DNA molecule was built using Amber1880  NAB module \nfrom a randomly generated sequence with 49% AT surrounded by 2 GC bp at the ‘fixed end’ and 60 \nGC bp at the ‘mobile end’, 362bp total; 300bp sequence was: \n1TGCAAGATTT 11GCAACCAGGC 21AGACTTAGCG 31GTAGGTCCTA 41GTGCAGCGGG 51ACTTTTTTTC \n61TATAGTCGTT 71GAGAGGAGGA 81GTCGTCAGAC 91CAGATACCTT 101TGATGTCCTG 111ATTGGAAGGA \n121CCGTTGGCCC 131CCGACCCTTA 141GACAGTGTAC 151TCAGTTCTAT 161AAACGAGCTA 171TTAGATATGA \n181GATCCGTAGA 191TTGAAAAGGG 201TGACGGAATT 211CGCCCGGACG 221CAAAAGACGG \n231ACAGCTAGGT 241ATCCTGAGCA 251CGGTTGCGCG 261TCCGAATCAA 271GCTCCTCTTT 281ACAGGCCCCG \n291GTTTCTGTTG   \nTo determine the default twist, we performed a σ=0 simulation. We used the 3DNA algorithm in \nCPPTRAJ64 as a reference to build under- and overtwisted straight DNA, σ=±0.1. \nMD. Simulations were done in Amber18, performed using CUDA implementation of AMBER’s \npmemd. DNA molecules were implicitly solvated using a generalised Born model, salt concentration \n0.2M with GBneck2 corrections, mbondi3 Born radii set and no cutoff for better reproduction of \nmolecular surfaces, salt bridges and solvation forces. Langevin dynamics was employed using similar \ntemperature regulation as above with collision frequency 0.01 ps to reduce solvent viscosity81. The \nBSC1 forcefield was used to represent the DNA molecule70. A single simulation was calculated for \neach combination of force and σ (=±0.1, force F=0.3, 0.7pN), following our protocols for \nminimisation and equilibration. In addition, we performed a control simulation at σ=0 . We \nperformed 40ns re-equilibration, applying the above restraints with exception of tensile force, for \nallowing plectoneme formation; restraints on the canonical WC H-bonds were added to avoid \npremature double helix disruption and allow distributions of twist and writhe to equilibrate. \nSimulations were extended 500ns-2.3μs depending on convergence, measured by cumulative end-\nto-end distance over time (Supplementary Fig. 15b), calculated using a single NVIDIA Tesla V100 GPU \nfrom the local York Viking cluster at 40ns/day. \nDetermination of bubbles in silico. Melting bubbles were easily identifiable by visual inspection. To \nquantify and ensure we only captured significant denaturation, we assumed a bubble was formed by \n≥3 consecutive bp that didn’t present WC H-bonds, whose angular bp parameters (propeller twist, \nopening and buckle) were ≥2 s.d. from the average obtained from relaxed DNA and that disruption \nlasted for more than 1ns. This information was acquired using the nastruct routine of CPPTRAJ82. \nPercentages of simulations where DNA presented bubbles were calculated considering the last \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n400ns. Standard deviations were calculated using bootstrapping, sampling 1% of each simulation \n200 times. \nWax melting. A 1% w/v suspension of each of three different alkane waxes was made, warmed in a \nwater bath 10°C hotter than the alkane melting point and sonicated 1min before plunge cooling in \nan ice bath, injected into a flow cell and incubating 20min prior to 100µL PBS washing, introducing \n50µg/mL magnetic beads, and sealing the flow cell.  The experiment comprised selecting a surface-\nimmobilized wax particle, trapping a nearby magnetic bead and bringing it in contact with the wax, \nthen leaving it in position for 3min, recording brightfield timelapse movies.  \nTEM/electron diffraction. 3µm diameter magnetic beads at initial concentration 50mg/ml \nwere pelleted by centrifugation 1,000xg 1min, washed in 100% Ethanol by \nresuspension/centrifugation, then infiltrated over 48h in LR White resin (Agar Scientific), and \nsubsequent polymerisation at 60°C for 48h. 70nm sections were cut (Leica Ultracut UCT7 \nultramicrotome and Diatome diamond knife). TEM was carried out on a JEOL 2100+, 200 kV using a \n150µm diameter condenser aperture, brightfield imaging achieved with a 120µm objective lens \naperture.  Simulated electron diffraction was performed for Fe3O4 nanoparticles using a standard Al \ndiffraction sample for calibration at camera length 25cm-1 compared to Selected Area Electron \nDiffraction (SAED) images taken from the edges of beads where nanoparticles were present. SAED \nwas also taken from within beads and showed no crystal structure. \nData availability \nExperimental and simulation data used are publicly available at DOI:10.5281/zenodo.7786636 and \nDOI:10.15124/fd0eb563-9a9f-4c0d-82dc-27a4bf071660 respectively. All graph data provided in \nSource Data file. \nCode availability \nAll code used for instrument design/control, data acquisition/processing/analysis and figure \ngeneration directly relied on: Matlab, LabVIEW, Mathematica, Autodesk Inventor, Jupyter, \nMatplotlib, NumPy, Pandas, SciPy. COMBI-Tweez source code available at https://github.com/york-\nbiophysics83 under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International \nLicense (CC-BY-NC-SA; https://creativecommons.org/licenses/by-nc-sa/4.0/). \nAcknowledgements \nThis work was supported by the Leverhulme Trust (RPG-2017-1340, RPG-2019-156), BBSRC \n(BB/R001235/1, BB/W000555/) and EPSRC (EP/N027639/1, EP/R513386/1, EP/R029407/1, \nEP/T022205/1). Thanks to Mark Dillingham (University of Bristol, UK) and Mauro Modesti (CRCM, \nFrance) for donation of cy3b-SSB and hRPA-eGFP respectively. \nAuthor contributions \nM.L. conceived the study. Z.Z. developed tweezers/imaging instrumentation and control software, \nwhich in the following years of development S.G., J.S. expanded. J.H. developed DNA chemistry. J.S., \nJ.H., S.G. collected data which J.S. analysed. A.N. conceived theoretical molecular dynamics \nsimulations, M.B. performed them. C.S. developed an analytical temperature model. C.S-K, A.K. \nperformed TEM and electron diffraction. J.S., M.L wrote the bulk of the manuscript, with \ncontribution, discussion and revision from all of the authors. M.L. was the project administrator. \nCompeting interests \nAll authors declare that they have no competing interests.  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n \nReferences \n1. Leake, M. C. The physics of life: one molecule at a time. Philos Trans R Soc Lond B Biol \nSci 368, 20120248 (2013). \n2. Bryant, Z. et al. Structural transitions and elasticity from torque measurements on \nDNA. Nature 424, 338–341 (2003). \n3. Wuite, G. J. L., Smith, S. B., Young, M., Keller, D. & Bustamante, C. Single-molecule \nstudies of the effect of template tension on T7 DNA polymerase activity. Nature 404, \n103–106 (2000). \n4. Wen, Y., He, M. Q., Yu, Y. L. & Wang, J. H. Biomolecule-mediated chiral \nnanostructures: a review of chiral mechanism and application. Adv Colloid Interface Sci \n289, (2021). \n5. Ashkin, A. Forces of a single-beam gradient laser trap on a dielectric sphere in the ray \noptics regime. Biophys J 61, 569–82 (1992). \n6. Smith, S. B., Finzi, L. & Bustamante, C. Direct mechanical measurements of the \nelasticity of single DNA molecules by using magnetic beads. Science 258, 1122–6 \n(1992). \n7. Strick, T. R., Allemand, J. F., Bensimon, D. & Croquette, V. Behavior of supercoiled \nDNA. Biophys J 74, 2016–2028 (1998). \n8. Smith, S. B., Cui, Y. & Bustamante, C. Overstretching B-DNA: The Elastic Response of \nIndividual Double-Stranded and Single-Stranded DNA Molecules. Science (1979) 271, \n795–799 (1996). \n9. Heller, I., Hoekstra, T. P., King, G. A., Peterman, E. J. G. & Wuite, G. J. L. Optical \ntweezers analysis of DNA-protein complexes. Chem Rev 114, 3087–3119 (2014). \n10. Baumann, C. G., Smith, S. B., Bloomfield, V. A. & Bustamante, C. Ionic effects on the \nelasticity of single DNA molecules. Proc Natl Acad Sci U S A 94, 6185–6190 (1997). \n11. Brouwer, I. et al. Sliding sleeves of XRCC4-XLF bridge DNA and connect fragments of \nbroken DNA. Nature 535, 566–569 (2016). \n12. Whitley, K. D., Comstock, M. J. & Chemla, Y. R. High-Resolution ‘Fleezers’: Dual-Trap \nOptical Tweezers Combined with Single-Molecule Fluorescence Detection. Methods \nMol Biol 1486, 183–256 (2017). \n13. Meijering, A. E. C. et al. Imaging unlabeled proteins on DNA with super-resolution. \nNucleic Acids Res 48, (2020). \n14. Sheinin, M. Y., Forth, S., Marko, J. F. & Wang, M. D. Underwound DNA under tension: \nstructure, elasticity, and sequence-dependent behaviors. Phys Rev Lett 107, (2011). \n15. Lionnet, T., Joubaud, S., Lavery, R., Bensimon, D. & Croquette, V. Wringing out DNA. \nPhys Rev Lett 96, (2006). \n16. Kriegel, F. et al. The temperature dependence of the helical twist of DNA. Nucleic \nAcids Res 46, 7998–8009 (2018). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n17. Kriegel, F. et al. Probing the salt dependence of the torsional stiffness of DNA by \nmultiplexed magnetic torque tweezers. Nucleic Acids Res 45, 5920–5929 (2017). \n18. Lipfert, J., Kerssemakers, J. W. J., Jager, T. & Dekker, N. H. Magnetic torque tweezers: \nmeasuring torsional stiffness in DNA and RecA-DNA filaments. Nat Methods 7, (2010). \n19. Allemand, J. F., Bensimon, D., Lavery, R. & Croquette, V. Stretched and overwound \nDNA forms a Pauling-like structure with exposed bases. Proc Natl Acad Sci U S A 95, \n14152–14157 (1998). \n20. Vlijm, R., Mashaghi, A., Bernard, S., Modesti, M. & Dekker, C. Experimental phase \ndiagram of negatively supercoiled DNA measured by magnetic tweezers and \nfluorescence. Nanoscale 7, 3205–3216 (2015). \n21. Long, X., Parks, J. W., Bagshaw, C. R. & Stone, M. D. Mechanical unfolding of human \ntelomere G-quadruplex DNA probed by integrated fluorescence and magnetic \ntweezers spectroscopy. Nucleic Acids Res 41, 2746–2755 (2013). \n22. van Loenhout, M. T. J., de Grunt, M. V & Dekker, C. Dynamics of DNA supercoils. \nScience 338, 94–7 (2012). \n23. Ganji, M., Kim, S. H., Van Der Torre, J., Abbondanzieri, E. & Dekker, C. Intercalation-\nBased Single-Molecule Fluorescence Assay To Study DNA Supercoil Dynamics. Nano \nLett 16, 4699–4707 (2016). \n24. King, G. A. et al. Revealing the competition between peeled ssDNA, melting bubbles, \nand S-DNA during DNA overstretching using fluorescence microscopy. Proc Natl Acad \nSci U S A 110, 3859–3864 (2013). \n25. Backer, A. S. et al. Single-molecule polarization microscopy of DNA intercalators sheds \nlight on the structure of S-DNA. Sci Adv 5, (2019). \n26. King, G. A., Burla, F., Peterman, E. J. G. & Wuite, G. J. L. Supercoiling DNA optically. \nProc Natl Acad Sci U S A 116, 26534–26539 (2019). \n27. Ouldridge, T. E., Louis, A. A. & Doye, J. P. K. Structural, mechanical, and \nthermodynamic properties of a coarse-grained DNA model. J Chem Phys 134, (2011). \n28. Wang, Q. & Pettitt, B. M. Modeling DNA thermodynamics under torsional stress. \nBiophys J 106, 1182–1193 (2014). \n29. Mosayebi, M., Louis, A. A., Doye, J. P. K. & Ouldridge, T. E. Force-Induced Rupture of a \nDNA Duplex: From Fundamentals to Force Sensors. ACS Nano 9, 11993–12003 (2015). \n30. Matek, C., Ouldridge, T. E., Doye, J. P. K. & Louis, A. A. Plectoneme tip bubbles: \ncoupled denaturation and writhing in supercoiled DNA. Sci Rep 5, (2015). \n31. Shepherd, J. W., Greenall, R. J., Probert, M. I. J., Noy, A. & Leake, M. C. The emergence \nof sequence-dependent structural motifs in stretched, torsionally constrained DNA. \nNucleic Acids Res 48, 1748–1763 (2020). \n32. Harris, S. A., Sands, Z. A. & Laughton, C. A. Molecular dynamics simulations of duplex \nstretching reveal the importance of entropy in determining the biomechanical \nproperties of DNA. Biophys J 88, 1684–1691 (2005). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n33. Mitchell, J. S., Laughton, C. A. & Harris, S. A. Atomistic simulations reveal bubbles, \nkinks and wrinkles in supercoiled DNA. Nucleic Acids Res 39, 3928–3938 (2011). \n34. Pyne, A. L. B. et al. Base-pair resolution analysis of the effect of supercoiling on DNA \nflexibility and major groove recognition by triplex-forming oligonucleotides. Nat \nCommun 12, (2021). \n35. Schakenraad, K. et al. Hyperstretching DNA. Nat Commun 8, (2017). \n36. Friese, M. E. J., Nieminen, T. A., Heckenberg, N. R. & Rubinsztein-Dunlop, H. Optical \nalignment and spinning of laser-trapped microscopic particles. Nature 1998 394:6691 \n394, 348–350 (1998). \n37. Forth, S. et al. Abrupt buckling transition observed during the plectoneme formation \nof individual DNA molecules. Phys Rev Lett 100, (2008). \n38. Kolbeck, P. J. et al. Molecular structure, DNA binding mode, photophysical properties \nand recommendations for use of SYBR Gold. Nucleic Acids Res 49, 5143–5158 (2021). \n39. Simpson, N. B., Allen, L. & Padgett, M. J. Optical tweezers and optical spanners with \nLaguerre–Gaussian modes. J Mod Opt 43, 2485–2491 (1996). \n40. Bustamante, C. J., Chemla, Y. R., Liu, S. & Wang, M. D. Optical tweezers in single-\nmolecule biophysics. Nature reviews. Methods primers 1, (2021). \n41. Miller, H., Zhou, Z., Shepherd, J., Wollman, A. J. M. M. & Leake, M. C. Single-molecule \ntechniques in biophysics: a review of the progress in methods and applications. \nReports on Progress in Physics 81, 024601 (2017). \n42. Gao, X., Hong, Y., Ye, F., Inman, J. T. & Wang, M. D. Torsional Stiffness of Extended and \nPlectonemic DNA. Phys Rev Lett 127, (2021). \n43. Gittes, F. & Schmidt, C. F. Interference model for back-focal-plane displacement \ndetection in optical tweezers. Opt Lett 23, 7 (1998). \n44. Fatima, H., Charinpanitkul, T. & Kim, K. S. Fundamentals to Apply Magnetic \nNanoparticles for Hyperthermia Therapy. Nanomaterials 11, (2021). \n45. Marko, J. F. & Siggia, E. D. Stretching DNA. Macromolecules 28, 8759–8770 (1995). \n46. Kriegel, F., Ermann, N. & Lipfert, J. Probing the mechanical properties, conformational \nchanges, and interactions of nucleic acids with magnetic tweezers. J Struct Biol 197, \n26–36 (2017). \n47. Linke, W. A. A. & Leake, M. C. C. Multiple sources of passive stress relaxation in muscle \nfibres. Phys Med Biol 49, 3613–27 (2004). \n48. Meng, H., Bosman, J., Van Der Heijden, T. & Van Noort, J. Coexistence of twisted, \nplectonemic, and melted DNA in small topological domains. Biophys J 106, 1174–1181 \n(2014). \n49. Strick, T. R., Allemand, J.-F., Bensimon, D., Bensimon, A. & Croquette, V. The Elasticity \nof a Single Supercoiled DNA Molecule. Science (1979) 271, 1835–1837 (1996). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n50. Plank, M., Wadhams, G. H. & Leake, M. C. Millisecond timescale slimfield imaging and \nautomated quantification of single fluorescent protein molecules for use in probing \ncomplex biological processes. Integrative Biology 1, 602–612 (2009). \n51. Tokunaga, M., Imamoto, N. & Sakata-Sogawa, K. Highly inclined thin illumination \nenables clear single-molecule imaging in cells. Nat Methods 5, 159–161 (2008). \n52. Gutiérrez Fosado, Y. A., Landuzzi, F. & Sakaue, T. Twist dynamics and buckling \ninstability of ring DNA: the effect of groove asymmetry and anisotropic bending. Soft \nMatter 17, 1530–1537 (2021). \n53. Kim, S. H. et al. DNA sequence encodes the position of DNA supercoils. Elife 7, (2018). \n54. Benham, C. J. & Bi, C. The analysis of stress-induced duplex destabilization in long \ngenomic DNA sequences. J Comput Biol 11, 519–543 (2004). \n55. Benham, C. J. Energetics of the strand separation transition in superhelical DNA. J Mol \nBiol 225, 835–847 (1992). \n56. Bell, J. C., Liu, B. & Kowalczykowski, S. C. Imaging and energetics of single SSB-ssDNA \nmolecules reveal intramolecular condensation and insight into RecOR function. Elife 4, \n(2015). \n57. Fogg, J. M., Judge, A. K., Stricker, E., Chan, H. L. & Zechiedrich, L. Supercoiling and \nlooping promote DNA base accessibility and coordination among distant sites. Nat \nCommun 12, (2021). \n58. Jost, D. Twist-DNA: computing base-pair and bubble opening probabilities in genomic \nsuperhelical DNA. Bioinformatics 29, 2479–2481 (2013). \n59. Backer, A. S. et al. Elucidating the Role of Topological Constraint on the Structure of \nOverstretched DNA Using Fluorescence Polarization Microscopy. J Phys Chem B 125, \n8351–8361 (2021). \n60. Shepherd, J. W., Payne-Dwyer, A. L., Lee, J. E., Syeda, A. & Leake, M. C. Combining \nsingle-molecule super-resolved localization microscopy with fluorescence polarization \nimaging to study cellular processes. Journal of Physics: Photonics 3, 034010 (2021). \n61. Yoshua, S. B. et al. Integration host factor bends and bridges DNA in a multiplicity of \nbinding modes with varying specificity. Nucleic Acids Res 49, 8684–8698 (2021). \n62. Velasco-Berrelleza, V. et al. SerraNA: a program to determine nucleic acids elasticity \nfrom simulation data. Phys Chem Chem Phys 22, 19254–19266 (2020). \n63. Randall, G. L., Zechiedrich, L. & Pettitt, B. M. In the absence of writhe, DNA relieves \ntorsional stress with localized, sequence-dependent structural failure to preserve B-\nform. Nucleic Acids Res 37, 5568–5577 (2009). \n64. Miller, H. et al. Biophysical characterisation of DNA origami nanostructures reveals \ninaccessibility to intercalation binding sites. Nanotechnology 31, (2020). \n65. Emile, O., Le Floch, A. & Vollrath, F. Biopolymers: shape memory in spider draglines. \nNature 440, 621 (2006). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n66. J. M Wollman, A. et al. Tetrameric UvrD Helicase Is Located at the E. Coli Replisome \ndue to Frequent Replication Blocks. J Mol Biol 436, 168369 (2024). \n67. Shashkova, S. & Leake, M. C. Single-molecule fluorescence microscopy review: \nshedding new light on old problems. Biosci Rep 37, (2017). \n68. Leake, M. C. C. Analytical tools for single-molecule fluorescence imaging in cellulo. \nPhys Chem Chem Phys 16, 12635–47 (2014). \n69. Harriman, O. L. J. & Leake, M. C. Single molecule experimentation in biological physics: \nexploring the living component of soft condensed matter one molecule at a time. J \nPhys Condens Matter 23, (2011). \n70. Zhou, Z., Miller, H., Wollman, A. & Leake, M. Developing a New Biophysical Tool to \nCombine Magneto-Optical Tweezers with Super-Resolution Fluorescence Microscopy. \nPhotonics 2, 758–772 (2015). \n71. Leake, M. C. et al. Stoichiometry and turnover in single, functioning membrane \nprotein complexes. Nature 443, 355–358 (2006). \n72. Paul, T. & Myong, S. Protocol for generation and regeneration of PEG-passivated slides \nfor single-molecule measurements. STAR Protoc 3, 101152 (2022). \n73. Shashkova, S. & Leake, M. C. Single-molecule fluorescence microscopy review: \nshedding new light on old problems. Biosci Rep 37, (2017). \n74. Leake, M. C., Wilson, D., Gautel, M. & Simmons, R. M. The elasticity of single titin \nmolecules using a two-bead optical tweezers assay. Biophys J 87, 1112–35 (2004). \n75. Dillingham, M. S. et al. Fluorescent single-stranded DNA binding protein as a probe for \nsensitive, real-time assays of helicase activity. Biophys J 95, 3330–3339 (2008). \n76. Modesti, M. Fluorescent labeling of proteins. Methods Mol Biol 783, 101–120 (2011). \n77. Dresser, L. et al. Amyloid-β oligomerization monitored by single-molecule stepwise \nphotobleaching. Methods 193, 80–95 (2021). \n78. Badrinarayanan, A., Reyes-Lamothe, R., Uphoff, S., Leake, M. C. & Sherratt, D. J. In vivo \narchitecture and action of bacterial structural maintenance of chromosome proteins. \nScience 338, 528–31 (2012). \n79. Shepherd, J. W., Higgins, E. J., Wollman, A. J. M. & Leake, M. C. PySTACHIO: Python \nSingle-molecule TrAcking stoiCHiometry Intensity and simulatiOn, a flexible, \nextensible, beginner-friendly and optimized program for analysis of single-molecule \nmicroscopy data. Comput Struct Biotechnol J 19, 4049–4058 (2021). \n80. Macke, T. J. & Case, D. A. Modeling Unusual Nucleic Acid Structures. ACS Symposium \nSeries 682, 379–393 (1998). \n81. Watson, G. D., Chan, E. W., Leake, M. C. & Noy, A. Structural interplay between DNA-\nshape protein recognition and supercoiling: The case of IHF. Comput Struct Biotechnol \nJ 20, 5264–5274 (2022). \n82. Roe, D. R. & Cheatham, T. E. PTRAJ and CPPTRAJ: Software for Processing and Analysis \nof Molecular Dynamics Trajectory Data. J Chem Theory Comput 9, 3084–3095 (2013). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint \n\n83. york-biophysics/COMBI-Tweez: COMBI-Tweez control panel v1.0. \nhttps://zenodo.org/doi/10.5281/zenodo.10779228 (2024). \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted March 5, 2024. ; https://doi.org/10.1101/2024.01.18.576226doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}