Interdomain Linker Effect on the Mechanical Stability of Ig Domains in Titin

Titin is the largest protein in humans, composed of more than one hundred immunoglobulin (Ig) domains, and plays a critical role in muscle’s passive elasticity. Thus, the molecular design of this giant polyprotein is responsible for its mechanical function. Interestingly, most of these Ig domains are connected directly with very few interdomain residues/linker, which suggests such a design is necessary for its mechanical stability. To understand this design, we chose six representative Ig domains in titin and added nine glycine residues (9G) as an artificial interdomain linker between these Ig domains. We measured their mechanical stabilities using atomic force microscopy-based single-molecule force spectroscopy (AFM-SMFS) and compared them to the natural sequence. The AFM results showed that the linker affected the mechanical stability of Ig domains. The linker mostly reduces its mechanical stability to a moderate extent, but the opposite situation can happen. Thus, this effect is very complex and may depend on each particular domain’s property.


Introduction
The giant muscle protein titin is a tandem modular construction designed polyprotein containing more than two hundred individually folded domains, such as immunoglobulinlike (Ig) and fibronectin-type III domains [1]. These domains are similar in size (~2 nm) and length (~90 residues). Interestingly, these domains are closely connected with very few residues in-between. Functionally, the I-band part of titin is extensible and plays a critical role in the passive elastic properties of muscles ( Figure 1a). Thus, the molecular design for such a giant polyprotein is of great interest in understanding its mechanical function [2][3][4][5]. Here, we are particularly interested in whether the interdomain amino acid sequence/linker affects the mechanical stability of the Ig domains in titin, which has been studied with great interest [6][7][8].
Thus, we choose six consecutive Ig domains of titin, including I27, I28, I29, I30, I31, and I32, as a representative unit (I27-I31), to study the interdomain linker effect (Figure 1b). The structures and mechanical stabilities of many of these domains have been well determined. The crystal structure of I27 is available, while structures for other domains are constructed by the I-TASSER server (Figure 1b). It is noted that I27 is a historical nomenclature,  Thus, we choose six consecutive Ig domains of titin, including I27, I28, I29, I30, I31,  and I32, as a representative unit (I27-I31), to study the interdomain linker effect ( Figure  1b). The structures and mechanical stabilities of many of these domains have been well determined. The crystal structure of I27 is available, while structures for other domains are constructed by the I-TASSER server ( Figure 1b). It is noted that I27 is a historical nomenclature, especially in the AFM-SMFS field, and we used it here. It is renamed I91 later [63]. There are 89 extensible residues in each Ig domain except for I30, leading to ΔLc of ~28 nm, as shown for I27. For I30, a disulfide bond may be formed between the Cys23 and Cys73, leading to a smaller ΔLc with only 51 extensible residues (Figure 1c). In addition, we designed nine glycines (9G) as an artificial linker due to its simple structure without a self-secondary structure.

Results
A high-precision AFM measurement system has been used for accurate measurement and comparison of Ig domains in titin [64][65][66][67] (Figure 1d). In short, the target polyprotein designed with a specific peptide sequence NGL was immobilized on a GL peptidecoated surface through AEP (asparaginyl endopeptidase)-mediated protein ligation between the two peptide sequences [68]. A GB1-XDoc coated-AFM tip was used to probe the target polyprotein ( Figure 2c). Here, GB1 with known properties (Force = 180 pN, ΔLc = 18 nm) was added, serving as an internal force caliper [69]. The reversible protein-protein interaction Cohesion:XDockerin (Coh:XDoc) was used to enable efficient protein pickup [64].
Thus, we built polyprotein Coh-I(27-32)/9G-NGL with a 9G linker for measurement. The 9G linker is present between each Ig domain except for the two end I27 and I32 domains ( Figure 2a). By approaching the AFM tip towards the surface, the polyprotein was picked up between the Coh:XDoc interaction. Upon stretching, the polyprotein was under mechanical manipulation and its corresponding force-extension curve showed characteristic sawtooth-like peaks from the stepwise unfolding of each domain, and a final rupture

Results
A high-precision AFM measurement system has been used for accurate measurement and comparison of Ig domains in titin [64][65][66][67] (Figure 1d). In short, the target polyprotein designed with a specific peptide sequence NGL was immobilized on a GL peptide-coated surface through AEP (asparaginyl endopeptidase)-mediated protein ligation between the two peptide sequences [68]. A GB1-XDoc coated-AFM tip was used to probe the target polyprotein ( Figure 2c). Here, GB1 with known properties (Force = 180 pN, ∆Lc = 18 nm) was added, serving as an internal force caliper [69]. The reversible protein-protein interaction Cohesion:XDockerin (Coh:XDoc) was used to enable efficient protein pick-up [64].
Thus, we built polyprotein Coh-I(27-32)/9G-NGL with a 9G linker for measurement. The 9G linker is present between each Ig domain except for the two end I27 and I32 domains ( Figure 2a). By approaching the AFM tip towards the surface, the polyprotein was picked up between the Coh:XDoc interaction. Upon stretching, the polyprotein was under mechanical manipulation and its corresponding force-extension curve showed characteristic sawtoothlike peaks from the stepwise unfolding of each domain, and a final rupture peak with a higher force of~600 pN was observed from the break of Coh:XDoc complex. By fitting the elasticity of the curve using the worm-like chain model [70], five unfolding events from Ig domains were obtained, showing a ∆Lc of 28 ± 2 nm (Figure 2b), agreeing well with the theoretical value. Moreover, the unfolding forces of different Ig domains are similar. The force histogram showed a single peak with an average unfolding force of 308 ± 64 pN (ave. + stdv. n = 1148, Figure 2c). Besides, an additional peak with a ∆Lc of 11 ± 2 nm was observed, which was from the partial unfolding of I30 ( Figure 1c). As described, a disulfide bond is indeed presented in I30. Thus, only 40 residues can be unfolded, leading to a theoretical value of 10.4 nm (40 × 0.36 − 4.0 nm). events from Ig domains were obtained, showing a ΔLc of 28 ± 2 nm (Figure 2b), agreeing well with the theoretical value. Moreover, the unfolding forces of different Ig domains are similar. The force histogram showed a single peak with an average unfolding force of 308 ± 64 pN (ave. + stdv. n = 1148, Figure 2c). Besides, an additional peak with a ΔLc of 11 ± 2 nm was observed, which was from the partial unfolding of I30 ( Figure 1c). As described, a disulfide bond is indeed presented in I30. Thus, only 40 residues can be unfolded, leading to a theoretical value of 10.4 nm (40 × 0.36 − 4.0 nm). Then, we used polyprotein Coh-I(27-32)-NGL with natural sequence for AFM measurement and comparison. The same cantilever used previously for the polyprotein with the linker was used here again to minimize the error. As expected, a similar unfolding pattern was observed (Figure 2b,d,e). However, the force is slightly lower, with a value of 324 ± 54 pN (n = 1808, Figure 2c) ( Table 1).
To confirm this effect, we chose three Ig domains only and constructed two shorter polyproteins, Coh-I(28-30)/9G-NGL and Coh-I(30-32)/9G-NGL, for measurement. Indeed, stretching these polyproteins resulted in a shorter force-extension curve with only two 28 nm-peaks from I28, I29, or I31, I32, and one 11 nm-peak from I30, as expected (Figures 3  and 4). For Coh-I(28-30)/9G-NGL, the histogram of unfolding forces from I28 and I29 showed a single peak with an average force of 330±36 pN (n = 860). Then, AFM measurement on Coh-I(28-30)-NGL with natural sequence showed a force of 325 ± 35 pN (n = 965). For Coh-I(30-32)/9G-NGL, the histogram of unfolding force from I31 and I32 showed a single peak with an average force of 320 ± 33 pN (n = 1120). However, AFM measurement on the natural sequence showed a different result. Two peaks were observed in the histogram, with a force of 276 pN and 345 pN, respectively (n = 1804), which has not been observed before. As a result, the unfolding force of Ig domains in titin is generally lower when the 9glycine length amino acids sequence is present as an interdomain linker. Moreover, the effect can be complex when considering the unfolding force of I31 and I32. Finally, we focused on the unfolding results of I30 in each polyprotein design. First, with an internal disulfide bond, I30 showed a unique 11 nm-peak which can be distinguished from other Ig domains. Thus, its unfolding force can be assigned unambiguously. Moreover, the linker situation for I30 in the three polyproteins is different ( Figure 5). For I (27)(28)(29)(30)(31)(32), 9G linker is present on both sides of I30 (Figure 5a). In this design, the unfolding force of I30 was 203 ± 34 pN (n = 360), and 186 ± 55 pN (n = 225) without linker. For I (28)(29)(30), 9G linker is only present on the N terminus. The force was 197 pN (n = 570), and 17 2± 24 pN (n = 910) without linker. Finally, for I(30-32), 9G linker is only present on the C terminus. The force was 159 ± 23 pN (n = 477), and 175 ± 24 pN (n = 404) without the linker ( Table 2). For Coh-I(30-32)/9G-NGL, the histogram of unfolding force from I31 and I32 showed a single peak with an average force of 320 ± 33 pN (n = 1120). However, AFM measurement on the natural sequence showed a different result. Two peaks were observed in the histogram, with a force of 276 pN and 345 pN, respectively (n = 1804), which has not been observed before.
As a result, the unfolding force of Ig domains in titin is generally lower when the 9-glycine length amino acids sequence is present as an interdomain linker. Moreover, the effect can be complex when considering the unfolding force of I31 and I32.
Finally, we focused on the unfolding results of I30 in each polyprotein design. First, with an internal disulfide bond, I30 showed a unique 11 nm-peak which can be distinguished from other Ig domains. Thus, its unfolding force can be assigned unambiguously. Moreover, the linker situation for I30 in the three polyproteins is different ( Figure 5). For I (27)(28)(29)(30)(31)(32), 9G linker is present on both sides of I30 (Figure 5a). In this design, the unfolding force of I30 was 203 ± 34 pN (n = 360), and 186 ± 55 pN (n = 225) without linker. For I (28)(29)(30), 9G linker is only present on the N terminus. The force was 197 pN (n = 570), and 172 ± 24 pN (n = 910) without linker. Finally, for I (30)(31)(32), 9G linker is only present on the C terminus. The force was 159 ± 23 pN (n = 477), and 175 ± 24 pN (n = 404) without the linker ( Table 2). Based on these results for I30, we found that this linker effect is much more complex than we thought before. Indeed, a few cases have been studied for the linker effect, both mechanically and thermodynamically [71,72]. No general trend/conclusion has been obtained. In this work, we found the linker can reduce the mechanical stability of I30 when present in I (27)(28)(29)(30)(31)(32) and I (30)(31)(32) while increasing it when present in I (28)(29)(30). Nevertheless, it is no doubt that the linker affects the domain stability in titin. In this work, we determined the effect of the interdomain linker for the Ig domain in titin. By measuring the mechanical stability of polyprotein containing multiple Ig domains in titin, with/without an artificial 9G linker, we found the linker indeed affects the Ig domain's stability. The force is reduced when an artificial linker is present in most cases. However, the extent can vary; sometimes the trend is reversed. Thus, we believe this linker effect is much more complex, and the intrinsic property of each domain and the linker itself play important roles. Nevertheless, this work provides a glimpse of the molecular design of the giant titin, and future studies are needed to understand this important molecule for humans and even for designing artificial muscle [73][74][75].
In this work, we determined the effect of the interdomain linker for the Ig domain in titin. By measuring the mechanical stability of polyprotein containing multiple Ig domains in titin, with/without an artificial 9G linker, we found the linker indeed affects the Ig domain's stability. The force is reduced when an artificial linker is present in most cases. However, the extent can vary; sometimes the trend is reversed. Thus, we believe this linker effect is much more complex, and the intrinsic property of each domain and the linker itself play important roles. Nevertheless, this work provides a glimpse of the molecular design of the giant titin, and future studies are needed to understand this important molecule for humans and even for designing artificial muscle [73][74][75].