Download PDF Article Open access Published: 24 June 2024 De novo DNA-based catch bonds Martijn van Galen  ORCID: orcid.org/0000-0002-5329-40871,2, Annemarie Bok  ORCID:

orcid.org/0000-0003-0665-89221, Taieesa Peshkovsky  ORCID: orcid.org/0009-0004-9724-60382, Jasper van der Gucht  ORCID: orcid.org/0000-0001-5525-83222, Bauke Albada  ORCID:

orcid.org/0000-0003-3659-24343 & …Joris Sprakel  ORCID: orcid.org/0000-0001-6532-45141 Show authors Nature Chemistry volume 16, pages 1943–1950 (2024)Cite this article

10k Accesses

All primary chemical interactions weaken under mechanical stress, which imposes fundamental mechanical limits on the materials constructed from them. Biological materials combine plasticity

with strength, for which nature has evolved a unique solution—catch bonds, supramolecular interactions that strengthen under tension. Biological catch bonds use force-gated conformational

switches to convert weak bonds into strong ones. So far, catch bonds remain exclusive to nature, leaving their potential as mechanoadaptive elements in synthetic systems untapped. Here we

report the design and realization of artificial catch bonds. Starting from a minimal set of thermodynamic design requirements, we created a molecular motif capable of catch bonding. It

consists of a DNA duplex featuring a cryptic domain that unfolds under tension to strengthen the interaction. We show that these catch bonds recreate force-enhanced rolling adhesion, a

hallmark feature of biological catch bonds in bacteria and leukocytes. This Article introduces catch bonds into the synthetic domain, and could lead to the creation of artificial

catch-bonded materials.

tethers for self-strengthening materials Article Open access 14 August 2023 Sequence-encoded intermolecular base pairing modulates fluidity in DNA and RNA condensates Article Open access 07

May 2025 Main

Mechanical stress accelerates the dissociation of all known primary chemical bonds, both covalent and supramolecular1,2,3,4. This so-called ‘slip bond’ nature of chemical interactions has

profound effects on the mechanical stability of molecules and their materials. In particular, bond strength and dissociation dynamics are intrinsically coupled—the stronger the bonds in a

material, the slower their exchange dynamics will be. As a result, for slip bonds, supramolecular dynamics and structural plasticity are at odds with toughness and mechanical strength.

In nature, cells continuously experience large mechanical stresses and deformations. Cellular materials must exhibit the strength and toughness to withstand these mechanics while

simultaneously remaining plastic and adaptive to accommodate the biological processes in which they are involved. Nature has evolved protein–protein and protein–ligand interactions with the

unique, counterintuitive feature of strengthening under tension5,6,7,8,9,10. Catch bonding often results from a force-gated conformational switch in a binding pocket, converting a weak

interaction into a mechanically stronger one as the tensile force on the bond increases9,11,12,13,14. Catch bonds are molecular equivalents of the fingertrap toy (Fig. 1a), which tightens

its grip under increasing tension—the harder one tries to escape, the more difficult it becomes12,15,16.

a, Schematic of a fingertrap toy that tightens under tension, forming a stronger interaction. b, Conceptual energy landscape of a two-state, two-pathway catch bond. The activation barriers

of transitions are indicated as Ex. At rest, catch bonds dissociate quickly from weak state 1. Tension 'tilts' the energy landscape, trapping the catch bond in strong state 2, which

dissociates more slowly. c, Mechanism of the de novo DNA catch bond. Weak state 1 dissociates quickly, via breaking of the crosslink. Tension opens the DNA hairpin in the hairpin strand,

forming an intermediate state. When the latch closes, the catch bond transitions to strong state 2, from which dissociation is slower. d, oxDNA simulation snapshots of a catch bond switching

from weak state 1 to strong state 2 under tension.

Despite being a recent discovery, first hypothesized in 198815 and confirmed experimentally in the early 2000s17,18, catch bonds are emerging as crucial design elements in nature’s

mechanical materials. Catch bonds have been identified in cell-adhesion complexes16,18,19,20,21, cytoskeletal crosslinkers5,8,9,22, protein–DNA bonds involved in cell division10,23 and

adhesive interfaces between plasma proteins and blood platelets during blood clotting24,25. Catch bonds play an important role in mechanotransduction26,27,28 and are proposed to allow

nature’s materials to combine structural plasticity and adaptivity in quiescent conditions, while being strong and tough under loading, when needed to protect cellular integrity9,29,30,31.

So far, catch bonds are only found in nature. Several theoretical models for the creation of artificial catch bonds have been proposed in recent years32,33,34,35. These rely on a variety of

mechanisms, including force-activated hinge-based switches33,34, self-interacting polymer chains inspired by the catch bond von Willebrand factor and load-sharing between tethered

nanoparticles35. Although these models provide interesting concepts, none has led to experimental realizations as yet. As a result, the promise of catch bonds as mechanoadaptive building

blocks for the creation of biomimetic materials that combine plasticity with mechanical strength remains untapped.

In this Article we report the design and realization of de novo catch bonds, constructed from engineered DNA duplexes. Based on a minimal thermodynamic model for catch bonds, we design DNA

duplexes that exhibit catch bonding. These duplexes contain a cryptic domain that is stable in the absence of force, but rapidly destabilizes under tension to expose additional base pairs

that strengthen the bond mechanically. To characterize these designs, we take inspiration from nature. A prototypical feature of cell-adhesion catch bonds, for example, in leukocytes, is

that they impart cells with the capacity to roll along tissue surfaces and become arrested at sites of high shear stresses36,37. We show, by conducting rolling adhesion experiments36,38,39

on microparticles decorated with our de novo catch bonds, that these designs can recreate this behaviour in a purely synthetic system. These experiments conclusively show how the catch bonds

enhance adhesive interactions under increasing load, while slip-bond control designs show the exact opposite.

Biological catch bonds, based on protein–protein or protein–ligand interactions, are incredibly complex, and the molecular mechanism underlying catch bonding has only been conclusively

evidenced for several examples12,13,14,21,40. To design a de novo catch bond, we started by abstracting the complexity of biological catch bonds to its bare essentials in the form of a set

of minimal thermodynamic design criteria.

We started from the established two-state/two-pathway model19,41,42, in which a bond can exist in two states: a thermodynamically favoured but mechanically weak state (1. Weak) and a

thermodynamically less favoured but mechanically strong state (2. Strong). Both states can reversibly convert into each other, or rupture, also reversibly, to an unbound state. This can be

schematically depicted as the energy landscape of a supramolecular reaction network (Fig. 1b). Within the Kramers–Bell picture of mechanically enhanced bond dissociation, tensile forces F on

the bonds reduce the activation barriers E for the transitions in this landscape, and thus the rates by which these barriers are crossed1. At rest, the weak state (1) is preferred. Under

tension, however, the energy landscape tilts by the mechanical work supplied to the bond, and this results in a conformational switch that accesses the strong state (2), which dissociates

more slowly.

To operate according to this mechanism, our catch bond should follow four thermodynamic design requirements. First, we require that state 2 is has greater mechanical strength than state 1,

which is the case when E2 > E1 (requirement I). The weak state can then either rupture, with barrier E1, or convert into the strong state, with barrier EIC. In the absence of tension, the

weak state should rupture and not convert to the strong state: that is, EIC > E1 at low tension (requirement II). By contrast, under tension, the opposite must happen—conversion to the

strong state happens more readily than bond rupture, such that EIC < E1 at high tension (requirement III). Finally, we require that the thermodynamically preferred state in the absence of

tension is the weak state, meaning that ΔE0 > 0 (requirement IV).

We used DNA duplexes to design a catch bond that meets these four requirements (Fig. 1c; Supplementary Table 1 presents the sequences). DNA interactions are ideal building blocks for

nanotechnological structures with tailored mechanical properties, as has been demonstrated in constructs such as DNA force sensors43,44 and artificial motor systems45,46. Double-stranded DNA

is remarkably elastic at low forces47,48 and offers good mechanical stability at larger forces due to π–π stacking interactions49,50. The large persistence length of the double helix allows

for the construction of rigid DNA-origami structures50,51,52. Moreover, the programmability of DNA provides a unique tunability in the interactions.

Our design consists of two DNA oligomers, a hairpin strand (HS, blue) and a complementary strand (CS, red). The hairpin strand contains three functional domains: the latch, toe-hold and

barrier domains. The latch domain is the cryptic binding site, which is complementary to a domain at the terminus of the complementary strand. In state 1, the hairpin strand and

complementary strand are bound by a short DNA duplex denoted as the crosslink (grey), which dissociates quickly. When tension is applied, the hairpin unfolds, yielding an intermediate state

in which the latch is exposed. On binding of the latch to its corresponding sequence on the complementary strand, the system interconverts to state 2. As this increases the number of base

pairs in the crosslink section, this state is more stable than state 1, meeting requirement I. In the absence of force, the crosslink is less stable than the hairpin due to the large number

of base pairs in the latter (19 base pairs) compared to the former (nine base pairs), ensuring that design requirement II is met. A mismatch of one base pair is present between the crosslink

and latch sequence to separate the domains.

Design requirement III is the most challenging to achieve. It requires that the interconversion EIC is more susceptible to tension than the dissociation barrier from state 1 (E1). We

accomplish this by using stress localization based on the mechanical geometry of loading. The crosslink (barrier E1) is loaded in a shear-type configuration, where the tension is distributed

evenly across all base pairs. By contrast, the hairpin (barrier EIC) is loaded in a zipper-type geometry; this localizes the tension on the first few base pairs at the start of the hairpin.

This stress localization makes DNA duplexes loaded in a zip geometry mechanically unstable as compared to identical duplexes loaded in the shear geometry. This difference in mechanical

stability between zip-loaded and shear-loaded DNA duplexes has been widely established in a range of studies38,53,54,55,56. It has been exploited previously to create DNA molecular motors

relying on asymmetric binding sites that respond differently depending on the direction of the applied force45. Our DNA catch bond differs from this mechanism, as it strengthens under

increasing force applied along a single direction. Furthermore, the difference between shear-type and zip-type unfolding pathways has been proposed previously as a mechanism for catch-bond

behaviour in natural proteins too57,58. Requirements II and III ensure that dissociation of the weak crosslink is more likely at low tension, and opening of the hairpin followed by

interconversion to the strong state is more likely at high tension. This force-dependent stability switching of the crosslink versus the hairpin lies at the core of our DNA catch-bond

design.

Finally, our design requires that in the absence of force, the construct is always fully in the weak state (design requirement IV). Through trial and error, we found that a short barrier

sequence (Fig. 1c, green) is required to achieve this. This barrier sequence is designed to be complementary to the base pairs within the hairpin, but non-complementary to the corresponding

poly(T) region on the complementary strand (detailed sketches are provided in Supplementary Figs. 9–20). This therefore increases the thermodynamic stability of weak state 1 relative to

strong state 2. Using Förster resonance energy transfer spectroscopy and computational modelling, we confirmed that a barrier sequence of at least five base pairs in length is essential to

ensure that requirement IV is fulfilled (Extended Data Fig. 1).

To verify our conceptual mechanism of catch-bond activation, we used oxDNA simulations59,60. These show that the design indeed transitions from weak state 1 to strong state 2 under tension,

via the anticipated reaction intermediate (Extended Data Fig. 2a and Supplementary Video 1). Moreover, these simulations revealed that catch-bond activation is reversible; catch bonds in

state 2 spontaneously return to state 1 following release of the mechanical tension by the formation of a toe hold and internal strand exchange (Extended Data Fig. 2b and Supplementary Video

2). We estimate the typical force required for catch-bond activation to be 12−18 pN, based on previously reported oxDNA simulations on zipper-loaded DNA duplexes61. We further quantified

the total energy Etotal over the course of our catch-bond activation simulations and found that state 1 is lower in energy than state 2 (Extended Data Fig. 3). This confirms that our catch

bond meets design requirement IV. We also found that the interconversion (IC) state has a larger Etotal than both state 1 and state 2, suggesting that a small energy barrier must be crossed

when returning from state 2 to state 1.

To confirm that tension changes the dissociation mechanism of the DNA catch bond, we performed an additional series of oxDNA simulations. Because the dissociation timescales are inaccessible

in our oxDNA simulations at room temperature, we carried out this series at an elevated temperature (40 °C) to accelerate the dissociation timescales (Extended Data Fig. 4). At low

tensional force, we often observe dissociation from the weak state of the catch bond. With increasing forces, we first observe a transition to the intermediate state, followed by a fraction

of bonds that stabilize in the strong state, revealing a change in dissociation mechanism. Further increases in tensional force result in rapid dissociation of the catch bond from the

intermediate state, before the latch is able to close.

To experimentally confirm the catch-bond behaviour of our DNA constructs, we made use of a rolling adhesion assay36,39. Cell-adhesion catch bonds, such as those on the surfaces of

leukocytes, enable these cells to stabilize their adhesive interfaces under shear as they roll along surfaces that display a catch-bonding ligand36,37,62. Rolling—the motion of a cell

without de-adhesion—occurs by the continuous exchange of a dissociating catch bond at the trailing edge with a newly formed one at the leading edge63,64. Rolling adhesion measurements of

microparticles bound to a surface by DNA slip bonds show how increased flow (that is, increased hydrodynamic forces on the particle) promotes supramolecular kinetics and thereby enhances

rolling. For catch bonds, in contrast, we expect the opposite—increased hydrodynamic forces stabilizing the adhesive contact.

In our rolling adhesion assay, we applied well-defined shear rates \({\dot{\gamma }}\) across microparticles bound to the surface of a microfluidics chip using the DNA molecule of interest,

placing the DNA bonds under tension (Fig. 2a). The tethered microparticles can respond in one of several ways. If the bonds in the interface reversibly break and form, we expect the

particles to roll forward, continuously breaking bonds at the rear and forming new ones in front. The rolling velocity is then dictated not by hydrodynamics, but by the supramolecular

reaction kinetics. If the interface is stable—for example, because catch bonds in the interface are activated—we expect the particles to become stationary (Fig. 2a). Indeed, in our

experiments we observe both (Fig. 2b). We confirmed that the observed interactions are caused by specific catch and slip bonds, rather than non-specific interactions, with a control

experiment in which microparticles were flowed over a flow cell surface functionalized with non-complementary filler DNA. In these measurements, particles moved rapidly across the surface at

\({\dot{\gamma }}\) values more than one order of magnitude lower than those required to observe rolling on surfaces containing slip or catch bonds (Extended Data Fig. 5). This proves that

any non-specific interactions present are much weaker than the specific slip and catch-bond interactions.

a, Schematic representation of a rolling adhesion experiment. Microparticles are bound to the surface of a microfluidic channel through catch bonds. A fluid flow through the channel imposes

a shear rate \({\dot{\gamma }}\) across the particles, which makes them roll forward, continuously breaking and forming bonds. Stabilization of the interface, through catch-bond activation

or an increase in bond density, causes particles to stop. b, Brightfield microscopy images showing a rolling and a stopped particle over time. c, Fraction of rolling particles froll as a

function of \({\dot{\gamma }}\) for the catch-bond candidate and the slip-bond control. For catch bonds, froll decreases with increasing \({\dot{\gamma }}\). d, Displacement trace of a

rolling particle, showing intermittent stop–roll behaviour, from which the stop time tstop and rolling velocity vroll are determined. e, A plot of the average vroll for particles bound with

slip and catch bonds shows that catch-bonded particles are much less sensitive to changes in \({\dot{\gamma }}\). f, A plot of the average tstop over \({\dot{\gamma }}\) indicates that

particles bound through catch bonds typically have larger stopping times. Shaded zones in e and f indicate standard deviations. The numbers of particle trajectories analysed per datapoint

are listed in Supplementary Tables 3c and 4e,f.

Through image analysis we can now compute, as a function of \({\dot{\gamma }}\), the fraction of all surface-localized beads that roll, froll, to quantify whether the bonds in the interface

weaken (slip) or strengthen (catch) under mechanical loading. We started with a control experiment consisting of a simple 7-nt slip-bonding DNA duplex. For this experiment, froll increases

with increasing shear rate (Fig. 2c and Supplementary Video 3). This is as expected—the increased tension weakens the slip bonds and thus enhances the supramolecular reaction kinetics and

thereby promotes rolling.

Strikingly, beads functionalized with our catch bonds show the opposite (Fig. 2c and Supplementary Video 4)—the larger the applied tension, the fewer particles roll and the more particles

become stuck to the chip. This is a clear demonstration that our DNA catch-bond design directly strengthens the interface under increasing shear forces. This increase in interfacial

stability under increasing shear stress is only possible if the bonds themselves gain mechanical strength under increasing tension. Hence, we can conclude that our DNA construct is the first

realization of an artificial catch bond.

The catch-bond nature of our de novo design is manifested also in other features of the same experiments. The displacement trajectories of the rolling particles show distinct intermittency,

with alternating periods of rolling and arrest (Fig. 2d). This behaviour is characteristic of rolling adhesion by supramolecular bonds and is the result of stochastic fluctuations in the

contact valency38,65. We observe this intermittency for both slip and catch bonds, but with distinct differences. For slip bonds, we observe a steep acceleration in motion with increasing

\({\dot{\gamma }}\) (Extended Data Fig. 6). By contrast, the rate of motion for catch-bonded particles appears almost insensitive to changes in \({\dot{\gamma }}\) (Extended Data Fig. 7). To

quantify these effects, we segmented the trajectories into rolling and stopping regimes to extract their rolling velocity, vroll, and stop times, tstop.

The rolling velocity of slip-bonded particles increases strongly with shear rate. This is as expected—tension accelerates bond rupture and thereby increases the rate by which the bead can

roll along the interface. Interestingly, the rolling velocity of the catch bonds is insensitive to changes in tension (Fig. 2e and Extended Data Fig. 8). This is another manifestation of

catch bonding, where the interconversion of catch bonds to the strong state under increased force, which strengthens the interface, compensates the acceleration of the particles under

increased shear rates, resulting in a velocity that is insensitive to shear. Besides the rolling velocity in the rolling regime, we also determined the overall average particle velocities,

taking into account both the rolling and stopping regimes, and found a similar insensitivity to shear for catch-bonded particles (Supplementary Fig. 5). Catch bonds here thus allow a

decoupling of the shear rate and the rate of particle motion. Also the average time particles remain stationary; the stopping time, tstop, is greater for catch bonds than for slip bonds

(Fig. 2f). We conjecture that the stochastic interconversion of a small number of catch bonds to the strong state 2 may result in a temporarily stabilized interface that results in longer

pauses.

We note that the distributions in vroll and tstop are broad, which can be caused by both stochastic fluctuations intrinsic to the system and experimental uncertainty. To assess the extent of

the intrinsic stochastic fluctuations, we compared the vroll distributions shown per rolling segment to those shown per particle trajectory, in which the intrinsic fluctuations are averaged

out (Supplementary Fig. 6). We found that averaging the per particle trajectory removes many of the high and low vroll outliers, confirming that much of the experimental uncertainty is

caused by intrinsic stochastic fluctuations. We also performed a series of control experiments to identify the causes of the remaining experimental uncertainty and found that variations in

bond density on both the channel and particle surfaces account for many of the fluctuations we observe (Supplementary Section Sources of experimental fluctuations).

The results from our rolling adhesion experiments suggest that catch bonds can enhance the stability of the rolling adhesive interface by decreasing the sensitivity of the particle–surface

interface to the shear rate. For a more mechanistic insight into this effect, we established a model for rolling adhesion based on steady-state kinetic Monte Carlo simulations (Fig. 3a).

These simulations allow us to predict the particle displacement trajectories, taking as input a molecular mechanotype, such as a slip-bond versus catch-bond mechanochemical interaction.

Catch bonds in the interface are simulated as being in either the weak or strong state. The shear field at the surface results in both a shear force (Fshear) and torque (M) acting on the

particles36. This displaces the particle, thereby stretching the bonds until mechanical equilibrium is reached. A stochastic algorithm then breaks, forms and interconverts the individual

bonds in the microparticle–surface interface, based on a force-dependent Kramers–Bell-type model for either faster (weak state and slip bond) or slower (strong state) dissociation under

tension. A detailed description of the simulation procedure and the assumptions made is provided in Supplementary Section Rolling adhesion simulation protocol. By simulating a rolling

microparticle in this way, we can reproduce many of the features in the experiments, but with additional insight into what happens at the molecular level.

simulations highlight the mechanism of interface reinforcement by catch bonds under shear.

a, Schematic of the simulation protocol followed by one kinetic Monte Carlo step in which a bond activates, forms or dissociates. By repeating these steps, we simulate a rolling particle

over time. b, Top–down heatmap at \({\dot{\gamma }}={560}\,{{{{\rm{s}}}}}^{-1}\), showing the spatial distribution of catch bonds in the weak and strong states relative to the centre of the

particle. The colour intensity indicates the average extensional force on the bonds. Catch bonds primarily activate in a small region at the trailing end of the particle, where tension is

highest. c, A simulated displacement trace of a catch-bonded particle shows intermittent stop–roll behaviour. d, A plot of the average rolling velocities, vroll, as a function of

\({\dot{\gamma }}\) reveals that catch bonds impart a lower sensitivity to the shear rate compared to slip bonds. e, Particle stopping times are higher, on average, for catch-bonded

particles over the range of \({\dot{\gamma }}\) evaluated. Shaded zones in d and e indicate standard deviations.

A heatmap that shows the distribution of catch bonds in the weak versus activated states (\({\dot{\gamma }}={560}\,{{{{\rm{s}}}}}^{-1}\)) reveals a clear localization of activated (strong)

catch bonds along a small ridge at the trailing edge of the particle, where tension on the interface stress is highest (Fig. 3b). This strong localization of catch-bond activation in zones

of high stress is also thought to contribute to their effectivity in promoting cohesive stability in networks9,29.

We performed simulations with both slip and catch bonds, and in both cases found intermittent stop–roll behaviour, similar to that observed in the experiments (Fig. 3c and Extended Data

Figs. 9 and 10). Similar to the experimental data analysis, we obtained the rolling velocity, vroll, and stopping time, tstop, as a function of shear rate (Fig. 3d,e). We found that these

trends reproduce the experimentally observed vroll and tstop data (Fig. 2d,e), using experimentally obtained mechanochemical parameters for the slip- and catch-bond DNA interactions as input

parameters. This further verifies that all of the features described above in the experiment are direct manifestations of catch bonding in our DNA designs. A complete list of the simulation

parameters and their justification is provided in Supplementary Section Rolling adhesion simulation protocol. At large shear rates, the simulated vroll increases exponentially for both the

slip-bond and catch-bond data (Fig. 3d). Although we do not observe this exponential increase in our experimental vroll data (Fig. 2d), we expect this should take place at larger shear

rates, which are inaccessible due to the practical limits in our flow control. The vroll histograms of the slip-bond simulations show similar broadening at increasing shear rates compared to

the experimental data. Simulated vroll distributions are furthermore narrower than the corresponding experimental distributions (Supplementary Fig. 7). We believe, here, that the terminal

regime of the catch bond is reached, where the strong state of the catch bond is overpowered by force, causing the catch bond to revert back to slip-bond behaviour.

Previous rolling adhesion studies on protein catch bonds have identified a regime in which the rolling velocity decreases with increasing shear rate36. To determine whether this regime is

accessible in our simulations, we performed an additional simulation series in which we systematically explored the parameter space in terms of the bond dissociation rates from the weak and

strong states, and the mechanical susceptibility towards catch-bond activation (Supplementary Fig. 8). Despite substantially varying these parameters and explicitly including catch-bond

behaviour, we do not observe a regime of decreasing rolling velocity. We therefore conclude that the decrease in vroll is not a feature of our synthetic DNA catch bond within the regimes we

can access.

Theoretical predictions and experimental evidence on biological catch bonds15,66 suggest that, between slip bonds and catch bonds, a regime exists where ‘ideal’ bonds can be found. Ideal

bonds neither weaken nor strengthen under tension, but their stability is insensitive to the applied force. In our catch-bond design, the 7-nt latch sequence strengthens the bond under

tension. To explore the regime between slip and catch bonds in our design, we systematically varied the length of the latch sequence Nlatch, expecting to gradually weaken the strong state.

We found that latch lengths of 2–5 nt result in a rolling velocity that is invariant to the applied shear stress, whereas for slip bonds it increases substantially (Fig. 4a). Clearly, all

designs with a latch of 2 nt or more do not behave as conventional slip bonds. However, these designs did not show the hallmark feature of catch bonds, which is a decreasing fraction of

mobile particles, froll, with increasing shear rate. Rather, within the experimental noise, froll appears constant or increases weakly with imposed hydrodynamic force. This behaviour, in

between catch and slip bonds, suggests ideal bonding. This illustrates how, in a single design, we can gradually tailor the molecular mechanics from slip to catch bonds, through an

intermediate regime that is reminiscent of ideal bonding. At the moment, we do not fully understand the exact origin of the role of this latch length in the switch in slip–ideal–catch

bonding behaviour. Nevertheless, it illustrates the narrow phase space in which true catch bonding is observed, which can be disrupted with a minimal design change.

latch sequence length reveal the emergence of catch-bond behaviour at high Nlatch.

a, Plots of the average vroll as a function of \({\dot{\gamma }}\) over a range of NLatch sequence lengths (green) in comparison to the slip bond (grey). Contrary to the slip-bond control,

vroll is insensitive to \({\dot{\gamma }}\) for each latch sequence variation. This indicates a deviation from slip-bond behaviour similar to that observed for the catch-bond design for

Nlatch = 7. b, Plots of froll as a function of \({\dot{\gamma }}\) over a range of Nlatch sequence lengths (blue) and the slip bond (grey). For Nlatch between 2 and 5, froll increases

slightly or remains constant. By contrast, a clear decreasing trend indicates catch bonding is only observed for Nlatch = 7. Filled zones in a indicate standard deviations. Measurements for

Nlatch = 7 at \({\dot{\gamma }} > {667}\,{{{{\rm{s}}}}}^{-1}\) contained fewer than 30 particles and were hence excluded due to a lack of statistics. The numbers of particle trajectories

analysed per datapoint are listed in Supplementary Tables 5b and 6a.

We have reported on the design and realization of an artificial de novo catch bond built using DNA duplexes. Despite several theoretical efforts to facilitate rational design in the

past32,33, an experimental realization had not yet been achieved. Starting from the conceptual framework of the two-state, two-pathway catch bonds found in nature19,41,42, and its

abstraction into four tangible design requirements, we have proposed a DNA-based construct capable of tension-induced and reversible switching between a thermodynamically stabilized weak

state and a mechanically stabilized strong state. Moreover, we have shown how collectives of these catch bonds, in direct analogy to their natural counterparts during leukocyte rolling36,37,

can promote the adhesive strength of an interface under tension, whereas slip bonds weaken as they experience increasing forces. This Article introduces catch bonds into the synthetic

domain, opening up a plethora of opportunities to create DNA-origami-based nanomachines and other man-made catch-bonded materials. These are theoretically predicted to result in fascinating

features, including the unification of structural plasticity with mechanical resilience9, improved fracture resistance and toughness29,31 and mechano-adaptivity67.

All chemicals were purchased from Sigma Aldrich, unless stated otherwise.

DNA oligonucleotides were purchased from Integrated DNA Technologies, and their sequences are reported in Supplementary Table 1. In the rolling adhesion experiments, we utilized catch-bond

(CB-HS and CB-CS) and slip-bond (SB-1 and SB-2) sequences that each contained a 5′ biotin group for attachment to streptavidin, followed by a 40-nt spacer sequence to repress non-specific

interactions and finally the target bond sequence. A filler sequence (filler) containing a 5′ biotin, the 40-nt spacer and a T5 end was mixed together with the catch- or slip-bond sequences

to tune the density of available bonds on the microfluidic channel surface. In the oxDNA simulations, we simulated only the target bond sequence, without the 40-nt spacer sequence, for

computational efficiency.

The following buffer solutions were used during sample preparation:

A: 10 mM Tris, 1 mM EDTA, 2 mM MgCl2, pH 6

B: buffer A, containing 0.1 mg ml−1 BSA, 0.2% wt/vol Pluronic F108, pH 6

C: buffer B, containing 100 mM NaCl, 0.03 mg ml−1 biotin, pH 6

First, DNA oligos (CB-HS, CB-CS, SB-1 or SB-2) were hybridized to the spacer-complement sequence by combining 10 μM target and 10 μM spacer-complement in buffer A. The mixtures were heated

for 5 min at 95 °C and gradually cooled to room temperature over 1.5 h. Hybridized oligos were kept at 4 °C and used within 24 h for functionalization.

All surface functionalization steps and rolling adhesion experiments were performed at 20 °C. Streptavidin-coated polystyrene microparticles (1.36-μm diameter, Spherotech Inc.) were first

transferred to buffer B as follows: 50 μl of a 0.5% wt/vol particle suspension was centrifuged for 6 min at 12,000g, the supernatant was removed with a pipette and the particles were

resuspended in 1 ml of buffer B. This washing–resuspension cycle was repeated three times. During the final cycle, particles were resuspended in 80 μl of hybridized slip- or catch-bond

oligos (CB-CS or SB-2, or latch variations CB-CS-L2 to CB-CS-L5; Supplementary Table 1) and incubated at room temperature for 5 min. Finally, functionalized particles were washed by three

more centrifugation–resuspension cycles using 1 ml of buffer B. On the final cycle, particles were resuspended in 500 μl of buffer C, which contained free biotin to irreversibly block any

remaining unoccupied streptavidin sites and prevent non-specific interactions arising from these. Finally, the particle suspension was diluted to the final concentration by adding 30 μl of

the particle suspension to 470 μl of buffer C. This suspension was stored at 4 °C until further use within two days.

Rectangular microfluidic channels (1,500 μm × 50 μm × 40 mm, Micronit Microtechnologies BV, cat. no. FC_FLC50.3) were functionalized with the hybridized target DNA (CB-HS or SB-1) using a

sequential functionalization with biotinylated BSA, streptavidin and finally DNA, as follows. First, the channels were mounted in a chip holder (Micronit Fluidic Connect Pro) and connected

with polypropylene tubing (1/32-inch inner diameter; 1/16-inch outer diameter). Before functionalization, the chip was filled with 1 M KOH for 5 min to etch the channel surface and

thoroughly flushed with 4 ml of Milli-Q water until neutralized. The chip holder was then mounted onto a Nikon Ti2 Eclipse brightfield microscope and connected to an Elveflow Sequential

Fluid injection set-up (Elvesys) composed of a pressure controller (OB1 Mk3+), a flow rate sensor (MFS3-80) and a 12-way distribution valve to sequentially inject different solutions (MUX).

All of the following functionalization steps were performed with the chip mounted in this set-up, immediately before the rolling adhesion experiment. First, 0.5 ml of a biotinylated BSA

(Pierce) solution (1 mg ml−1) in buffer A was introduced into the channel, followed by a 5-min incubation at a flow rate of 0 μl min−1. Buffer A is slightly acidic (pH 6) to facilitate

better absorption of the BSA to the glass surface68. Next, the channel was flushed with 5 ml of buffer B to remove any remaining unbound biotinylated BSA. The channel was then functionalized

with streptavidin (ThermoFisher) by flowing 0.5 ml of a streptavidin solution (0.02 mg ml−1) in buffer A through the channel at 50 μl min−1. After another flushing step with 5 ml of buffer

B, a mixture of 0.15 ml of 10 μM hybridized DNA (CB-HS or SB-1) and 0.15 ml of 10 μM hybridized filler DNA and 0.6 ml buffer B was introduced to the channel, followed by 5 min of incubation

at 0 μl min−1. This 1:1 ratio of target DNA and filler DNA was found to be optimal for achieving stable rolling behaviour. Finally, the remaining excess of DNA was removed by flowing 5 ml of

buffer B.

After functionalizing the surface of the microfluidic chips with the hairpin strand of the catch bond, and attaching the complementary strand to the microparticles, we performed our rolling

adhesion assay as follows. Within 1 h after microfluidic channel functionalization, a batch of functionalized microparticles was introduced into the functionalized microfluidic channel. The

flow was switched to 0 μl min−1 for 5 min to allow the particles to settle to the bottom of the flow cell. Sedimented particles in the centre of the channel were brought into focus under a

brightfield microscope, using a ×40 objective (Nikon Plan apo lambda) and a Fastec HiSpec 2G Mono camera. We then used the Elveflow microfluidic flow controller to apply well-defined shear

rates to the particles. To begin a measurement, the microfluidic flow rate was set to the flow rate of interest and allowed to stabilize, after which a movie of 1,634 frames was recorded at

60 f.p.s., using a resolution of 1,280 × 1,024 pixels. In a typical measurement, we imaged between 100 and 400 particles in the field of view at a time. Measurements were repeated for a

range of flow rates, Q. The imposed flow rate was converted to the wall shear rate \({\dot{\gamma }}\) assuming Poisseuille flow:

where h denotes the height of the channel and w the width of the channel. Equation (1) is a valid approximation for channels containing a Newtonian fluid with rectangular shape and w ≫ h.

The recorded movies were analysed using a custom-built Python particle-tracking script that employs the TrackPy package69, which has been made publicly available at

https://github.com/jorissprakel/DNACatchBond. Before particle tracking, the raw movies were pre-processed to convert the brightfield spots into trackable Gaussian features through a series

of steps. First, an intensity threshold step was used, setting all values below a threshold intensity to 0. This threshold step removed the out-of-focus particles flowing past and the dark

rings surrounding each particle spot, leaving only the bright spots at the centres of the particles. This resulted in bright, truncated features. The threshold value was determined manually

for each dataset to ensure that it was higher than the background intensity of out-of-focus particles, but lower than the maximum intensity of the in-focus particles. Threshold values of

between 25 and 45 were used. Next, a Gaussian blur (width 1 pixel) was used to transform the truncated features into Gaussian spots, which were then tracked by the TrackPy locate function.

Particles that displaced more than 1 μm over the course of their trajectories were considered rolling and included in the fraction of rolling particles froll. Only those particles tracked

for at least 30 frames were considered in the analysis, and datasets containing fewer than 30 traced particles were excluded from the analysis owing to a lack of statistics. Trajectories of

the rolling particles were separated into stopping and rolling regimes using an instantaneous velocity threshold of 0.5 μm s−1, which was empirically found to provide good separation between

the regimes. A Savitzky–Golay filter was used to smooth noise in the instantaneous velocity profile and allow for a more precise determination of the regimes. Finally, the average stop

time, tstop, was computed from the stopping regime duration. The average rolling velocity, vroll, was computed per particle trajectory by computing the average velocity over all rolling

regimes experienced by the particle.

Rolling adhesion simulations were performed with a custom-developed Python program. The simulation and data-analysis code has been made publicly available at

https://github.com/jorissprakel/DNACatchBond. A thorough description of the simulation protocol and motivation for the parameters is provided in Supplementary Section Rolling adhesion

simulation protocol. Our simulations made use of a Gillespie kinetic Monte Carlo algorithm to determine which step took place at a given moment (bond dissociation, bond formation or bond

activation). Single-particle simulations were performed on particles bound via slip bonds and particles bound via catch bonds, over a range in shear rates \({\dot{\gamma }}\). Each

simulation was run for a total of 5 × 105 simulation steps and was repeated 50 times. As the Kinetic Monte Carlo simulations are event-driven, the displacement data are inhomogeneously

distributed in time. For this reason, displacement data were linearly interpolated over time, using an interpolation time step of 0.01. The interpolated displacement traces were split into

rolling and stopped regimes using an instantaneous velocity cutoff of 0.5, which was empirically found to provide good separation between the regimes. A Savitzky–Golay filter was again used

to smooth noise in the instantaneous velocity and allow for a more precise determination of the regimes. From these separated regimes, the average stopping times of the stopping regimes were

determined. Finally, the average rolling velocity was computed from the rolling regimes, weighted following the duration of the regimes. To generate the catch-bond activation heatmap at

high shear, a separate set of 100 simulations were performed for a total of 5 × 105 simulation steps at \({\dot{\gamma }}={560}\,{{{{\rm{s}}}}}^{-1}\), during which the position, state and

extensional force on each linker were stored every 2,500 steps. The catch-bond activation heatmap (Fig. 3b) was then generated as follows. First, a two-dimensional array of bins was defined

on the channel surface, relative to the centre of the particle. Within each bin, the average extensional force on catch bonds in the weak and strong states was determined over the course of

the simulation. Averaging the heatmaps further over the 100 simulated repeats resulted in the activation heatmap in Fig. 3b.

The raw data associated with the figures in this manuscript are publicly available at https://doi.org/10.4121/96e43d14-80a6-46e2-819c-9c627cedf10e.v2.

All simulation and data analysis codes used to perform and analyse the work in this article have been made publicly available at https://github.com/jorissprakel/DNACatchBond/.

References Bell, G. Models for specific adhesion of cells to cells. Science 200, 618–627 (1978).

Article  CAS  PubMed  Google Scholar 

Evans, E. & Ritchie, K. Dynamic strength of molecular adhesion bonds. Biophys. J. 72, 1541–1555 (1997).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Grandbois, M., Beyer, M., Rief, M., Clausen-Schaumann, H. & Gaub, H. E. How strong is a covalent bond? Science 283, 1727–1730 (1999).

Article  CAS  PubMed  Google Scholar 

Beyer, M. K. & Clausen-Schaumann, H. Mechanochemistry: the mechanical activation of covalent bonds. Chem. Rev. 105, 2921–2948 (2005).

Article  CAS  PubMed  Google Scholar 

Buckley, C. D. et al. The minimal cadherin-catenin complex binds to actin filaments under force. Science 346, 1254211 (2014).

Article  PubMed  PubMed Central  Google Scholar 

Kong, F. et al. Cyclic mechanical reinforcement of integrin-ligand interactions. Mol. Cell 49, 1060–1068 (2013).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Boettiger, D. Mechanical control of integrin-mediated adhesion and signaling. Curr. Opin. Cell Biol. 24, 592–599 (2012).

Article  CAS  PubMed  Google Scholar 

Huang, D. L., Bax, N. A., Buckley, C. D., Weis, W. I. & Dunn, A. R. Vinculin forms a directionally asymmetric catch bond with F-actin. Science 357, 703–706 (2017).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Mulla, Y. et al. Weak catch bonds make strong networks. Nat. Mater. 21, 1019–1023 (2022).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Sarangapani, K. K. & Asbury, C. L. Catch and release: how do kinetochores hook the right microtubules during mitosis? Trends Genet. 30, 150–159 (2014).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Sauer, M. M. et al. Catch-bond mechanism of the bacterial adhesin FimH. Nat. Commun. 7, 10738 (2016).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Le Trong, I. et al. Structural basis for mechanical force regulation of the adhesin FimH via finger trap-like β sheet twisting. Cell 141, 645–655 (2010).

Article  PubMed  PubMed Central  Google Scholar 

Manibog, K., Li, H., Rakshit, S. & Sivasankar, S. Resolving the molecular mechanism of cadherin catch bond formation. Nat. Commun. 5, 3941 (2014).

Article  CAS  PubMed  Google Scholar 

Wang, A., Dunn, A. R. & Weis, W. I. Mechanism of the cadherin-catenin F-actin catch bond interaction. ELife 11, e80130 (2022).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Dembo, M., Torney, D., Saxman, K. & Hammer, D. The reaction-limited kinetics of membrane-to-surface adhesion and detachment. Proc. R. Soc. B 234, 55–83 (1988).

CAS  Google Scholar 

Thomas, W. E., Vogel, V. & Sokurenko, E. Biophysics of catch bonds. Annu. Rev. Biophys. 37, 399–416 (2008).

Article  CAS  PubMed  Google Scholar 

Thomas, W. E., Trintchina, E., Forero, M., Vogel, V. & Sokurenko, E. V. Bacterial adhesion to target cells enhanced by shear force. Cell 109, 913–923 (2002).

Article  CAS  PubMed  Google Scholar 

Marshall, B. T. et al. Direct observation of catch bonds involving cell-adhesion molecules. Nature 423, 190–193 (2003).

Article  CAS  PubMed  Google Scholar 

Thomas, W. et al. Catch-bond model derived from allostery explains force-activated bacterial adhesion. Biophys. J. 90, 753–764 (2006).

Article  CAS  PubMed  Google Scholar 

Sundd, P., Pospieszalska, M. K., Cheung, L. S.-L., Konstantopoulos, K. & Ley, K. Biomechanics of leukocyte rolling. Biorheology 48, 1–35 (2011).

PubMed  PubMed Central  Google Scholar 

Liu, Z. et al. High force catch bond mechanism of bacterial adhesion in the human gut. Nat. Commun. 11, 4321 (2020).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Owen, L. M., Bax, N. A., Weis, W. I. & Dunn, A. R. The C-terminal actin-binding domain of talin forms an asymmetric catch bond with F-actin. Proc. Natl Acad. Sci. USA 119, e2109329119

(2022).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Akiyoshi, B. et al. Tension directly stabilizes reconstituted kinetochore-microtubule attachments. Nature 468, 576–579 (2010).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Yago, T. et al. Platelet glycoprotein Ibα forms catch bonds with human WT vWF but not with type 2B von Willebrand disease vWF. J. Clin. Invest. 118, 3195–3207 (2008).

CAS  PubMed  PubMed Central  Google Scholar 

Interlandi, G. & Thomas, W. The catch bond mechanism between von Willebrand factor and platelet surface receptors investigated by molecular dynamics simulations. Proteins 78, 2506–2522

(2010).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Swaminathan, V., Alushin, G. M. & Waterman, C. M. Mechanosensation: a catch bond that only hooks one way. Curr. Biol. 27, R1158–R1160 (2017).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Hoffman, B. D., Grashoff, C. & Schwartz, M. A. Dynamic molecular processes mediate cellular mechanotransduction. Nature 475, 316–323 (2011).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Liu, B., Kolawole, E. M. & Evavold, B. D. Mechanobiology of T-cell activation: to catch a bond. Annu. Rev. Cell Dev. Biol. 37, 65–87 (2021).

Article  PubMed  Google Scholar 

Ruiz-Franco, J., Tauber, J. & Van Der Gucht, J. Cross-linker mobility governs fracture behavior of catch-bonded networks. Phys. Rev. Lett. 130, 118203 (2023).

Article  CAS  PubMed  Google Scholar 

Mbanga, B. L., Iyer, B. V., Yashin, V. V. & Balazs, A. C. Tuning the mechanical properties of polymer-grafted nanoparticle networks through the use of biomimetic catch bonds. Macromolecules

49, 1353–1361 (2016).

Article  CAS  Google Scholar 

Iyer, B. V., Yashin, V. V. & Balazs, A. C. Harnessing biomimetic catch bonds to create mechanically robust nanoparticle networks. Polymer 69, 310–320 (2015).

Article  CAS  Google Scholar 

van Galen, M., van der Gucht, J. & Sprakel, J. Chemical design model for emergent synthetic catch bonds. Front. Phys. 8, 361 (2020).

Article  Google Scholar 

Dansuk, K. C. & Keten, S. A simple mechanical model for synthetic catch bonds. Matter 1, 911–925 (2019).

Article  Google Scholar 

Dansuk, K. C. & Keten, S. Self-strengthening biphasic nanoparticle assemblies with intrinsic catch bonds. Nat. Commun. 12, 85 (2021).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Dansuk, K. C., Pal, S. & Keten, S. A catch bond mechanism with looped adhesive tethers for self-strengthening materials. Commun. Mater. 4, 60 (2023).

Article  CAS  Google Scholar 

Yago, T. et al. Catch bonds govern adhesion through L-selectin at threshold shear. J. Cell Biol. 166, 913–923 (2004).

Article  CAS  PubMed  PubMed Central  Google Scholar 

McEver, R. P. & Zhu, C. Rolling cell adhesion. Annu. Rev. Cell Dev. Biol. 26, 363–396 (2010).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Porter, C. L., Diamond, S. L., Sinno, T. & Crocker, J. C. Shear-driven rolling of DNA-adhesive microspheres. Biophys. J. 120, 2102–2111 (2021).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Lou, J. et al. Flow-enhanced adhesion regulated by a selectin interdomain hinge. J. Cell Biol. 174, 1107–1117 (2006).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Xu, X.-P. et al. Structural basis of αE-catenin–F-actin catch bond behavior. ELife 9, e60878 (2020).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Evans, E., Leung, A., Heinrich, V. & Zhu, C. Mechanical switching and coupling between two dissociation pathways in a P-selectin adhesion bond. Proc. Natl Acad. Sci. USA 101, 11281–11286

(2004).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Prezhdo, O. V. & Pereverzev, Y. V. Theoretical aspects of the biological catch bond. Acc. Chem. Res. 42, 693–703 (2009).

Article  CAS  PubMed  Google Scholar 

Zhang, Y., Ge, C., Zhu, C. & Salaita, K. DNA-based digital tension probes reveal integrin forces during early cell adhesion. Nat. Commun. 5, 5167 (2014).

Article  CAS  PubMed  Google Scholar 

Liu, Y. et al. DNA-based nanoparticle tension sensors reveal that T-cell receptors transmit defined pN forces to their antigens for enhanced fidelity. Proc. Natl Acad. Sci. USA 113,

5610–5615 (2016).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Loh, I. Y., Cheng, J., Tee, S. R., Efremov, A. & Wang, Z. From bistate molecular switches to self-directed track-walking nanomotors. ACS Nano 8, 10293–10304 (2014).

Article  CAS  PubMed  Google Scholar 

Chen, Y., Wang, M. & Mao, C. An autonomous DNA nanomotor powered by a DNA enzyme. Angew. Chem. Int. Ed. 116, 3638–3641 (2004).

Article  Google Scholar 

Bustamante, C., Marko, J. F., Siggia, E. D. & Smith, S. Entropic elasticity of λ-phage DNA. Science 265, 1599–1600 (1994).

Article  CAS  PubMed  Google Scholar 

Bustamante, C., Bryant, Z. & Smith, S. B. Ten years of tension: single-molecule DNA mechanics. Nature 421, 423–427 (2003).

Article  PubMed  Google Scholar 

Kilchherr, F. et al. Single-molecule dissection of stacking forces in DNA. Science 353, aaf5508 (2016).

Article  PubMed  Google Scholar 

Ji, J., Karna, D. & Mao, H. DNA origami nano-mechanics. Chem. Soc. Rev. 50, 11966–11978 (2021).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Saccà, B. & Niemeyer, C. M. DNA origami: the art of folding DNA. Angew. Chem. Int. Ed. 51, 58–66 (2012).

Article  Google Scholar 

Pfitzner, E. et al. Rigid DNA beams for high-resolution single-molecule mechanics. Angew. Chem. Int. Ed. 52, 7766–7771 (2013).

Article  CAS  Google Scholar 

Albrecht, C. et al. DNA: a programmable force sensor. Science 301, 367–370 (2003).

Article  CAS  PubMed  Google Scholar 

Kufer, S., Puchner, E., Gumpp, H., Liedl, T. & Gaub, H. Single-molecule cut-and-paste surface assembly. Science 319, 594–596 (2008).

Article  CAS  PubMed  Google Scholar 

Wang, X. & Ha, T. Defining single molecular forces required to activate integrin and notch signaling. Science 340, 991–994 (2013).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Merindol, R., Delechiave, G., Heinen, L., Catalani, L. H. & Walther, A. Modular design of programmable mechanofluorescent DNA hydrogels. Nat. Commun. 10, 528 (2019).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Guo, S. et al. Structural-elastic determination of the force-dependent transition rate of biomolecules. Chem. Sci. 9, 5871–5882 (2018).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Huang, W. et al. Mechanical stabilization of a bacterial adhesion complex. J. Am. Chem. Soc. 144, 16808–16818 (2022).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Šulc, P. et al. Sequence-dependent thermodynamics of a coarse-grained DNA model. J. Chem. Phys. 137, 135101 (2012).

Article  PubMed  Google Scholar 

Bohlin, J. et al. Design and simulation of DNA, RNA and hybrid protein–nucleic acid nanostructures with oxView. Nat. Protoc. 17, 1762–1788 (2022).

Article  CAS  PubMed  Google Scholar 

Mosayebi, M., Louis, A. A., Doye, J. P. & Ouldridge, T. E. Force-induced rupture of a DNA duplex: from fundamentals to force sensors. ACS Nano 9, 11993–12003 (2015).

Article  CAS  PubMed  Google Scholar 

Yago, T., Zarnitsyna, V. I., Klopocki, A. G., McEver, R. P. & Zhu, C. Transport governs flow-enhanced cell tethering through L-selectin at threshold shear. Biophys. J. 92, 330–342 (2007).

Article  CAS  PubMed  Google Scholar 

Caputo, K. E., Lee, D., King, M. R. & Hammer, D. A. Adhesive dynamics simulations of the shear threshold effect for leukocytes. Biophys. J. 92, 787–797 (2007).

Article  CAS  PubMed  Google Scholar 

Hammer, D. A. Adhesive dynamics. J. Biomech. Eng. 136, 021006 (2014).

Article  PubMed  Google Scholar 

Hammer, D. A. & Apte, S. M. Simulation of cell rolling and adhesion on surfaces in shear flow: general results and analysis of selectin-mediated neutrophil adhesion. Biophy. J. 63, 35–57

(1992).

Article  CAS  Google Scholar 

Rakshit, S., Zhang, Y., Manibog, K., Shafraz, O. & Sivasankar, S. Ideal, catch and slip bonds in cadherin adhesion. Proc. Natl Acad. Sci. USA 109, 18815–18820 (2012).

Article  CAS  PubMed  PubMed Central  Google Scholar 

Tabatabai, A. P. et al. Detailed balance broken by catch bond kinetics enables mechanical-adaptation in active materials. Adv. Funct. Mater. 31, 2006745 (2021).

Article  CAS  PubMed  Google Scholar 

Kubiak-Ossowska, K., Tokarczyk, K., Jachimska, B. & Mulheran, P. A. Bovine serum albumin adsorption at a silica surface explored by simulation and experiment. J. Phys. Chem. B 121, 3975–3986

(2017).

Article  CAS  PubMed  Google Scholar 

Allan, D. B., Caswell, T., Keim, N. C., van der Wel, C. M. & Verweij, R. W. soft-matter/trackpy: Trackpy v.0.5.0 https://doi.org/10.5281/zenodo.4682814 (2021)

Download references

The work of M.v.G. is supported financially by the VLAG Graduate School. J.S., A.B. and M.v.G. are funded by the European Research Council (ERC), project CoG-CATCH. We thank L. Michels for

the design of Fig. 1a.

Author informationAuthors and Affiliations Laboratory of Biochemistry, Wageningen University & Research, Wageningen, Netherlands

Martijn van Galen, Annemarie Bok & Joris Sprakel

Physical Chemistry and Soft Matter, Wageningen University & Research, Wageningen, Netherlands

Martijn van Galen, Taieesa Peshkovsky & Jasper van der Gucht

Laboratory of Organic Chemistry, Wageningen University & Research, Wageningen, Netherlands

Bauke Albada

AuthorsMartijn van GalenView author publications You can also search for this author inPubMed Google Scholar

Annemarie BokView author publications You can also search for this author inPubMed Google Scholar

Taieesa PeshkovskyView author publications You can also search for this author inPubMed Google Scholar

Jasper van der GuchtView author publications You can also search for this author inPubMed Google Scholar

Bauke AlbadaView author publications You can also search for this author inPubMed Google Scholar

Joris SprakelView author publications You can also search for this author inPubMed Google Scholar

J.S. conceived of the project. M.v.G., J.S., B.A. and A.B. designed the DNA catch bond. M.v.G. and A.B. performed and analysed the oxDNA simulations. A.B. performed and analysed the

fluorescence resonance energy transfer experiments. T.P., M.v.G. and A.B. performed and analysed the rolling adhesion experiments. J.v.d.G. designed the rolling adhesion simulation protocol,

and M.v.G. performed and analysed the rolling adhesion simulations. All authors discussed the data and their interpretation and edited the paper. The funders had no role in study design,

data collection and analysis, decision to publish or preparation of the paper.

Corresponding authors Correspondence to Bauke Albada or Joris Sprakel.

The authors declare no competing interests.

Nature Chemistry thanks Zhisong Wang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

measurements reveal that the catch bond construct assembles in the weak state 1.

a: Catch bond candidates with varying latch sequence length NLatch and barrier sequence length NBarrier (sequences: CB-CS and CB-HS) were hybridized to Alexa Fluoro labeled DNA oligos

(A488-spacer and A565-spacer). The distance between the FRET-dyes is longer in strong state 2, compared to weak state 1. This results in a lower FRET efficiency E, allowing us to distinguish

the two states. b: A comparison between experimentally obtained FRET efficiencies (triangles, in duplo) and predicted FRET efficiencies using oxDNA simulations in the weak and strong states

(circles) reveals that weak state 1 is the preferred state for each value of Nlatch, including the main catch bond design (NLatch = 7). c: A similar comparison for FRET measurements at

varying NBarrier reveals that a minimum of 5 barrier nt is necessary to ensure that weak state 1 is preferred over state 2. The main catch bond design has NBarrier = 6. Experimental details

are provided in SI section FRET experiments on the de novo catch bond.

and strong state 2 under tension (a) and at rest (b).

We track the state of the catch bond using the fraction of base-pairs formed in the toe-hold region (ftoe−hold) and those formed between the complementary strand and hairpin strand in the

latch region flatch. Simulation snapshots are shown as insets. a: When weak state 1 is placed under tension, the hairpin opens, followed by a closing of the latch to yield strong state 2. b:

At rest, strong state 2 relaxes to weak state 1 via an initial formation of the toe-hold, which displaces the latch on the complementary strand. Simulation details are provided in the SI

section Coarse-Grained molecular dynamics simulations.

a: We separated the simulation in states 1, intermediate and 2 using the simulation snapshots. b: Time-evolution of Etotal over the course of the simulation. Starting from state 1.weak, the

hairpin first unfolds reversibly at t = 0.5 ⋯ 108 steps, causing a spike in Etotal. Between t = 1.5 ⋯ 108 and t = 2 ⋯ 108 steps, the hairpin fully unfolds to the intermediate state (IC),

which increases Etotal. Finally, a closing of the latch sequence at t = 6 ⋯ 108 steps again lowers Etotal. c: Etotal distributions recorded in each of the three states reveal that an energy

barrier must be crossed when transitioning from state 1 via the intermediate IC to state 2. Moreover, Etotal is lower for state 1 than for state 2, confirming that weak state 1 is the

thermodynamically favored state. Here, the Etotal distribution of state 1 represents Etotal during the first 0.25 ⋯ 108 simulation steps, prior to the reversible hairpin unfolding event at t

 = 0.5 ⋯ 108. Violin plots in c are composed of N=59 (weak), N=1036 (IC), and N=512 (Strong) simulation snapshots. Error bars in c represent standard deviations. Simulation details are

provided in the SI section Coarse-Grained molecular dynamics simulations.

increasing tension.

a: Overview of the number of simulations ending up in each catch bond state at varying tensional forces F. States include 1. weak, interconversion (IC), 2. strong and dissociation from these

states. b. Simulation snapshots of each state included in a. At F = 15 pN the bonds are either stable in 1. weak or dissociate from 1. weak. At increasing force, the bond interconverts and

in a few simulations stabilizes in state 2. strong. Larger forces result in rapid dissociation from the intermediate state before the catch bond has a chance to activate. simulations were

performed at 40°C for 1.6 ⋯ 109 simulation steps, with stepsize Δt = 0.005. Each simulation was repeated 20 times.

functionalised with 100% of the non-specific Filler DNA, at γ⋅= 16 s−1.

a: Microparticles move rapidly across the flow cell at a shear rates as low as 16 s−1, more than an order of magnitude lower than the shear rates required to achieve rolling on the catch

bonded or slip bonded interfaces (>200 s−1), confirming that any non-specific interactions in our rolling adhesion assay are much weaker than the specific slip bond and catch bond

interactions. Hence, non-specific interactions do not play a substantial role in the rolling behaviour we observe for the slip and catch bonded interfaces. b: Examples of particle

displacement traces recorded during these control experiments. Scale bars denote 10 μm.

Overview of a few typical displacement traces recorded for slip bonded particles at shear rates γ⋅ of (a) 200 s−1, (b) 400 s−1, and (c) 533 s−1. Rolling regimes are shown in blue, and

stopped regimes in yellow.

Overview of a few typical displacement traces recorded for catch bonded particles at shear rates γ⋅ of (a) 200 s−1, (b) 400 s−1, and (c) 533 s−1. Rolling regimes are shown in blue, and

stopped regimes in yellow.

Histograms are generally well-described by a Gaussian distribution (black lines). Distributions for slip bonded particles broaden and move to higher vroll at increasing γ⋅, while

distributions of catch bonded particles remain stable at lower values of vroll.

of (a) 187 s−1, (b) 400 s−1, and (c) 560 s−1.

Rolling regimes are shown in blue, and stopped regimes in yellow.

s−1 (b) 400 s−1, and (c) 560 s−1.

Rolling regimes are shown in blue, and stopped regimes in yellow.

DNA sequence used in the study, Discussion, Rolling adhesion simulation protocol, videos, Rolling adhesion particle statistics, Tables 1–6 and Figs. 1–20.

Render of an oxDNA MD simulation showing a catch bond interconverting from state 1 to state 2 under 30 pN of overall tension. Under tension, the hairpin opens quickly, which is then followed

by a closing of the latch to yield state 2. The simulation consisted of 1.6 × 109 MD steps in total, and a snapshot is shown every 2 × 106 simulation steps.

Render of an oxDNA MD simulation showing a catch bond relaxing from state 2 to state 1 in the absence of tension. Initially, the toe hold forms rapidly, followed by an internal strand

exchange that displaces the complementary strand and reforms the hairpin to yield state 1. The simulation consisted of 1.6 × 109 MD steps in total, and a snapshot is shown every 2 × 106

simulation steps.

Close-up of a typical rolling adhesion experiment using slip bonds (SB-1 and SB-2) performed at γ⋅ = 400 s−1 (left) and γ⋅ = 667 s−1 (right). Particle traces are shown as yellow lines, and

tracked particles are colour-coded based on whether they are in a rolling regime (green) or stopping regime (red). A clear increase in rolling velocity is observed at increasing

γ⋅.

Close-up of a typical rolling adhesion experiment using catch bonds (CB-HS and CB-CS) performed at γ⋅ = 400 s−1 (left) and γ⋅ = 667 s−1 (right). Particle traces are shown as yellow lines,

and tracked particles are colour-coded based on whether they are in a rolling regime (green) or stopping regime (red). At γ⋅ = 400 s−1, a larger fraction of particles display rolling

behaviour than at the higher shear rate of γ⋅ = 667 s−1.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or

format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or

other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in

the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the

copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this articleCite this article van Galen, M., Bok, A., Peshkovsky, T. et al. De novo DNA-based catch bonds. Nat. Chem. 16, 1943–1950 (2024). https://doi.org/10.1038/s41557-024-01571-4

Download citation

Received: 26 June 2023

Accepted: 06 June 2024

Published: 24 June 2024

Issue Date: December 2024

DOI: https://doi.org/10.1038/s41557-024-01571-4

Share this article Anyone you share the following link with will be able to read this content:

Get shareable link Sorry, a shareable link is not currently available for this article.

Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative