Molecular engineering of a minimal E-cadherin inhibitor protein derived from Clostridium botulinum hemagglutinin

Hemagglutinin (HA), a nontoxic component of the botulinum neurotoxin (BoNT) complex, binds to E-cadherin and inhibits E-cadherin-mediated cell–cell adhesion. HA is a 470 kDa protein complex comprising six HA1, three HA2, and three HA3 subcomponents. Thus, to prepare recombinant full-length HA in vitro, it is necessary to reconstitute the macromolecular complex from purified HA subcomponents, which involves multiple purification steps. In this study, we developed NanoHA, a minimal E-cadherin inhibitor protein derived from Clostridium botulinum HA with a simple purification strategy needed for production. NanoHA, containing HA2 and a truncated mutant of HA3 (amino acids 380–626; termed as HA3mini), is a 47 kDa single polypeptide (one-tenth the molecular weight of full-length HA, 470 kDa) engineered with three types of modifications: (i) a short linker sequence between the C terminus of HA2 and N terminus of HA3; (ii) a chimeric complex composed of HA2 derived from the serotype C BoNT complex and HA3mini from the serotype B BoNT complex; and (iii) three amino acid substitutions from hydrophobic to hydrophilic residues on the protein surface. We demonstrated that NanoHA inhibits E-cadherin-mediated cell–cell adhesion of epithelial cells (e.g., Caco-2 and Madin–Darby canine kidney cells) and disrupts their epithelial barrier. Finally, unlike full-length HA, NanoHA can be transported from the basolateral side to adherens junctions via passive diffusion. Overall, these results indicate that the rational design of NanoHA provides a minimal E-cadherin inhibitor with a wide variety of applications as a lead molecule and for further molecular engineering.

E-cadherin is an important cell-adhesion molecule that mediates cell-cell adhesion at adherens junctions (AJs) in epithelial cells (15,16). AJs are located beneath tight junctions, which prevent diffusion of molecules between apical and basolateral membranes (17). Type I classic cadherins, including E-cadherin, contain five extracellular domains (1)(2)(3)(4)(5), a single transmembrane domain, and an intracellular domain. The extracellular 1-2 domains of type I classic cadherins bind another cadherin to form transdimer and cisdimer in a calcium ion-dependent manner. E-cadherin-mediated cell-cell adhesion plays a pivotal role in the development and maintenance of tissue organization. E-cadherin-mediated cell contacts are known to inhibit cell proliferation through the Hippo signaling pathway (18). HA/A and HA/B inhibit E-cadherin-mediated cell contacts (12,13,19) and promote the proliferation of Caco-2 cells (human colon carcinoma-derived epithelial cell line), T84 cells (human colon carcinoma-derived epithelial cell line), and Madin-Darby canine kidney type I (MDCK-I) cells (canine kidney epithelial cell line) (19). Furthermore, HA disperses human-induced pluripotent stem (iPS) cell aggregates in 3D suspension culture, resulting in a larger number of live cells, higher cell density, and higher-fold expansion than those of cells treated with conventional digestive enzymes (20). Therefore, human iPS cell culture systems with HA facilitate simple and robust maintenance of undifferentiated cells (20)(21)(22).
HA forms a heterododecameric complex that adopts a large triskelion-shaped structure (Fig. 1A) (3). In a previous study, the recombinant HA/B complex was reconstituted in vitro, by mixing the recombinant proteins of HA1, HA2, and HA3 and incubating at 37 C for 3 h (3). Thus, three protein expression systems (HA1, HA2, and HA3) and four purification steps (three HA subcomponents and in vitro reconstitution) are required to obtain the HA/B complex.
In the present study, we developed NanoHA, a minimal E-cadherin inhibitor protein derived from HA (termed NanoHA). NanoHA inhibits cell-cell adhesion and disrupts the epithelial barrier, similar to full-length WT HA. Furthermore, NanoHA has one-tenth the molecular weight of full-length HA and can be purified in sufficient quantities using a simple purification strategy.

Minimization of HA
To identify the essential HA fragments that disrupt the epithelial barrier, we tested the barrier-disrupting activity of 1A, 2D and S1). Within 4 h post-addition, 100 (protomer) nM HA/B, HAΔ1/B, and Mini-HA/B (#1) reduced the transepithelial electrical resistance (TER) of the Caco-2 cell monolayer to 25% of that at 0 h post-addition, whereas 100 nM of Mini-HAΔ1/B (#2) did not affect the TER (Fig. 1B). In this study, we calculated protein concentration in protomer units for HA/B and HAΔ1/B because HA comprises three protomers (3,23), and each protomer has one E-cadherin-binding site (14): that is, 1 nM HA/B is equal to three protomer nM HA/B.

Inhibition of cell-cell adhesion
To test the effect of NanoHA on cell-cell adhesion in different cell lines, we added NanoHA to Caco-2, HT-29 (human colon carcinoma-derived epithelial cell line), CMT-93 (mouse rectum carcinoma-derived epithelial cell line), and MDCK-I cells. We found that NanoHA inhibited cell-cell adhesion similarly to HA/B (Figs. 5 and S2). In particular, the colonies of HT-29, CMT-93, and MDCK-I cells were dispersed by these treatments (Figs. 5 and S2). The treated cells largely remained attached to the culture plates (Figs. 5 and S2), suggesting that HA and NanoHA did not inhibit cell-matrix adhesion.
Binding to the basolateral cell surface at 4 C To approach AJs from the basolateral side, basolaterally applied HA/B initially binds to the basal surface of Caco-2 cells. The cell-bound HA/B is then transported to the lateral surface of the cell in an E-cadherin-binding ability-dependent manner (24). We propose that cadherin flow (28) transports HA from the basal to lateral surface. NanoHA comprises the Engineering of E-cadherin inhibitor protein core components of HA, which are essential to inhibit Ecadherin and has one-tenth molecular weight (47 kDa versus 488 kDa) and one-fourth molecular size (65 Å versus 280 Å diameter) compared with that of full-length HA (Figs. 1A and 6C). Therefore, we hypothesized that NanoHA is advantageous in terms of accessibility to intercellular spaces. To test this hypothesis, we added HAs to the basolateral sides of Caco-2 cell monolayers and incubated them at 4 C. Consequently, NanoHA was localized to both the basal and lateral cell surfaces (Fig. 6, A and B), whereas HA/B bound only to the basal surface (Fig. 6, A and B) (24). This indicates that NanoHA can access AJs by passive diffusion, and that HA/B is too large to penetrate lateral intercellular spaces (Fig. 6C).

Discussion
In this study, we developed a minimal E-cadherin inhibitor protein, NanoHA, derived from C. botulinum HA through rational design. scMini-HAΔ1/CB-LD-YFDY (construct #2-3-2, termed NanoHA) has three types of modifications: a short  Engineering of E-cadherin inhibitor protein peptide linker to preserve structural integrity, chimerization, and three point mutations (Figs. 2B, 3, B and C and S1). NanoHA inhibited cell-cell contacts and disrupted the epithelial barrier (Figs. 3E, 5 and S2). To obtain high-quality purified recombinant protein, full-length HA needs four purification steps: purification of HA1, HA2, and HA3 and fulllength HA after in vitro reconstitution. In contrast, NanoHA can be purified in a single step and shows a 10-fold higher protein yield than that of full-length HA (data not shown).
Cadherin-mediated cell-cell adhesions are abrogated by other cadherin inhibitors, such as ADH-1 (also known as Exherin), a cyclic peptide composed of the His-Ala-Val sequence of N-cadherin (29); Epep, an E-cadherin mimic peptide H-SWELYYPLRANL-NH2 (30); and E-cadherinneutralizing antibodies (31,32). These E-cadherin inhibitors induce apoptosis in some cell lines by abrogating cell-cell contact (32)(33)(34). In contrast, HA/A and HA/B are not toxic to epithelial cells, such as Caco-2 cells and MDCK-I cells, but  Engineering of E-cadherin inhibitor protein rather promote cell proliferation (19). NanoHA also does not affect cell viability of Caco-2, HT-29, CMT-93, MDCK, and HeLa cells (Fig. S3). Recently, we reported a simple and robust method to maintain iPS cells in an undifferentiated state using full-length HA (20)(21)(22). We demonstrated that NanoHA, as well as full-length HA, disrupts the epithelial barrier (Fig. 3E) and inhibits E-cadherin-mediated cell-cell contacts in epithelial cells (Figs. 5 and S2). Thus, NanoHA can be used in novel iPSC culture systems. NanoHA (47 kDa) has one-tenth the molecular weight of full-length HA (native HA, 470 kDa; recombinant HA including affinity tags, 488 kDa) (Fig. 1A) and is favorable for penetrating intercellular spaces (Fig. 6). Therefore, NanoHA has the advantage of being an E-cadherin inhibitor for tightly packed 3D-cultured iPSC aggregates compared with full-length HA. Furthermore, NanoHA provides a wide variety of applications as a useful basal protein tool for further molecular engineering, such as directed evolution using phage display (35), and addition of other protein components (36) or functional signal peptides such as celltargeting peptides (37).

Plasmid construction
Genomic DNA was extracted and purified from C. botulinum serotype B strain Okra and serotype C strain Stockholm. HA1 (amin acids 7-294) derived from the serotype B BoNT complex (HA1/B)-encoding gene, excluding the stop codon, was cloned into the NheI-SalI site of the pET28b(+) vector (Merck), and an oligonucleotide encoding a FLAG tag was inserted at the C terminus of HA1/B. HA2 (amino acids 2-146) derived from the serotype B BoNT complex (HA2/B) or serotype C BoNT complex (HA2/C) encoding gene was cloned into the NheI-SalI site of the pET28b(+) vector. Fulllength HA3 (amino acids 19-626) derived from the serotype B BoNT complex (HA3/B)-encoding gene, excluding the stop codon, was cloned into the NcoI-SalI site of the pET52b(+) vector (Merck). The truncated mutant of HA3 (termed HA3 mini , amino acids 380-626) derived from the serotype B BoNT complex (HA3 mini /B)-encoding gene, excluding the stop codon, was cloned into the NcoI-SalI site of the pET52b(+) vector. An oligonucleotide encoding a Strep-tag II tag was inserted at the C termini of HA3 and HA3 mini . For single-chain Mini-HAΔ1 (scMini-HAΔ1) proteins, HA3 mini , an oligonucleotide encoding a short linker sequence (GSGGDDPPG), HA2, and an oligonucleotide encoding a Strep-tag II tag were cloned into the NcoI-SalI site of the pET52b(+) vector. Sitedirected mutagenesis was performed using PrimeSTAR Max Polymerase (TaKaRa Bio). The inserted regions of these plasmids and the presence of mutations were confirmed by DNA sequencing.

Protein expression and purification
Rosetta2 (DE3) Escherichia coli cells (Merck) were grown in Terrific broth medium. The expression of HA proteins was induced using Overnight Express Autoinduction System 1 (Merck) at 18 C for 48 h. Cells were harvested and lysed in PBS (pH 7.4) by sonication. His-HA1-FLAG, His-HA2, and His-NanoHA were purified using HisTrap HP (Cytiva). Streptag II-tagged proteins were purified using StrepTrap HP (Cytiva). These column purification steps were performed following the manufacturer's protocols. All proteins were dialyzed against PBS (pH 7.4) and stored at −80 C until further analysis. The protein concentration of the samples was determined using the Pierce bicinchoninic acid assay (Thermo Fisher Scientific).

In vitro reconstitution and purification
The HA protein complexes were reconstituted and purified as previously described (3). For HA and Mini-HA, purified HA1, HA2, and HA3 (or HA3 mini ) were mixed at a molar ratio of 4:4:1. For HAΔ1 and Mini-HAΔ1, purified HA2 and HA3 (or HA3 mini ) were mixed at a molar ratio of 4:1.
TER assay TER was measured using Millicell-ERS (Merck) as previously described (12). Briefly, HAs were added to the basolateral chambers of a Transwell (Corning) with the Caco-2 cell monolayer, and the plates were incubated at 37 C with 5% CO 2 . TER was measured at time points up to 24 or 72 h postaddition.

Optical microscopy
Caco-2, HT-29, CMT-93, and MDCK cells were incubated with 0 to 100 (protomer) nM of HA/B or NanoHA at 37 C with 5% CO 2 for 24 h in 24-well plates (Iwaki). The plates were stained with Giemsa Stain Solution (FUJIFILM Wako Chemicals) according to the manufacturer's protocol and observed using a BZ-X700 all-in-one fluorescence microscope (KEYENCE).

Immunofluorescence
The basolateral side of Caco-2 cells grown on a Transwell was treated with 17 nM (51 protomer nM) of HA/B (His-HA1/ B-FLAG + His-HA2/B + HA3/B-strep), or 51 nM of His-NanoHA were added to the basolateral chamber of Transwell with the Caco-2 cell monolayer. The plates were then incubated at 4 C for 40 min. After washing, the cells on Transwell filter membranes were fixed with 4% PFA at room temperature for 15 min, permeabilized with 0.5% Triton X-100/PBS at room temperature for 5 min, and blocked with 5% bovine serum albumin/PBS. The cells were then incubated with mouse anti-His-tag monoclonal antibody (OGHis, MBL; 1:1000 dilution) and rat anti-E-cadherin monoclonal antibody (DECMA-1, Merck; 1:1000 dilution), followed by Alexa Fluor 488-conjugated anti-mouse immunoglobulin G antibody (Thermo Fisher Scientific; 1:400 dilution) and Cy3-conjugated anti-rat immunoglobulin G antibody (Jackson ImmunoResearch; 1:400 dilution). The slides were mounted with Pro-Long Diamond Antifade mountant. Images were acquired by confocal microscopy using an IX71 microscope (Olympus) and a CSUX1 confocal scanner unit (Yokogawa Electric) and processed using Metamorph software (Molecular Devices).
Engineering of E-cadherin inhibitor protein Structure modeling of NanoHA The molecular model of MiniHAΔ1/CB was built by manually docking the crystal structures of HA2 from the HA1/ D-HA2/D complex (Protein Data Bank ID: 2E4M) onto that of HA/B (Protein Data Bank ID: 3WIN), as HA2/D shows 99% amino acid sequence homology with HA2/C. A short peptide linker sequence, Gly-Ser-Gly-Gly-Asp-Asp-Pro-Pro-Gly, was inserted between the C terminus of HA3 mini /B and the N terminus of HA2/D, and the model was refined by simulated annealing using GROMACS (GROMACS Development Team) (38). The figures were generated using PyMOL (The PyMOL Molecular Graphics System, version 2.4.0, Schrödinger, LLC).

Data availability
All experimental data are contained within the article.
Supporting information-This article contains supporting information.