The anti-ssDNA antibody has the innate conformation for binding to the ssDNA
The apo (non-binding form) and ssDNA binding forms of anti-ssDNA antibody has been identified by the X-ray crystallography. Their crystal structures, were downed from the protein database bank and with IDs as 5GKR and 5GKS, respectively [7], which were shown in Fig. 1A and B. The detailed binding behavior of antibody and ssDNA were analyzed and visualized, as shown in Fig. 1C, a considerable number of non-bonding interactions between the ssDNA and the antibody were found, e.g. the H-bonds were formed between the amino acid residues Y50, S54, S56, N58, H95 from Heavy chain and the ssDNA, and the Y95 and Y96 from the light chain could form the H-bonds with the ssDNA through their hydroxy groups. In addition, the pi-pi stacking interactions were observed between the aromatic group of side chains of residues and the base groups of the ssDNA, which are Y95L with the dT2 of the ssDNA, W98H, F33L with the dT3 of ssDNA.
To explore whether the ssDNA binding affects the arrangement orientation of amino acid residues on the anti-DNA antibody, the key residues derived from the crystal structures of apo (5GKS) or ssDNA binding form (5GKR) were analyzed and the results were shown in Fig. 1D-F. As shown, the residual orientation between two forms (apo vs ssDNA binding) was consistent and the corresponding RMSD was 0.481 Å, which suggests that the orientation of the key residues would not be influenced by the ssDNA binding. And moreover, the overlapping of CDR loops was also performed to explore the impact on the conformation change of CDR loops due to ssDNA binding, The obtained results showed that no obvious deviation was observed in the CDR loops between apo and ssDNA binding antibody forms (Fig. 1F). Especially for the H-CDR3, very few or no visible conformation changes were observed after ssDNA binding, take together, these observations suggested that the conformation of this anti-DNA antibody CDR have not been influenced by the binding to antigen ssDNA, implying that this anti-ssDNA antibody has the innate conformation for binding to the ssDNA.
Characterize the influence of F33Y mutation on the conformation change of the Variable fragment region
It was reported that single-point mutation F33Y in the heavy chain of anti-ssDNA antibody could abolish its ssDNA binding ability [7]. To investigate the structural change of anti-ssDNA antibody caused by F33Y mutation, 200ns-molecular dynamic simulations procedures were employed. Firstly, the RMSD analysis was performed to evaluate the average amount of movement of backbone atoms throughout the entire protein structure, and the corresponding RMSD values fluctuation were shown in Fig. 2A. As we can see, the RMSD value of the heavy chain of apo-anti-ssDNA antibody mainly varied between the 1.6Å and 2.8 Å, whereas, the light chain showed relatively high variation of the RMSD value mainly between the 1.8Å and 4Å. In contrast, the RMSD values of the Y33F mutant heavy chain and light chain showed similar variation (between the 1.6Å and 2.8Å). In addition, the superimposition analysis was performed between the crystal structure and the MD-optimized structure, as can be seen in the MD-optimized wildtype anti-ssDNA antibody has only small conformational change relative to the crystal structure (Fig. 2B). Moreover, the overall architecture of F33Y mutant is similar to the crystal structure of wildtype antibody (Fig. 2C). These data revealed that F33Y single-point mutation had less or none significant influence on the architecture of this antibody.
To assess the conformational flexibility of the wildtype and mutant antibody, the root mean square fluctuations (RMSF) analysis were performed. The RMSF values were calculated by measuring atomic fluctuations after superimposing each structure of a MD trajectory onto the initial structure by means of least-squares fitting, to remove rotational and translational motion [13]. The results were shown that the corresponding RMSF value variation of amino acids located on the H-CDR1, H-CDR2 and H-CDR3, had the similar trend both in the wildtype and F33Y mutant antibodies, and that the value slightly decreased closed to the 72th residues of F33Y mutant heavy chain (Fig. 2D), presumably because this region orientated outside of the antibody leading to more flexibility. Interestingly, no obvious difference in the RMSF value was observed in the 33th residues where the mutation F33 to Y happened. However, the RMSF value in the constant 1 region (CH1) of F33Y mutant heavy chain was smaller than that of wildtype, which indicates that the CH1 may be influenced by the H-F33Y mutation although this region is “spatially” far away from the mutation site. For the light chain of the antibody, similar trends were observed that the overall fluctuation of RMSF of the F33Y mutant were similar to those in the wildtype antibody, but some regions were exception, e.g, RMSF values of the resides in the F33Y mutant light chain CDR1 (L-CDR1) was less, whereas the values of L-CDR3 and L-CH1 is larger than those in wildtype. Taken together, these results demonstrated that the F33Y mutation in the heavy chain of anti-ssDNA antibody had more impact on the flexibility/mobility of light chain than heavy chain.
PCA helps to determine the most significant motion in dynamics trajectory. It was also carried out to investigate the important motions during the dynamic simulation process, and the covariance matrix of atomic fluctuations was diagonalized for predicting the eigenvalues. As the first few eigenvectors play a central role in the motions of protein, thus the first two eigenvectors of wildtype and F33Y mutant antibody were shown in the Fig. 2E and F, respectively, as we can see, there exist significant differences in the CDR loops of the light chain, in comparison, less difference existed in the H-CDR loops. But for the constant domains (CH1-CL), the motion of the top 2 eigenvectors are quite different between wildtype and F33Y mutant antibodies, it is suggested that F33Y mutation can influence the dynamic behaviors of the CH1 region of antibody.
The mutation F33Y impairs the ssDNA binding ability of anti-DNA antibody
To further explore the influence of the F33Y mutation on the ssDNA binding of this antibody, the MD simulation analysis was employed to imitate the dynamic process of wildtype and mutant antibody binding to the ssDNA. After 200ns MD simulation, the RMSD of the Cα for the wildtype antibody-ssDNA complex and F33Y mutant antibody-ssDNA complex were detected and collected. After binding to ssDNA, the RMSD values of heavy chains in wild-type antigen-antibody complex was mostly distributed between 2 Å and 3 Å, while the RMSD values of the light chains and the ssDNA in the complex were mostly between 1.6 Å and 2.4 Å and between 0.8 Å and 1.6 Å, respectively (Fig. 3A). which indicates that the conformation of ssDNA has slightly change in relevant to its initial crystal structure. In contrast, for the F33Y mutant antibody-ssDNA complex, although the RMSD values of the heavy chain were still between 2 Å to 3 Å, both the light chain and ssDNA had increased, especially it reached to 2.2 Å or above for the ssDNA (Fig. 3B). which indicated that great conformation change happened in ssDNA during its binding to F33Y mutants.
To further evaluate the impact of F33Y mutation on antibody heavy and light chains, RMSF analysis were performed. The obtained results were shown in the Fig. 3C, the superimposition of RMSF plot curves between the F33Ymutant and WT antibodies fitted very well, means that no obvious difference were observed. But for the ssDNA, the RMSF plot of ssDNA in the mutant complex was lower than that in the wildtype complex, which indicated that ssDNA suffered more limitation in the mutant complex.
Motivated by these observations, the equilibrated snapshots of wildtype and mutant antibody-ssDNA complexes were collected and compared them with the corresponding initial structures of MD. The superimposition structure was shown in Fig. 3D and 3G, obviously, there is no visual differences observed throughout the MD simulation for the WT and mutant complex. But PCA analysis on the dynamic behaviors of the mutant and wildtype antibody showed the CH1 region of the light chain shows quiet different orientation of movement between the WT and F33Y mutant (Fig. 3E and 3H).
Finally, the binding manner between the ssDNA and antibody were explored and shown in the Fig. 3F and 3I, as we can see, when the F33 mutated to Y, the hydroxy group of Y33 can hydrogen-bond with the thymine base of ssDNA (Fig. 3I), where F33 in the wildtype antibody cannot form the H-bond with the ssDNA (Fig. 3F). and the dT2 formed more H-bonds in F33Y mutant than in wildtype. Hence, these observations clearly showed that the F33Y mutation could change binding manner between the antibody and ssDNA.
F33Y mutation undermine/impair the interaction energy between antibody and ssDNA
To investigate the potential interaction mechanism between the ssDNA and antibody based on the energetic changes, the binding free energy of all ssDNA-antibody complexes was evaluated using the MM-GBSA methodology. The predicted binding free energies together with their corresponding energy contributions are summarized in Table 2. As we can see, the major contribution to stabilize the ssDNA-antibody complex is the electrostatic energy(ΔEele), However, comparing the ΔEele values between the wildtype (-78.4 ± 2.9 kcal/mol) and F33Y mutant (-72.0 ± 4.5kcal/mol) antibodies, it showed that there was no significant difference between them, which suggests that F33Y mutation has weak influence on the electrostatic interactions between the antigen ssDNA and the antibody. The second contributor for the stabilization of ssDNA-antibody complex are van der Waals(vdW) forces (ΔEvdW), which mainly form non-polar stacking interaction and play a crucial role in the binding between the protein and nucleotide[14, 15]. The results showed that ΔEvdW declined from − 54.0 ± 3.6 kcal/mol in wildtype antigen-antibody complex to -41.2 ± 7.1kcal/mol in mutant complex (Table 2), which indicates that F33Y mutation in H-CDR1 resulted in significant decrease in vdW interaction. whereas for other energetic items (ΔEGB and ΔESA), the difference is less significant in compared with ΔEvdW. So, it can be concluded that the loss of binding ability of F33Y mutant antibody to ssDNA was largely due to the loss the vdW interaction between the antigen ssDNA and F33Y mutant antibody.
Table 2
Binding free energies between antibody and ssDNA (kcal∙mol− 1) obtained via the MM/GBSA approach for the two models.
Antibody | Antigen | Energy components |
ΔEele | ΔEvdW | ΔEGB | ΔESA | ΔEtot |
WT | ssDNA | -78.4 ± 2.9 | -54.0 ± 3.6 | 80.7 ± 2.4 | -6.6 ± 0.5 | -58.3 ± 3.6 |
F33Y | ssDNA | -72.0 ± 4.5 | -41.2 ± 7.1 | 74.2 ± 4.5 | -5.7 ± 0.5 | -44.7 ± 6.9 |
ΔEtot = ΔEele + ΔEvdW + ΔEGB + ΔESA. The stand errors of the mean are listed in parentheses. |
To further investigate the energy contributions of individual residues, the residue-based free energy decomposition analysis was performed to calculate the energy of residues in the CDR of antibody heavy and light chain. The results showed that the interactions of 23 residues Y32, F33, Y50, I51, Y52, Y53, S54, G55, S56, T57, N58, Y59, K64, R94, H95, R96, N97, W98 in the heavy chain, and Y32, S94, Y95, Y96 in the light chain, were identified as contributor to the binding with the ssDNA (Fig. 4A, Table 3). For the wildtype, the residues F33H, W98 H, and Y96L were identified as the top three contributors for the ssDNA binding of the antibody (Fig. 4A, Table 3). For the F33Y mutant antibody-ssDNA complex, the residues Y/F33 H, W98 H, and Y95L were identified as key contributors. These 23 residues could be divided into four groups according to their properties, 11 amino acids in aromatic groups (11/23), 6 in polar groups (6/23), 3 in charged groups (3/23) and 3 in nonpolar group (3/23) (Table 3). Comparing the energy changes among groups, the aromatic groups contribute the most, followed by polar group and charged group, which further demonstrated that the electrostatic interactions does not play a critical role during the antibody-antigen binding process.
Table 3
The energy contribution of residues of antibody to its binding with the ssDNA
The properties of amino acid | Source | Amino acid | Energy contribution (kcal/mol) |
WT | F33Y |
Aromatic amino acids | H-CDR3 | W98 | -4.5 | -18.36 | -4.28 | -14.28 |
L-CDR | Y95 | -4.22 | -2.93 |
H-CDR1 | F/Y33 | -3.89 | -3.54 |
H-CDR2 | Y52 | -3.23 | -1.72 |
H-CDR2 | Y50 | -1.01 | -0.36 |
L-CDR | Y96 | -0.43 | -0.52 |
H-CDR3 | H95 | -0.33 | -0.31 |
H-CDR1 | Y32 | -0.3 | -0.28 |
H-CDR2 | Y53 | -0.21 | -0.19 |
L-CDR | Y32 | -0.14 | -0.13 |
H-FRW3 | Y59 | -0.1 | -0.02 |
Polar amino acid | H-CDR2 | S56 | -3.34 | -8.49 | -2.34 | -5.6 |
H-FRW3 | T57 | -2.06 | -0.52 |
H-CDR3 | N97 | -1.19 | -1.36 |
H-FRW3 | N58 | -1.06 | -0.2 |
H-CDR2 | S54 | -0.73 | -1.11 |
L-CDR | S94 | -0.11 | -0.07 |
Charged amino acid | H-CDR3 | R96 | -0.45 | -0.68 | -0.45 | -0.63 |
H-FRW3 | K64 | -0.13 | -0.08 |
H-CDR3 | R94 | -0.1 | -0.1 |
Nonpolar amino acid | H-CDR2 | G55 | -0.45 | -0.77 | -0.69 | -0.88 |
H-CDR2 | I51 | -0.23 | -0.11 |
H-CDR3 | L99 | -0.09 | -0.08 |
Furthermore, analyzing the location of these residues found that they were mainly located in the heavy chain CDR1 (H-CDR1, 2/23), CDR2 (H-CDR2, 7/23), CDR3 (H-CDR3, 6/23), framework 3 (H-FRW3, 4/23), and light chain CDRs (L-CDRs, 4/23) (Fig. 4A). For the wildtype, the H-CDR2 contributed greatest in compared to other regions, followed by the H-CDR3, then L-CDRs, H-CDR1 and H-FRW3 (Fig. 4B). For the F33Y mutant, H-CDR2 and H-CDR3 have the greatest contributions to the ssDNA binding, followed by H-CDR1 and L-CDRs, then H-FWR3. Yet it is worth noting that the binding energies of H-FR3, H-CDR2, and L-CDRs were largely reduced in F33Y mutant complex comparing with wildtype. Taken together, these results implied that the residues located in these regions might be the most crucial for the binding of antibodies to ssDNA, and that F33Y mutation causing the antibody to lose its antigen binding ability was achieved by changing the energy of the key amino acids in these regions.
Pi-pi stacking interaction but not hydrogen bonds between antibody and ssDNA was undermined/impaired upon F33Y mutation
Previous study showed that the aromatic pi-pi stacking interaction, a major type of vdW force, was one of the most fundamental interaction forces for the high affinity and specificity of autoantibody [16]. We identified Y/F33H, W98H and Y95Las key amino acid residues for ssDNA binding in both wild-type and mutant antibody. All of them have aromatic cyclic structures, indicating that they are likely to form stable pi-pi interactions with DNA bases. To verify this, the non-bond interactions detection analyses were performed to inspect the pi-pi stacking interactions between the key residues and the nucleobase of ssDNA. The results showed that F33H, W98H and Y95L could form the stable pi-pi stacking interaction with the nucleotide base group of ssDNA in the wildtype antibody during the whole MD process (Fig. 5A and 5B). While for the F33Y mutant, the Y33H and W98H could also form the stable pi-pi interaction with the ssDNA, which were similar in the wildtype. However, for the Y95L in the F33Y mutant, the distance between dT2 and Y95 (ddT2−−−Y95) suddenly increased from the 4 Å to more than 5 Å (critical distance of pi-pi stacking is less than 5 Å) at the end of the MD process, which means that this pi-pi stacking was finally impaired (Fig. 5C and 5D). Moreover, it also could see that the dT2 ring of the ssDNA was sandwiched by the F33 and W98 to form the parallel pi-pi-pi stacking interaction in the wildtype antibody-ssDNA complex (Fig. 5A). However, F33Y mutant, the parallel sandwich pi-pi-pi stacking were destroyed due to that the Y33 cannot form the pi-pi stacking interaction with the dT2 (Fig. 5C). Hence, these results demonstrated that F33Y mutation weaken the pi-pi stacking interactions between the antibody and ssDNA.
Hydrogen bond interaction is also one of the most important non-bonding interactions between the antibody and antigen. To explore whether the hydrogen bond interaction contributes to the ssDNA binding of the wild type and the mutant antibody, the MD simulation timescale was performed. The results showed that Y50H, T57H, Y96H, and W98H could form the same number of stable hydrogen bonds with ssDNA in both the wild type and the mutant (Table 4 and Fig. 6), indicating that the mutation did not alter the way and quantity of these residues binding to DNA. S56H could form more hydrogen bonds in the wild type, but the N58H residue could form more stable hydrogen bonds with ssDNA in the F33Y mutant. Additionally, although Y33H and S54H in the mutant could form additional hydrogen bonds with ssDNA, these hydrogen bonds had very short duration throughout the entire MD process, implying that these bonds may not be the contributors to the overall stability of the binding process.
Table 4
Intermolecular hydrogen bonds between ssDNA and antibodya.
Amino acid | WT | F33Y |
Acceptor | Donor | Ocpyb | Distc | Angd | Acceptor | Donor | Ocpyb | Distc | Angd |
Y50 | dT3@O4 | Y50H@HH Y50H@OH | 99.52 | 2.78 | 164.43 | dT3@O4 | Y50H@HH Y50H@OH | 99.8 | 2.8 | 163.9 |
Y96 | dT3@O4 | Y96L@HH Y96L@OH | 98.57 | 2.82 | 160.31 | dT3@O4 | Y96L@HH Y96L@OH | 98.5 | 2.8 | 160.51 |
W98 | dT3@O2 | W98H@H W98H@N | 96.36 | 2.98 | 154.36 | dT3@O2 | W98H@H W98H@N | 95.4 | 3 | 154.09 |
T57 | dT1@O2 | T57H@H T57H@N | 29.79 | 2.98 | 147.58 | dT1@O2 | T57H@H T57H@N | 53.4 | 3 | 142.21 |
T57H@O | dT2@H3 dT2@N3 | 32.85 | 2.95 | 151.48 | T57H@O | dT1@H3 dT1@N3 | 54.7 | 3.1 | 146.99 |
S56 | dT2@O4' | S56H@HG S56H@OG | 44.12 | 2.97 | 146.9 | dT2@O4' | S56H@HG S56H@OG | 56.3 | 2.9 | 148.47 |
dT1@O3' | S56H@HG S56H@OG | 28.95 | 2.9 | 144.58 | / | / | / | / | / |
N58 | dT2@O2 | N58H@HD21 N58H@ND2 | 55.67 | 2.86 | 162.22 | dT2@O2 | N58H@HD21 N58H@ND2 | 99.6 | 2.9 | 162.42 |
N58H@OD1 | dT2@H3 dT2@N3 | 41.74 | 2.92 | 162.62 | N58H@OD1 | dT2@H3 dT2@N3 | 96.8 | 2.9 | 161.86 |
/ | / | / | / | / | N58H@OD1 | dT1@H3 dT1@N3 | 28.9 | 3.1 | 136.85 |
Y33 | / | / | / | / | / | dT2@O3' | Y33H@ HH Y33H@ OH | 36.7 | 3 | 138.13 |
/ | / | / | / | / | dT2@O2 | Y33H@ HH Y33H@ OH | 35.8 | 2.9 | 146.07 |
S54 | / | / | / | / | / | dT2@OP1 | S54H@HG S54H@OG | 45.8 | 2.7 | 163.57 |
a The subscripted H and L indicate the residues reside in the antibody heavy and light chains, respectively. |
b Hydrogen bond occupancy during MD (%). |
c Time averaged hydrogen bond length (Å). |
d Time averaged hydrogen bond angle (º). |
Pi-pi stacking interactions dominated the binding between antibody and ssDNA
In order to confirm the dominated role of the pi-pi stacking interaction between the anti-DNA antibody and the ssDNA, the F33YH, F33AH, W98AH, Y95AL mutants and the wildtype antibody were expressed and purified (Fig. 7A). Their binding performance towards the ssDNA were evaluated by the ELISA experiments. The corresponding results showed that almost all the mutants lost the ssDNA binding ability (Fig. 7B) due to impairing of pi-pi stacking interaction, which further demonstrated that pi-pi stacking was one of the key contributors for this anti-ssDNA antibody.