The Smc5/6 complex is a DNA loop-extruding motor – Nature

Gene synthesis, subcloning and strain creation for Smc5/6 overexpression

Genes for all Smc5/6 subunits were synthesized by GeneArt Gene Synthesis (Thermo Fisher Scientific), with codon optimization and introduced into pJF2, pJF3, pJF4 and pJF5 yeast integrative vectors (kindly provided by J. Diffley) under the bidirectional GAL1-10 promoter²⁹. The TAP-tag sequence derived from pBS1479 (EUROSCARF) was introduced into the C terminus of Smc6 or Nse5 using standard methods. The following plasmids—CD373; SMC5-GAL1-10-SMC6-TAP, CD380; NSE5-pGAL1-10-NSE6, CD395; NSE3-pGAL1-10-NSE4, CD377; and NSE1-pGAL1-10-NSE2—were integrated into CB3245 using auxotrophic markers, then the TOP1 gene was deleted using standard gene-replacement methods. For ATPase mutants, point mutations were introduced using standard methods at appropriate positions (Walker A (KE): Smc5K75E, Smc6K115E; Walker B (EQ): Smc5E1015Q, Smc6E1048Q)³¹. A codon-optimized SNAP tag was synthesized by GeneArt Gene Synthesis and introduced into pFA6a with a KAN marker. The SNAP tag was introduced into the C terminus of ectopic NSE2, NSE4 and NSE5 after the 6×His tag using standard methods. A list of yeast strains and plasmid DNA used in this study can be found in Supplementary Tables 1 and 2.

Overexpression and purification of Smc5/6 and subcomplex Nse5/6

Overexpression strains were grown at 30 °C in 1 or 2 l of YEP-lactate medium to optical density (OD₆₀₀) 0.8–1.0, then protein expression was induced for 4 h by the addition of 2% galactose. After harvesting, cells were disrupted using Freezer mill 6870 (SPEX), and proteins were extracted by the addition of one cell volume of IPP150 buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 10% glycerol, 0.1% IGEPAL CA-630, 1 mM DTT) containing 10 mM MgCl₂ and complete EDTA-free protease inhibitor (Roche Applied Science), after which treatment with benzonase (Merck) was performed for 1 h at 4 °C. Cleared extracts were mixed with IgG Sepharose 6 FF (Merck) for 2 h at 4 °C and washed with IPP150 buffer. IPP150 buffer was then replaced with GF500 buffer (20 mM HEPES-NaOH pH 7.5, 500 mM NaCl, 10% glycerol, 0.1% IGEPAL CA-630, 1 mM DTT) and the resin treated with tobacco etch virus (TEV) protease (kind gift from H. Schüler) at 4 °C overnight. The fraction eluted by TEV proteinase treatment was diluted fourfold in CBB500 buffer (50 mM Tris-HCl pH 8.0, 500 mM NaCl, 1 mM Mg(CH₃COO)₂, 1 mM imidazole, 2 mM CaCl₂, 1 mM DTT, 0.1% IGEPAL CA-630) supplemented with 1 M CaCl₂ (30 µl for 40 ml of mixture) and incubated with calmodulin Sepharose 4B (Merck) for 2 h at 4 °C. After washing with CBB500 buffer, proteins were eluted using CEB500 buffer (50 mM Tris-HCl pH 8.0, 500 mM NaCl, 1 mM Mg(CH₃COO)₂, 1 mM imidazole, 20 mM EGTA, 1 mM DTT, 0.1% IGEPAL CA-630). The eluate was concentrated by around 50-fold using a Vivaspin20 ultrafiltration unit (100 K MWCO, Sartorius) concomitant with an exchange to STO500 buffer (50 mM Tris-HCl pH 8.0, 500 mM NaCl, 2 mM MgCl₂, 0.5 mM tris(2-carboxyethyl)phosphine (TCEP), 10% glycerol, 0.1% IGEPAL CA-630). Concentration of the complex was determined by Bradford assay using bovine serum albumin (BSA) as standard. The integrity of purified Smc5/6 was tested using size-exclusion chromatography on a Superose 6 Increase 10/300 GL column (GE healthcare), pre-equilibrated with STO500 buffer and subsequent SDS–polyacrylamide gel electrophoresis (SDS–PAGE) analysis of eluted fractions (see Extended Data Fig. 1b for the WT octameric complex and Extended Data Fig. 1j for a hexameric complex lacking subunits Nse5 and Nse6).

Fluorescent labelling of Smc5/6

The Smc5/6 complexes containing C-terminally tagged Nse2-6xHis-SNAP, Nse4-6xHis-SNAP or Nse5-6xHis-SNAP were overexpressed and purified using IgG Sepharose 6 FF as described above. After TEV protease cleavage, the eluate was concentrated by around 50-fold using a Vivaspin20 ultrafiltration unit (100 K MWCO, Satorius) concomitant with an exchange to STO500 buffer. For fluorescent labelling, the eluate was mixed with 20 µM SNAP-Surface Alexa Flour 647 (NEB) in 50 µl of STO500 buffer supplemented with 50 mM DTT and incubated overnight at 4 °C. The mixture was concentrated by approximately tenfold using an Amicon Ultra centrifugal filter (100 K MWCO, Merck) concomitant with buffer exchange to fresh STO500 for removal of free Alexa Fluor 647.

Labelling efficiency estimation

Labelling efficiency was calculated in two steps using Smc5/6 containing Nse4- (or Nse2-) 6xHis-SNAP-Alexa 647. The amount of Smc5/6 was first estimated by Bradford assay using BSA as standard, which was 7.56 ± 0.5 µM. The amount of label (Alexa 647) was then estimated by comparison of both absorption and fluorescence intensity of a known concentration (for example, 1 µM) of Alexa 647 in the same storage buffer used for labelled Smc5/6. Both absorption and fluorescence measurements yielded a labelling efficiency of 68 ± 10%, within their respective error.

ATPase assay

Smc5/6 (0.5 µl, final concentration 30 nM) was incubated with 4 nCi [α-³²P]ATP in 5 µl of the reaction buffer (50 mM Tris-HCl pH 7.6, 40 mM KCl, 1 mM MgCl₂, 1 mM DTT, 0.1 mg ml^–1 BSA) containing 1 mM ATP and various concentrations of pRS316 at 30 °C. Aliquots (1 µl) were collected every 30 min for 90 min and mixed with 1.5 µl of 1% SDS to stop the reaction. Then, 1 µl of the mixture was spotted on TLC PEI cellulose F plates (MERCK) and developed in 1 M HCOOH/0.5 M LiCl. Radiolabelled ATP and ADP were quantified using a LAS-3000 imager (Fujifilm). ATPase rates at each DNA concentration were calculated by linear regression using the least-squares method. Maximum ATPase rate and 95% confidence interval for the WT complex were obtained by fitting of a stimulatory dose–response model to experimental data by nonlinear regression using Prism 9 software (GraphPad).

Highly inclined optical light sheet microscopy and data collection

A custom-built microscope was used for single-molecule visualization of DNA and labelled Smc5/6. Lasers with wavelengths of 638 nm (Cobolt) and 561 nm (Coherent) were coupled to a Zeiss (AxioVert200) microscope body through a single-mode fibre in wide-field illumination mode with the potential of changing the illumination angle. This setup allowed us to use highly inclined optical light sheet illumination using a total internal reflection fluorescent objective (alpha-Plan-APOCHROMAT ×100/1.46 numerical aperture, oil) for selective imaging of DNA and Smc5/6 while minimizing out-of-focus fluorescence background and bleaching. The fluorescence signal from the sample was spectrally selected by a dichroic filter (no. t405/488/561/640rpc2, Chroma) and recorded with a sCMOS (PCO edge 4.2) camera. Light from the excitation lasers (638 and 561 nm) was additionally suppressed using a multiband notch filter (no. NF03-405/488/561/635E-25, Semrock) located in front of the camera. For simultaneous imaging of DNA and Smc5/6, alternative excitation between the 561 and 638 nm lasers was used through electronic triggering of an acoustic-optic tunable filter (MPDSnCxx-ed1-18 and AOTFnC_MDS driver from AA-Optoelectronic). The temperature of the flow cell was controlled by adjustment of electric current sent through a self-adhesive heating foil (Thermo TECH Polyester Heating foil self-adhesive 12 V DC, 12 V AC 17 W IP rating IPX4 (L × W) 65 × 10 mm²) attached to the top of the glass slide. The temperature was set to 30 °C for all experiments unless stated otherwise. A custom-written python software was used for recording, storing and visualization of data. Specifically, we utilized PyQtGraph (https://github.com/pyqtgraph/pyqtgraph) and napari (https://github.com/Napari/napari)³² for visualization and export of images. Typically, images were recorded at 100 ms exposure time per frame for a duration of 1,000–2,000 s unless stated otherwise.

Mass photometry experiments

Mass photometry measurements were carried out on a TwoMP device (Refeyn). Glass coverslips were rinsed in the following order: deionized water, 50% isopropanol, deionized water, 50% isopropanol and water, followed by drying in a clean nitrogen stream. The flow chamber was assembled as described in ref. ³³. Before measurements, samples were diluted to a final concentration of 10 nM and incubated in assay buffer containing 40 mM Tris-HCl pH 7.5, 100 mM potassium glutamate and 7.5 mM MgCl₂ at 30 °C for 10 min. All buffers used for mass photometry experiments were filtered with a 0.22 µm syringe filter (with a polyvinylidene difluoride membrane, Merck Millex). Mass photometry experiments with oligonucleotides were performed with 10 nM Smc5/6 hexameric or octameric complexes (WT or EQ mutant) and 5 nM 200 base pair (bp), linear double-stranded DNA(5′-TGGTTTTTATATGTTTTGTTATGTATTGTTTATTTTCCCTTTAATTTTAGGATATGAAAACAAGAATTTATCTGGTTTTTATATGTTTTGTTATGTATTGTTTATTTTCCCTTTAATTTTAGGATATGAAAACAAGAATTTATCTGGTTTTTATATGTTTTGTTATGTATTGTTTATTTTCCCTTTAATTTTAGGATATG-3′), unless stated otherwise. If stated, 10 nM Nse5/6 and 2.5 mM ATP were supplemented to the reaction. Immediately after injection of the sample into the flow chamber, images were acquired for 60 s at 135 Hz in all measurements. After each measurement the chamber was rinsed in the following order: water, 1 M NaCl, water and assay buffer. Data analysis was performed by DiscoveryMP (Refeyn). For contrast to mass conversion, the mass of the Smc5/6 hexamer EQ mutant without DNA and ATP was used as calibrant on the same day as each measurement. All samples were measured at least three times, unless stated otherwise.

Single-molecule loop-extrusion assay

Flow cell preparation

The single-molecule assay used throughout this work was prepared as described previously^9,25,30 with the following slight modifications: microscope slides were cleaned with acid piranha (sulfuric acid (five parts) and hydrogen peroxide (one part)) and silanized with 3-[(2-aminoethyl)aminopropyl] trimethoxysilane in methanol containing 5% glacial acetic acid, which leaves free amine groups on the surface. Slides were then treated with 5 mg ml^–1 methoxy-PEG-N-hydroxysuccinimide (no. MW 3500, Laysan Bio) and 0.05 mg ml^–1 biotin-PEG-N-hydroxysuccinimide (no. MW3400, Laysan Bio) in 50 mM borate buffer pH 9.0. The pegylation step was repeated five times to minimize nonspecific surface sticking of proteins. Pegylated slides were dried under a gentle flow of nitrogen, sealed and stored at −20 °C until further use. Flow cells were then assembled with the functionalized glass slides as previously described⁹. Each flow cell contained one inlet and two outlet channels to facilitate buffer flow application perpendicular to the axis of the immobilized DNA. The fluidic channels were first incubated with 100 nM streptavidin in T50 buffer (40 mM Tris-HCl pH 7.5, 50 mM NaCl) for 1 min and then washed thoroughly with T50 buffer. Subsequently, 10 pM Phage λ-DNA molecules, labelled with biotins at both ends³¹, were introduced to the flow cell at a constant speed of 3 μl min^–1, resulting in surface immobilization of DNA molecules with relative DNA extensions ranging from 0.1 to 0.6. Unbound DNA molecules were later washed out. To minimize unwanted surface sticking of Smc5/6, the flow cell was further passivated by incubation with 0.5 mg ml^–1 BSA for 5 min.

Single-molecule imaging of Smc5/6-mediated loop extrusion

Real-time imaging of Smc5/6-mediated loop extrusion was carried out as follows. The imaging buffer (40 mM Tris-HCl pH 7.5, 100 mM NaCl, 7.5 mM MgCl₂, 0.5 mg ml^–1 BSA, 1 mM TCEP, 2 mM ATP, 200 nM SxO, 30 mM d-glucose, 2 mM trolox, 10 nM catalase, 37.5 µM glucose oxidase) containing Smc5/6 (2 nM, unless stated otherwise) was introduced into the flow cell at a flow rate of 30 µl min^–1 for 1 min and flow was stopped thereafter. For the side-flow experiment, a larger volume (200–300 µl) of the sample of the same composition was continuously flowed into the channel at 15 µl min^–1. If only SxO-stained DNA was imaged, only the 561 nm laser was used, at an intensity of 0.1 W cm^–2 whereas, for dual-colour imaging, SxO-stained DNA and Alexa 647-labelled Smc5/6 were imaged by alternating excitation using 531 nm (0.1 W cm^–2) and 638 nm (about 150 W cm^–2) lasers.

Single-molecule analysis of high-salt-resistant Smc5/6–DNA binding

For the estimation of high-salt-resistant DNA entrapment by octameric Smc5/6 containing Nse5/6, Nse2-SNAP-Alexa 647-labelled Smc5/6 was incubated with DNA in our single-molecule assay for 1 h at 2 nM concentration, which led to the accumulation of Smc5/6 at the ends of DNA (Extended Data Fig. 7k). We further enhanced the number of Smc5/6 bindings on DNA using 100 mM potassium glutamate rather than 100 mM NaCl in the imaging buffer (Extended Data Fig. 9e). Subsequently we exchanged the imaging buffer with 1 M NaCl containing 500 nM SxO at a flow rate of 10 µl min^–1 and imaged the labelled Smc5/6 during high-salt washing to estimate the number of remaining Smc5/6 molecules after washing. To minimize the effect of bleaching, and thus to avoid underestimation of the number of Smc5/6, labelled Smc5/6 was imaged for short intervals during high-salt washing. In the case of hexameric complexes the same experimental conditions were used, including 2 nM protein concentrations and Nse4-SNAP-Alexa 647-labelled hexamers.

Data analysis

Fluorescence images were analysed using a custom-written python software^32,34 Regions containing λ-DNA molecules were chosen manually, cropped and saved into TIFF format. For snapshots of the molecules shown in this paper (Fig. 1e and others), the background was subtracted using the ‘white_tophat’ filter in scipy³³. For further quantification of fluorescence intensity, an additional median filter (radius two pixels) was applied. Snapshots of labelled Smc5/6 (Fig. 2a,b,e) were denoised using a machine learning-based method (Noise2void) for better visualization³⁴. For further quantification of fluorescence intensity (that is, for building of kymographs), however, the same median filter used for the DNA was employed. From the median filtered images, kymographs were subsequently built by summation of fluorescence intensity greater than 11 pixels along lines centred around the DNA axis (Fig. 1g,i). Intensities in the kymograph were normalized such that values outside of DNA approached zero. Each vertical pixillated line in the kymograph corresponds to one time point (one image-frame) of the image sequence.

Estimation of DNA size

The position of the DNA punctum—the centre position of a DNA loop—in the DNA kymograph was found using the ‘find_peaks’ algorithm in scipy, which determines the positions of local peak maxima for each line of the kymograph. We selected the most intense peak along each line on the kymograph. The intensity of DNA puncta—the entire region containing the DNA loop—was obtained by summing over the area of a square with side length seven pixels and centred around the punctum position, termed Int_loop. The remaining DNA regions outside of the puncta were separated into two categories, termed Int_up and Int_down, where ‘up/down’ is the intensity from the DNA region above/below the puncta (Fig. 1h). The amounts of DNA in, above and below the loop were estimated by multiplying the fraction of the intensity in the respective regions by 48.5 kbp (the length of the used lambda DNA in bp):

Changes in DNA size in the respective regions were plotted as a function of time, as shown in Fig. 1h,j. These data were plotted, together with the smoothed data, using a Savitzky–Golay second-order filter and window size of 50 points (solid curve, Fig. 1h,j). Looking at Fig. 1n, the loop extrusion was termed two-sided if the increase in I_loop was correlated with a decrease in both I_up and I_down, otherwise the events were deemed one-sided. To estimate the rate (k) of loop extrusion, the initial 5 s of the loop growth curve was fitted (Fig. 1l and Extended Data Fig. 3b) with a line I_loop = kt + c, where c compensates for the initial DNA amount before loop extrusion. Time traces of loop-extrusion rates were then obtained using (kleft(tright)=left[{I}_{{rm{loop}}}left(t+dtright)-,{I}_{{rm{loop}}}left(t-dtright)right]/2dt), with dt = 1 s and consequent smoothing with the Savitzky–Golay filter as described above (Extended Data Fig. 3c). For force estimation, relative extension was first calculated as ({R}_{{rm{ext}}}=48502d/left(left({I}_{{rm{up}}}+{I}_{{rm{down}}}right){L}_{{rm{C}}}right)), where L_C = 16 μm is the contour length of λ-DNA and d is the end-to-end distance of double-tethered DNA (in μm) (Extended Data Fig. 3d,e). Subsequently, relative extension was converted to force (Extended Data Fig. 3f) via linear interpolation of the force–extension curve obtained by magnetic tweezer force spectroscopy³⁵. The force at maximum loop size was taken as the stalling force (Fig. 1m).

Estimation of number of Smc5/6

For determination of the number of Smc5/6 required for loop extrusion (Fig. 2c,d,f,g), image sequences recorded using ALEX were used to build kymographs of DNA and labelled Smc5/6 intensities. The areas used to determine intensities of DNA puncta were then utilized to determine Smc5/6 intensities and build intensity time traces, which were then used to count bleaching steps (Fig. 2d) and to build intensity histograms (Fig. 2h). For determination of the photobleaching statistics shown in Fig. 2i we included only molecules for which we observed loop initiation during the recording interval—that is, loops already initiated before recording started were excluded from analysis. Bleaching times for one-step bleaching (Δτ₁₁) and those for two-step bleaching (Δτ₂₁, Δτ₂₂) were than used to calculate the respective average bleaching times and to build the histogram shown in Extended Data Fig. 5a,c,e,f.

To quantify the fluorescence intensity of labelled Smc5/6, those Smc5/6 complexes that did not perform loop extrusion were localized and separately categorized as ‘nonlooping Smc5/6’. The intensity trace of Smc5/6 was calculated by summing intensities over the square centred around the localized position of Smc5/6 in the kymograph. Smc5/6 molecules that bound and were stuck at the end of the DNA near the PEG surface were not considered for further analysis.

For estimation of the number of Smc5/6 during the high-salt wash, the intensity of a single label was estimated from those surviving at the end of the salt wash (Fig. 4j, inset). The number of Smc5/6 that survived the high-salt wash was further verified from the number of bleaching steps with the measurement at 100 mM NaCl post high-salt wash.

The MSD of nonlooping Smc5/6 molecules was calculated from traces of their respective positions determined using trackpy³⁶. Positions were tracked until individual Smc5/6 had reached either end of a DNA construct. The MSD (Fig. 3b) was fitted with a directed motion equation: ({rm{MSD}}(t)={v}^{2}{t}^{2}+4Dt), where v is mean velocity, D is the diffusion coefficient and t is lag time. The velocity (in μm s^–1) obtained from these fits was then converted to kbp s^–1 as (vleft(frac{kbp}{s}right)=vleft(frac{mu {rm{m}}}{{rm{s}}}right)times 48.5,{rm{k}}{rm{b}}{rm{p}}/{L}_{{rm{a}}{rm{v}}{rm{g}}}), where L_avg = 9 μm is the average end-to-end distance of those DNAs on which translocation was observed.

Langmuir–Hill plot

Loop-extrusion experiments were performed at different concentrations of WT Smc5/6. These measurements, each of 10 s duration, were recorded after an incubation period of 15 min. At this time point an equilibrium state is reached and the fraction of looped DNA remains almost constant. The fraction of DNA constructs that had formed loops ((f[L])) is determined and plotted as a function of Smc5/6 concentration (([L])) and fitted with the Hill–Langmuir function: (f([L])={left[Lright]}^{n}/left({left({K}_{{rm{a}}}right)}^{n}+{left[Lright]}^{n}right)), where K_a is the concentration of Smc5/6 at which half of the DNA is looped and n is the Hill coefficient. DNA molecules of end-to-end distance greater than 10 µm were not counted for this analysis, because they are unlikely to act as DNA substrates for loop extrusion due to the high tension/stall force on the stretched DNA (Fig. 1m and Extended Data Fig. 4b).

Probability of bleaching steps derived from dimer:monomer ratio

We determined the probability P(n) of observing either n = 0, 1 or 2 bleaching steps as a function of dimer fraction x with the following formula:

where p represents labelling efficiency. The respective errors were calculated using (sigma left(Pleft(nright)right)=pm frac{text{d}Pleft(nright)}{text{d}p}{sigma }_{p}), where σ_p is the error of labelling efficiency.

Quantification and statisitcal analysis

The fitting of curves in Figs. 2l and 3b and Extended Data Figs. 3b and 10b was done with the Scipy package in python (v.3.9)³⁵. Smoothing of data in Fig. 1h,j and Extended Data Fig. 3a,c–f was done with interpolation by the Savitzky–Golay method in Scipy using 50 data points. Error bars with 95% confidence interval in Figs. 1k, 2i–k, 3c,d and 4a,d,e and Extended Data Fig. 9b,d,e were calculated using the ‘binomial proportion confidence interval’. Box whisker plots in Figs. 1l,m and 4b,c contain the respective median values (horizontal white lines), with the box extending from Q1–Q3 quartile values of the data and bars extending no more than 1.5× IQR from the edges of the box. P values throughout the manuscript were calculated with a two-sided Student’s t-test unless otherwise stated.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.