Linker peptide acts as an external mechanical clamp in a designed polyprotein

Rational protein design has successfully enhanced the mechanical stability of biomaterials by optimizing the internal topology of force-bearing domains. However, current design strategies predominantly focus on isolated monomers, paying less attention to the immediate oligomeric environment, such as the connecting linkers, modulates mechanical resistance. Here, we report a “Neighbor Peptide Effect” in SuperMyo A339, a recently de novo designed mechanostable protein with an engineered shear topology. Using single-molecule force spectroscopy, we demonstrate that oligomerization via a specific flexible linker (RSGGS) amplifies mechanical stability (∼15%), elevating the unfolding force from ∼350 pN in the monomer to ∼400 pN in the tetramers. Steered molecular dynamics simulations, together with contact-occupancy analysis, elucidate the molecular mechanism underlying this enhancement, revealing a discrete, sequence-specific interaction between the linker and the force-bearing loop of the adjacent domain. Specifically, an interfacial hydrogen bond between linker residue Ser117 and domain residue Lys218, which exhibits a 0.21 occupancy increase at the transition state, acts as an external mechanical clamp. This interaction may restrict loop stochasticity, optimize the pulling geometry and stabilize the core hydrogen-bond network, raising the energy barrier for unfolding without altering the protein’s core fold. These findings indicate that interfacial peptide design may serve as an external means to adjust the mechanical properties of protein.

Mechanical proteins serve as the building blocks of force-bearing tissues, providing the elasticity and toughness required for biological functions ranging from muscle contraction to cellular mechanotransduction (Labeit and Kolmerer, 1995; Figure 1a). The immunoglobulin-like domain I27 from human cardiac titin has long served as a model for understanding protein mechanics (Rief et al., 1997; Brockwell et al., 2003). Single-molecule force spectroscopy (SMFS), particularly using Atomic Force Microscopy (AFM), has revealed that the mechanical stability of such domains is largely governed by the topology of their force-bearing hydrogen bonds (Klimov and Thirumalai, 2000; Müller and Dufrene, 2008; Hoffmann et al., 2013). In the canonical I27 fold, a “shear topology” involving multiple hydrogen bonds between the A′ and G β-strands acts as a mechanical clamp, resisting unfolding until a force of approximately 200 pN is applied (Carrion-Vazquez et al., 1999).

Driven by the demand for mechanically superior biomaterials, rational protein design has focused on engineering artificial proteins that surpass the mechanical limits of naturally occurring domains. Recently, we developed a de novo designed protein termed “SuperMyo A339.”(Zheng et al., 2026; Figure 1b) By elongating the force-bearing β-strands of the I27 scaffold and engineering the hydrogen bond network, we successfully increased the number of shear-topology hydrogen bonds from the native count to eight. This structural modification yielded a substantial enhancement in mechanical stability, elevating the unfolding force to ∼350 pN (Figure 1c). These findings underscored the efficacy of topological engineering and AI design in tuning the mechanical properties of isolated protein monomers (Liu et al., 2025; Meng et al., 2025).

Interestingly, in both physiological environments and synthetic applications, mechanical proteins not only function as isolated monomers (Li et al., 2002). In sarcomeres, titin domains are arranged in tandem arrays (Figure 1a). A prevailing assumption in single-molecule biophysics has been that domains within a polyprotein behave as independent mechanical entities, with the linker regions serving merely as passive tethers. Consequently, current design strategies often focus on the internal core of the monomer, overlooking how the immediate local environment, specifically interactions with neighboring short peptide sequences or linkers, might modulate mechanical resistance (Chen, et al., 2013; Zacharchenko et al., 2015; Lanzicher et al., 2020; Tong et al., 2022).

In this study, we report a “neighbor peptide effect” where the oligomerization of SuperMyo A339 significantly amplifies its mechanical strength beyond the intrinsic stability of the monomer. Using force spectroscopy on engineered polyproteins (e.g., A339 tetramers; Figure 1d) containing a specific and indispensable flexible enzymatic cleavage sequence (RSGGS), we observed an unfolding force of ∼400 pN, a distinctive increase compared to the ∼350 pN characteristic of the monomer. To elucidate the molecular mechanism underlying this enhancement, we employed Molecular Dynamics (MD) simulations and Steered Molecular Dynamics (SMD) simulations (Karplus and McCammon, 2002). Our simulations reveal that in the oligomeric state, the enhancement is driven by a discrete interaction between the connecting short peptide and the loop region of the force-bearing domain (Figure 1e). Specifically, residues within the linker form a critical extra hydrogen bond with the domain’s loop, which not only fortifies the resistance network but also optimizes the pulling geometry by restricting loop stochasticity, thereby raising the energy barrier for mechanical unfolding.

These results demonstrate that the mechanical phenotype of a protein is defined not only by its internal topology but also by its neighbor environment. By validating that oligomeric construction with fitted peptide connected can synergistically enhance mechanical resistance, our findings provide a novel strategy to tune protein mechanical stability without altering the core fold. This approach helps bridge molecular-level design with macroscopic mechanical behavior and may inform the development of protein-based hydrogels and biomimetic materials with improved performance.

Protein expression and purification

The proteins were ordered from General Biol. Each protein was constructed into the pET30a, resulting in constructs of pET30a-sdrG-GB1-POI-GB1-his6-NAL for unfolding force spectroscopy experiments. The plasmids were transformed into BL21(DE3) cells. Single colonies were selected and cultured overnight at 37 °C in LB medium containing kanamycin for pET30a vector. The culture was then transferred to 800 mL of kanamycin-containing LB and incubated for 2 h at 37 °C on a shaker. When the OD 600 of the culture reached 0.6–0.8, isopropyl-β-D-1 thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM, and the culture was incubated at 18 °C on the shaker for 16 h.

The bacteria were harvested by centrifugation at 13,000 × g for 10 min at 12 °C, and resuspended by the lysis buffer containing 20 mM Tris-HCl, 150 mM NaCl, pH 7.4. Prior to cell lysis, DNase, RNase, and PMSF were added to the cell suspension. The suspension was then lysed using a homogenizer for 3 min, followed by centrifugation at 39,000 × g for 15 min at 4 °C to separate the supernatant fraction.

Protein sdrG-GB1-POI-GB1-his6-NAL was purified by Ni-NTA affinity chromatography column. Impurities were washed with a Wash buffer (20 mM Tris-HCl, 400 mM NaCl, 20 mM imidazole, pH 7.4) and bound proteins were eluted by Elution buffer (20 mM Tris-HCl, 150 mM NaCl, 250 mM imidazole, pH 7.4). Protein concentrations were quantified by measuring the absorbance at 280 nm and adjusted to a final concentration of ∼5 mg/mL for subsequent experiments.

AFM-SMFS experiment

Single-molecule force spectroscopy experiments were performed using a NanoWizard 4 atomic force microscope (JPK) equipped with the functionalized silicon nitride cantilevers. The spring constant of each cantilever was calibrated using the thermal noise method (Florin et al., 1995; Shi et al., 2022).

To ensure site-specific and directional pulling, we employed a receptor-ligand handling system based on the ultra-stable Fgβ–SdrG complex (Milles et al., 2018). The GL-functionalized cantilevers were coupled with Fgβ–GB1–NGL, while the GL-functionalized glass substrates were coupled with the target polyprotein SdrG–GB1–POI–GB1–NAL. Protein immobilization was achieved via Oa AEP1(C247A)-catalyzed peptide ligation, which covalently linked the NGL/NAL tags on the proteins to the GL peptides on the surfaces (Yang et al., 2017).

In this setup, the C-terminal GB1-NAL domain anchored the protein to the surface, while the N-terminal SdrG domain was available to bind Fgβ on the cantilever tip. The high mechanical stability of the Fgβ–SdrG complex (>2 nN) ensured robust maintenance of the tether during unfolding experiments.

Each measurement cycle consisted of tip–surface contact at a force of 300 pN for 400 ms to allow bond formation, followed by vertical retraction at a constant velocity of 1 μm/s. Force-extension traces were recorded and analyzed using JPK Data Processing software.

It should be noted that the difference in unfolding-event numbers between the constructs (n = 1657 vs. n = 515) arises from differences in single-molecule pickup probability and distinct adsorption/surface-interaction behaviors inherent to the different linker compositions. The same stringent event-selection criteria were applied to all constructs, and both datasets amply meet the statistical requirements for reliable single-molecule analysis.

AFM-SMFS data analysis

All recorded force-extension traces were screened to identify specific single-molecule interactions based on the characteristic unfolding fingerprints of the flanking GB1 domains. The selected traces were analyzed using the worm-like chain (WLC) model (Marko and Siggia, 1994) implemented in Python to characterize the elasticity of the unfolded polypeptide chain. The WLC model describes the force-extension relationship as follows: where F(x) is the applied force, x is the extension of peptide, l p is the persistence length, L c is the contour length, k B is the Boltzmann constant, and T is the Kelvin temperature.

SMD simulation

Protein structure files were pre-processed with the Biopython package to align the N-terminus and C-terminus along the z-axis, the designated pulling direction. The size of the simulation water box was determined based on the protein size and the pulling distance. Consequently, the simulation water box was constructed with dimensions of 4.978 × 5.02 × 20.43 nm, providing a simulation environment that accommodates the full pulling pathway. The systems were then solvated and neutralized using 150 mM NaCl, resulting in a total of 51,845 atoms. All simulations were performed using GROMACS 2025.3 with the CHARMM36m force field and the TIP3P water model (Van Der Spoel et al., 2005; Huang et al., 2017).

To generate a fully equilibrated starting configuration for mechanical unfolding, the system first underwent 5000 steps of energy minimization, followed by 125 ps of NVT equilibration and a 5 ns pre-equilibration in the NPT ensemble. Subsequently, a 100 ns production Molecular Dynamics (MD) simulation was conducted to relax the protein conformation. Then, the final frame from this 100 ns trajectory would serve as the initial configuration for the SMD experiments.

Prior to the SMD simulations, this template structure underwent a second round of 5000 steps of energy minimization to eliminate steric clashes, followed by 125 ps of NVT simulations with positional restraints on the protein backbone atoms. A subsequent 5 ns pre-equilibration in the NPT ensemble was conducted under periodic boundary conditions, maintaining a temperature of 310 K using the V-rescale thermostat and a pressure of 1 bar using the C-rescale barostat. From this equilibrated state, 10 independent 5 ns MD simulations were performed to sample diverse conformations. RMSD and inter-domain contact analysis were used to evaluate whether the simulations had reached equilibration, which was determined by the emergence of stable plateaus in both metrics (details in the Supplementary Figure S5). The last frame of each equilibrated 5 ns trajectory was then extracted and defined as the starting structure for the subsequent pulling experiments.

For the SMD simulations, 15 independent pulling runs were performed for each of the 10 starting structures, resulting in a total dataset of 150 trajectories per protein. The SMD pulling experiments were carried out in the NVT ensemble, maintaining a constant temperature of 310 K using only the V-rescale thermostat. In each simulation, the C-terminal of the protein was restrained with a fixed harmonic potential (1000 kJ/mol/nm 2 for x, y, z direction), while a harmonic pulling potential (1000 kJ/mol/nm 2) was applied along the positive z-axis to the Cα atoms of the N-terminal residues to induce unfolding with a constant pulling speed 1 nm/ns. The simulation time required for the pulling was determined by the pulling length and pulling speed, and was calculated as follows: time = pulling length (nm)/pulling speed (nm/ns). The force-distance curves generated from the SMD simulations provided the unfolding forces of the proteins. The mechanical strength of the protein was evaluated by averaging the unfolding forces obtained from the 150 times simulations.

Contact analysis

To quantify the force-bearing interactions and identify mechanostable contacts during both equilibrium and dynamic unfolding processes, we employed the Overlap and residue Contact Surface Units (OV + rCSU) method (Wołek et al., 2015; Moreira et al., 2022). The Overlap (OV) criterion treats each residue as a collection of van der Waals (vdW) spheres and a contact is assigned when any pair of spheres between residues sterically overlaps. The repulsive Contacts of Structural Units (rCSU) criterion further refines this by incorporating the chemical and electrostatic properties of the interacting atoms, categorizing atomic interactions into attractive andrepulsive types. A residue–residue contact is considered a valid stabilizing interaction only when the number of attractive interactions exceeds that of repulsive ones. By combining these criteria, the OV + rCSU scheme provides the contacts that capture both steric packing and chemical complementarity.

For the equilibrium MD simulations, the OV + rCSU algorithm was used to perform a contact evolution analysis, monitoring the persistence of inter-domain interactions to define the stability of the “closed” conformation.

For the SMD pulling experiments, a more granular analysis was conducted to capture the structural features of the unfolding transition. Specifically, we extracted the single frame immediately preceding the maximum force from each of the 150 independent trajectories. These frames were then used to compute the contact occupancy, defined as the probability of a specific residue pair being in contact across all 150 structures. To isolate the mechanical contribution of the RSGGS linker, a differential contact analysis was performed by subtracting the average contact matrix of A339EAAAK5 from A339RSGGS. Due to the difference in linker length (5 residues for RSGGS vs. 25 residues for EAAAK5), the residue indices were re-mapped by aligning both A339 variants to the C-terminal domain.

Neighboring peptide sequences enhance mechanical stability

Our investigation into the mechanical properties of SuperMyo A339 began with an unexpected observation. A339 is a de novo designed protein developed in our previous work, where we utilized a diffusion model to rationally expand the force-bearing hydrogen bond network of the native I27 template. By extending the force-bearing β-strands to accommodate eight concurrent backbone hydrogen bonds parallel to the force vector (Figures 1b,c), A339 requires the simultaneous rupture of this entire network via a shearing geometry to unfold. This process is similar to the unfolding of I27. (Lu and Schulten, 2000). While A339 monomers typically unfold at ∼350 pN, preliminary AFM experiments on A339 homo-tetramers revealed a distinct population of unfolding force approaching ∼400 pN. To assess whether the higher unfolding force was inherent to the A339 building block rather than a consequence of the multimeric arrangement, we reduced the system to its simplest functional unit by engineering dimeric A339 constructs suitable for precise single-molecule analysis.

To determine whether the force enhancement arose from the specific characteristic of the linker, we constructed two distinct A339 dimers fused with the GB1 domain as an internal force caliper (Figure 2a). The first construct, A339RSGGS, utilizes the short, flexible enzymatic cleavage sequence (RSGGS) originally present in the tetramer design. The second construct, A339EAAAK5, employs a rigid helical linker, (EAAAK) 5, which acts as a stiff spacer to physically separate the domains and minimize interfacial contact. This experimental design allows us to isolate the contribution of the short connecting peptide.

Single-molecule force spectroscopy (SMFS) was performed using a high-precision AFM system (Figure 2a; Tian et al., 2021). To ensure data reliability, we employed the GB1 fingerprinting method, using the well-established unfolding force of GB1 (corrected to 213 pN) as an internal standard to normalize force measurements across different cantilevers (Cao et al., 2006). Force-extension curves were analyzed using the worm-like chain (WLC) model to verify contour length increments (ΔL c), confirming that the observed peaks correspond to the specific unfolding of A339 domains (Figure 2b).

The comparative analysis of the force distributions suggests a clear dependency on the linker type. The force histogram for the rigid-linker control, A339EAAAK5, exhibits a single Gaussian distribution with an average unfolding force of 344 ± 31 pN (mean ± stdv, n = 1657, Figure 2c). This value aligns well with previously reported values for the isolated A339 monomer (352 pN), confirming that the rigid helical linker effectively isolates the domains and precludes mechanical crosstalk. In stark contrast, the A339_RSGGS dimer displays a shifted distribution with a higher average unfolding force of 372 ± 38 pN (n = 515, Figure 2C). To assess whether the two distributions differ statistically, an independent t-test was performed, revealing a significant difference between the two linker types (p < 0.05).

This force increase of approximately 30 pN in the dimer confirms that the mechanical enhancement observed in the tetramer is indeed reproducible and localized to the interface between the domain and its immediate neighbor. Although the magnitude of enhancement in the dimer is slightly lower than that observed in the tetramer—a discrepancy likely attributable to the reduced statistical probability of finding the weakest link in a shorter chain—the qualitative effect is robust. These results suggest that the enhancement is not an artifact of aggregation, but rather the result of a specific “neighbor peptide effect,” where residues within the short RSGGS sequence interact with the A339 domain to reinforce the force-bearing hydrogen bond network. This implies that simple peptide additions, often treated as passive linkers in protein engineering, can be rationally exploited to engineer significant gains in mechanical stability.

Static interfacial interactions and the limits of equilibrium analysis

To elucidate the structural origin of the mechanical enhancement in A339RSGGS and to generate equilibrated configurations for subsequent SMD, we performed molecular dynamics (MD) simulations. We hypothesized that the flexible linker might facilitate direct domain-domain interactions, potentially stabilizing the dimer in a “closed” state. Analysis of 100 ns equilibrium trajectories confirmed that the A339RSGGS relaxes into a compact conformation where the two domains maintain persistent contact (Figure 3a). The occupancy of this “closed” state was estimated to be ∼40%, as supported by inter-domain interaction energy and OV + rCSU contact evolution analysis (Figure 3b). Specifically, the C-terminal domain’s loop region forms a stable interface with the surface of the α-helix in the adjacent domain. In contrast, the control construct with the rigid linker, A339EAAAK5, cannot form such interdomain contacts.

While the identification of this closed conformation provides an initial explanation for the stability difference, it is crucial to distinguish between thermodynamic stability and mechanical resistance (Li et al., 2000). The existence of a stable interface in the ground state does not necessarily translate to a higher energy barrier for unfolding under load. Mechanical unfolding is a non-equilibrium kinetic process, and though a stable interface in equilibrium may still indirectly influence the load-bearing β-sheet by reorganizing the hydrogen-bond network, the static “closed” conformation is still insufficient to explain the 30–50 pN force enhancement. Thus, we proceeded to simulate the dynamic stretching process for further research.

The neighboring peptide effect: a dynamic stabilization mechanism

In order to probe the dynamic mechanism underlying the force enhancement, we performed constant-velocity SMD simulations on A339 dimers. Given the inherent stochasticity of non-equilibrium pulling processes, single SMD trajectory is insufficient to capture the full mechanical landscape. Therefore, we conducted a large battery of independent pulling simulations to generate statistically robust force distributions for three distinct dimer constructs: the rigid control (A339EAAAK5), the target sequence (A339RSGGS), and a specificity control (A339RSGGG), where the critical serine is mutated to glycine to test sequence dependence.

Meanwhile, it should be noted that a significant six-order-of-magnitude discrepancy exists between the pulling speeds of AFM-SMFS (∼1 μm/s) and SMD simulations (∼1 nm/ns). This high pulling rates typically used in silico tend to produce higher rupture forces and may bias unfolding pathways toward non-equilibrium events (Isralewitz et al., 2001; Franz et al., 2020). Therefore, we emphasize that SMD is used only to characterize mechanistic trends and relative mechanical behavior. The results presented here are intended to provide qualitative mechanistic insights and capture relative differences in mechanical stability, rather than to quantitatively reproduce AFM force magnitudes.

The force histograms (Figure 4a) reveal distinct results. The rigid A339EAAAK5 construct exhibits a force distribution centered at 1340 ± 165 pN, consistent with the rupture of isolated domains. As the simulation progresses, the α-helical linker unravels first, effectively decoupling the two domains before the load propagates to the core. Subsequently, the force rises monotonically until it exceeds the threshold of the β-strand clamp, triggering a catastrophic, avalanche-like rupture of the hydrogen bond network. Interestingly, the flexible specificity control, A339RSGGG, exhibits a similar force profile 1333 ± 142 pN, suggesting that flexibility alone cannot drive the enhancement.

In contrast, the A339RSGGS construct displays a bimodal force distribution characterized by two distinct populations centered at 1244 pN and 1464 pN. The two-peak analysis reveals that while the lower-force population falls below the unfolding force of the isolated domain, a substantial second population shifts into a higher-force regime. Furthermore, statistical testing indicates that the overall mechanical behavior of A339RSGGS is distinct from that of the rigid A339EAAAK5 control (p = 7.446e-2). The trajectory analysis of A339RSGGS reveals a mechanism that does not primarily rely on the equilibrium “closed” state. During the initial stages of pulling, the force induces a transition from the closed to an open conformation, straightening the flexible linker and separating the domain cores to a sufficient distance (Figures 4b,c). By the time the critical unfolding force is reached, the domains are spatially distant, precluding direct domain-domain interaction.

However, a closer inspection of the rupture event uncovers a subtle yet critical interaction: despite the global separation, the “neighboring peptide”—specifically residue Ser117 within the linker—remains intimately engaged with Lys218 located on the loop region of the force-bearing domain (Figures 1e, 4b), as supported by contact occupancy analysis (Figure 4c). This interaction forms an additional hydrogen bond, which we hypothesize to be a key contributor to the enhancement of unfolding force. In differential contact map, we found that the RSGGS linker not only introduces a specific new contact between linker residue Ser117 and domain residue Lys218 with a marked occupancy increase, but also enhances the internal contacts within the force-bearing β-sheet, effectively optimizing the force-bearing core of the domain (Figure 4c). In contrast, the long EAAAK5 spacer unfolds into an extended chain and does not form such stabilizing interactions.

We propose that this extra hydrogen bond is the linchpin of the observed mechanical reinforcement. Crucially, this bond is geometrically aligned to rupture concomitantly with the internal 8-bond network, effectively acting as the “9th bond” at the moment of peak load. Furthermore, this interaction likely imposes a steric constraint on the loop, acting as a “lever” that restricts the conformational entropy of the β-sheet. By minimizing the stochastic fluctuations of the loop (Lys218 region), the neighboring peptide forces the β-sheet to remain geometrically aligned with the pulling vector, thereby increasing the shearing component of the unfolding pathway and suppressing unzipping-type motions. These results highlight the complexity of the “Neighbor Peptide Effect.” It is not a generic consequence of oligomerization, but a specific, sequence-dependent phenomenon. The enhancement requires both an extended domain interface (the loop) and a complementary short peptide (RSGGS), demonstrating that simple, rational modifications to connecting sequences can function as external mechanical regulators, boosting stability by 30–50 pN without mutating the protein core and offering a new dimension for rational protein design with finer control over mechanical performance.

In this study, we have combined single-molecule force spectroscopy with molecular simulations to elucidate a novel mechanical reinforcement mechanism, termed the “Neighbor Peptide Effect,” within the oligomeric structures of the de novo designed protein SuperMyo A339. Our AFM results provide compelling evidence that the mechanical stability of a protein domain is not solely an intrinsic property of its fold; rather, the inclusion of a simple enzymatic cleavage sequence (RSGGS) as a connecting linker unexpectedly elevates the unfolding force by ∼30–50 pN compared to isolated monomers or rigid-linker controls.

Steered molecular dynamics (SMD) simulations provided key insight into the microscopic origin of this enhancement with the analysis of OV + rCSU contact occupancy map, revealing that the reinforcement operates through a specific, dynamic interaction during the critical phase of extension. Although the global tertiary structure transitions from a closed to an open conformation and the linker straightens under load, a critical hydrogen bond is preserved between residue Ser117 of the neighboring peptide and Lys218 on the force-bearing loop of the A339 domain. This interaction acts synergistically with the core hydrogen bond network, effectively raising the energy barrier required to cross the transition state. Furthermore, we propose that this external bond functions via a “lock-and-lever” mechanism: by constraining the conformational entropy of the loop, it optimizes the alignment of the force vector relative to the β-sheet, ensuring the unfolding event proceeds in a geometry maximizing the shearing component of β-strand separation.

These findings bridge a critical gap between molecular design and the complex microenvironment of polyproteins. We demonstrate that mechanical stability can be modulated by simple external peptide modifications, independent of the protein’s core topology. This establishes the “neighbor peptide” as a tunable design element, suggesting a potential strategy for engineering protein-based hydrogels and biomaterials in which mechanical toughness can be adjusted through the rational design of interfacial linkers, in addition to modifications to the protein core. Future work may further explore how short peptides with distinct sequence features influence proteins with diverse mechanical architectures, while incorporating the advanced coarse-grained approaches such as GoMartini 3 or OLIVES (Souza et al., 2025; Pedersen et al., 2024) could help bridge the remaining timescale gap between experiments and simulations, providing deeper molecular insight into force-dependent behavior and enabling more precise control over mechanical performance.