Grafting T7 loop to CDRs in VHH
FtsZ polymerization is performed by the interaction between the T7 loop in FtsZ and another FtsZ. Therefore, inhibition of the T7loop binding can lead to the inhibition of cell division. To interfere with the interaction between the T7loop and another FtsZ, in this study, we attempted to create a molecule that overcomes competitive inhibition with FtsZ by enhancing the affinity of VHH grafted with T7loop by evolutionary molecular engineering. First, parent VHH (cAbCII10, VHH fragments of camel anti-BcII b-lactamase antibody) were genetically engineered by grafting T7loops in CDR1, CDR2, and CDR3. The amino acids involved in the interaction of the T7 loop with other FtsZ were determined by crystal structural analyses. The I201, A202, E206, N208, and D210 residues of the T7loop interact with other FtsZ monomers, and N208, D210, and D213 residues are related to the GTPase activity. To retain the function of the T7loop, we designed it to contain as many amino acid residues as possible. Furthermore, given that the T7 loop contains a part of the helix at both ends, and the CDR in the scaffold VHH does not have a helix, we designed the T7 loop of the three different lengths to reduce the structural distortion. Specifically, loop A: loop with N-terminal and C-terminal helices, loop B: loop with only N-terminal helix left, or loop C: loop with both removed (interaction residues I201 and A202 are also removed). The location for grafting to the CDR was also carefully chosen, while the structure of the T7loop was maintained. The differences in the distance between the ends of T7loop A, B, and C predicted from the three-dimensional structure analysis was 10 Å, 5.3 Å, and 8.8 Å. We chose the length of the replacement CDR, which would be the same end distance of these loops. In loop A: GGSEYSYSTF (endo-to-endo distance of 13.1 Å), SGGSEYSYSTF (12.8 Å) in CDR1, and YFM RLP (9.9 Å) in CDR3, in loop B: ASMGGL (5.2 Å) and SMGG (5.5 Å) in CDR2, or in loop C: RGYFMRLPSSHNF (8.3 Å) and RGYFMRLPSSHNFR (8.6 Å) in CDR3 were grafted (Table 1). As a result, seven variants, namely, VHHA11, VHHA12, VHHA33, VHHB21, VHHB22, VHHC31, and VHHC32, were constructed. These variants were expressed in Escherichia coli (E. coli), purified to homogeneity, and were subsequently used in the following experiment.
Characterization of grafted VHH variants
Changes in the secondary structure of VHH variants constructed by T7 loop grafting were evaluated with the use of circular dichroism spectrometry (Fig. S1). None of the mutants yielded remarkable structural differences from the parent VHH. Instead, they yielded a typical immunoglobulin fold structure pattern. In VHHA11 and VHHA12, the molar ellipticity at approximately 200 nm decreased slightly, thus indicating that the random structures increased by loop grafting. The binding affinity of VHH variants for FtsZ was evaluated with the use of the enzyme-linked immunosorbent assay (ELISA) (Fig. 2). All variants had an increased binding affinity to the parent. In particular, VHHB22 bound most strongly to FtsZ in all the variants. The binding affinity for FtsZ was measured using surface plasmon resonance (SPR) for parent and VHHB22 in detail (Fig. 3). The parent could not detect the binding affinity for FtsZ, whereas VHHB22 showed binding affinity when Kd = 4.3.
Random mutagenesis of another CDR in VHHB22 variants and screening of variants with phage display
Furthermore, we tried to improve the affinity of the obtained VHHB22 with the use of an evolutionary engineering approach based on the phage display method. The VHHB22 gene was inserted into the phagemid vector to display VHHB22 to the pIII of M13 filamentous phage, and random mutations of CDR1 or CDR3 in VHHB22 were performed by overlap-extension polymerase chain reaction (PCR) (Table 2). A convenient NNK degenerate codon that can cover all 20 amino acids was employed for random mutation using primers, as shown in Table S1. Because the CDR1 and CDR3 libraries have 17 and 20 residues, respectively, the theoretical sequence space is estimated to be contain 2017 (1.3 × 1022) and 2020 (1.0 × 1026). The actual magnitudes of the obtained phage libraries were 4.1 × 105 colony forming units. Therefore, it is shown that this library is a condition that does not meet the theoretical sequence space of the mutant. This phage was subjected to biopanning while the concentration of the target FtsZ was gradually reduced from round to round. After four rounds of biopanning, the eluted phages were infected with E. coli, and the formed 186 colonies were monocloned and cultured. Recombinant phages in the supernatant were analyzed by phage ELISA to evaluate the binding affinity of FtsZ (Fig. S2a and b).
Consequently, signal improvements were observed in 174 out of 186 clones in the CDR1 library and in 184 clones out of 186 in the CDR2 library, despite not satisfying the theoretical mutation sequence space. Sixteen clones were selected from each library and sequenced. In particular, the VHHB22-35 sequence occupied nine out of 16 clones of the CDR3 library (Table S2). As a result, three types of sequences from CDR1 and seven sequences from CDR3 were identified (Table S2). These genes were transferred to an E. coli expression vector, and the protein was purified (Fig. S3). Finally, only four mutants (VHHB22-3, VHHB22-23, VHHB22-35, and VHHB22-97) were functionally expressed in E. coli. One mutant (VHHB22-42) interacted unspecifically with the Size exclusion chromatography (SEC) column, and the other four mutants formed soluble aggregates.
Characterization of evolved VHH variants
ELISA analysis was performed to analyze the binding affinity for FtsZ of the purified VHHB22-3, VHHB22-23, VHHB22-35, and VHHB22-97 (Fig. 4). All variants showed enhanced activity compared with the parent VHH and VHHB22. In particular, VHHB22-23 and VHHB22-97 showed significantly enhanced activities. Subsequently, VHHB22-3 and VHHB22-35 showed enhanced activity. To eliminate the possibility that the apparent affinity was increased by becoming a multimer and to calculate the KD value precisely, SPR analysis was performed with a small amount of immobilized FtsZ (Fig. 5, Table S3). Given that VHHB22-35 showed the highest binding affinity (KD = 0.042 mM), it was found that VHHB22-35 had the highest binding activity. Therefore, we investigated whether VHHB22-35 could inhibit the polymerization of FtsZ (Fig. 6a, b and c). In the presence of GTP and MgCl2, where FtsZ polymerization occurred, the parent VHH, VHHB22, and VHHB22-35 were mixed with FtsZ and reacted for 30 min. After centrifugation, the soluble (Fig. 6a) and insoluble fractions (Fig. 6b) were analyzed with the use of polyacrylamide gel electrophoresis (SDS-PAGE). When FtsZ and parent VHH were mixed, the polymer was approximately half polymerized, whereas when VHHB22 was mixed with FtsZ, the polymer was approximately one-fourth. At VHHB22-35, the polymer completely disappeared; this indicated that a strongly inhibited VHH variant could be obtained (Fig. 6c). In addition, E. coli growth in isopropyl β-D-thiogalactopyranoside (IPTG)-induced conditions was observed (Fig. 6d). E. coli growth was not delayed when parent VHH was mixed, whereas it was significantly delayed when VHHB22 was added. At VHHB22-35, it completely inhibited growth.