Crystal structure of a Vip3B family insecticidal protein reveals a new fold and a unique tetrameric assembly

Abstract

Vegetatively expressed insecticidal proteins (VIPs) produced by Bacillus thuringiensis fall into several classes of which the third, VIP3, is known for their activity against several key Lepidopteran pests of commercial broad acre crops and because their mode of action does not overlap with that of crystalline insecticidal proteins. The details of the VIP3 structure and mode of action have remained obscure for the quarter century that has passed since their discovery. In the present article, we report the first crystal structure of a full‐length VIP3 protein. Crystallization of this target required multiple rounds of construct optimization and screening—over 200 individual sequences were expressed and tested. This protein adopts a novel global fold that combines domains with hitherto unreported topology and containing elements seemingly borrowed from carbohydrate‐binding domains, lectins, or from other insecticidal proteins.

1. INTRODUCTION

Since its discovery in 19011 Bacillus thuringiensis (Bt, renamed in 1911–19152) has been in the research spotlight due to its ability to control susceptible insect larvae. Shortly after its discovery, Bt and its derivatives have found use as biopesticides (as fermented spore suspensions); in 1955 Hannay and Fitz‐James3 demonstrated that the major insecticidal principle of Bt is contained within the proteinaceous, crystalline parasporal inclusion bodies. With the advent of biotechnology, several of these crystalline (Cry) insecticidal proteins of Bt have been introduced into transgenic plants,4, 5, 6 culminating in the development of insect‐resistant crops which are now widely used across the world.

In addition to the numerous and highly diverse δ‐endotoxins produced during sporulation, Bt also produces vegetative insecticidal proteins (VIPs) that belong to several superfamilies numbered in chronological order of their discovery. The present work focuses on the third superfamily (VIP3), discovered in 1996 by Estruch et al.7 who have shown that these proteins are potent, broad‐spectrum (mainly against Lepidoptera) insecticidal agents secreted by vegetative cells of certain Bt strains. The original paper and the subsequent reports on VIP3 proteins revealed a number of significant insights into their sequence diversity, properties, and mode of action (MOA): it is now known that VIP3s belong to three or more subfamilies (VIP3A‐C) with 100+ reported members, that they are soluble, secreted with their signal sequence intact and that they can undergo specific proteolysis resulting in two or more fragments8, 9, 10, 11 that likely remain non‐covalently attached and are indeed potently insecticidal. Even though both VIP3s and Cry proteins attack the host via binding to the insect midgut epithelial membranes, the insecticidal receptor binding site(s) of VIP3 proteins (or MOA) is known to be distinct from that of the Cry proteins—therefore insect colonies resistant to Cry proteins are found to be sensitive toward VIP3s10 and vice versa.10, 12

Despite significant efforts from both the academic and industrial communities, and orthogonal to the wealth of functional knowledge that has been revealed during the ~25 years since the original VIP3A discovery, there is a distinct lack of structural information for these important proteins. Recent studies11, 13, 14 revealed that VIP3A is a tetramer that retains its assembly and general architecture following trypsinolysis—however, the resolution of the 2017 electron microscopy density envelope11 is too low to provide insight into the atomic‐level details of the protein.

We have set out to discover the structure of a representative member of the VIP3 superfamily of insecticidal proteins. The present article summarizes the atomic resolution model obtained by X‐ray diffraction studies of VIP3B2160, revealing the overall fold and key observations derived from the structure.

2. MATERIALS AND METHODS

Several internally sourced VIP3‐like sequences were computationally evaluated for completeness and similarity to the VIP3 superfamily. The chosen sequence shares 74% identity and 85% similarity with the sequence of Vip3Ba1—GenBank entry AAV70653.1 by Rang et al.15 As the subject protein currently lacks official designation, it will be referred to as VIP3B2160 throughout this work. For clarity, throughout this article all residue numbering refers to the native, un‐modified sequence without tags.

Extended Materials and Methods are provided as Supporting Information. In brief, the following work was performed to elucidate the structure of VIP3B2160.

Protein‐coding sequence was obtained directly from Bt DNA by PCR and cloned into pET28 vector with an N‐terminal His‐tag. Protein was expressed in BL21(DE3) RIPL cells using an autoinduction medium.16 Despite the presence of the intact secretion signal the protein was not released into growth medium—it is likely that the protein accumulated in the periplasmic space, similar to the observations made by Chen et al.17 who observed periplasmic accumulation of Vip3A during Escherichia coli expression. Purification of VIP3B2160 was accomplished by means of immobilized metal affinity chromatography (IMAC) on His‐SELECT resin (Sigma) followed by preparative size exclusion chromatography (SEC). Despite numerous attempts, crystallization of the intact, native VIP3B2160 did not occur.

Tryptic digest of VIP3B2160 under alkaline conditions resulted in the formation of a ~65 kDa stable tryptic core. While the native core did not crystallize by itself, it produced a number of promising granular precipitates in a wide range of crystallization conditions. Based on MS identification of the tryptic cut after Lys‐208 several constructs were designed starting with N‐205 (NVTK…) and an uncleavable N‐terminal His‐7 tag, such that the Cys‐Glu‐Asp repeat region 462–479 (…KVESWKEKSCEEDSCEDE…) was pared down to several shorter versions. The winning variant (Δ466‐474‐>S, …KVES‐S‐CEDE…) yielded crystals; this variant was produced as Selenomethionine‐substituted protein and its structure was determined at 2.7 Å resolution by multiwavelength anomalous diffraction analysis. Large portions of protein were disordered and the overall quality of the structure was not sufficient for detailed interpretation. Based on this partial structure, loop predictions, and crystal contact engineering considerations nearly 200 constructs were designed (in four sequential cohorts) of which approximately 125 expressed soluble proteins, 36 of which crystallized, and ultimately a single construct gave crystals that diffracted to 3.2 Å resolution. The winning construct was VIP3B2160‐FL with mutations E514A, K515A, Q517A, K518A as well as deletions Δ466‐474‐>S, and Δ581‐584. It crystallized in the space group P1 with four protomers in the asymmetric unit. Structure of the full‐length VIP3B2160 was solved by molecular replacement (Phaser18) using the truncated partial structure as the search model. Building of the four chains required considerable manual (Coot19) and automated (Buccaneer20) rebuilding, averaging‐assisted electron density modification (DM,21 Resolve22) and local model generation (YASARA23). Iterative refinement using Refmac24 resulted in a reasonably good quality final model of the full‐length protein based on MolProbity25 scores. Relevant details of data collection and refinement as well as salient model quality parameters for the full‐length structure and for the recombinant tryptic core are summarized in Supporting Information Tables I and II, respectively. The final structure of VIP3B2160 was deposited with the PDB/RCSB under accession code http://firstglance.jmol.org/fg.htm?mol=6V1V.

3. RESULTS AND DISCUSSION

The crystal structure of full‐length VIP3B2160 insecticidal protein was solved by means of extensive crystal engineering. The structure reveals an entirely new protein fold (Figure 1a) as established by comparative analysis conducted by means of DALI26 and VAST27 algorithms. The VIP3B2160 monomer is comprised of five domains, as follows: the N‐terminal 331 residues fold into two α‐helical domains (DI: residues 47–120 and DII: 22–46 and 121–331) connected by a twisted 70‐residue contiguous helix (amino acids 88–158) whereas the C‐terminal 450 residues form three discrete β‐sheet domains (DIII: 332–538, DIV: 543–668, and DV: 675–785). Even though domains DIV and DV share only ~20% sequence identity they are clearly from the same structural family (C_α rmsd = 4.8 Å).

Figure 1.

(a) Monomer of VIP3B2160 with domains DI…DV labeled and colored individually. (b) VIP3B2160 DII (top) versus Domain I of Cry1Ac from RCSB ID 4W8J (bottom). (c) VIP3B2160 DIII (top) versus Domain II of Cry4Aa from RCSB ID 2C9A (bottom). (d) VIP3B2160 DV (top) versus carbohydrate‐binding domain of the Caldanaerobius S‐layer endoglucanase from RCSB ID 2ZEW (bottom)

The two helical domains of VIP3B2160 share superficial similarity with helical bundles found in miscellaneous unrelated proteins such as a yeast transport protein (RCSB ID 3ETV), a Lon‐like protease from thermophilic bacterium (RCSB ID 4FWV), a human mitochondrial RNAse (RCSB ID 4XGL) and many others—which is not surprising since helical bundles are ubiquitous protein building blocks. There is at least conceptual similarity between DII and Domain I of the three‐domain CRY (3dCry) insecticidal toxin family (Figure 1b): both folds contain a central helix surrounded by several (five for 3dCry and four for VIP3B2160) relatively straight helices and the sixth broken/bipartite helix. Domain DIII has a few distant relatives – notably it is like Domain II of Cry4Aa (RCSB ID 2C9K), Domain II of Cry1Ac (RCSB ID 4w8j) (Figure 1c), and the beta‐sheet domain of Vibrio hemolysin (RCSB ID 1XEZ). The β‐sheet domains DIV and DV show meaningful similarity to the beta‐roll superfamily of carbohydrate‐binding modules from enzymes such as Caldanaerobius S‐layer endoglucanase (RCSB ID 2ZEW) (Figure 1d), Streptococcus hyaluronate lyase (RCSB ID 4D0Q), Streptococcus TIGR4 O‐glycoside hydrolase (RCSB ID 5A5A), or Cellulomonas b‐1,4‐glucanase C (RCSB ID 1CX1) with the pairwise C_α rmsd of 2.2–3.0 Å.

The five domains of VIP3B2160 come together to form a novel fold with several unique features—however, perhaps the most intriguing feature of this protein is its quaternary assembly. The tetramer of VIP3B2160 resembles a Galaga^© spaceship with a pointed “nose” (DI) and a wide “body” (DII/DIII) supported by four “fins” (DIV/DV) (Figure 2a,b). Despite the initial impression that the oligomeric structure of VIP3B2160 is a fourfold symmetrical tetramer; the protein is in fact a dimer of dimers composed of two pairs of conformationally distinct monomers. Geometric inequivalence between monomers is generated by ~45° rotation of DI with respect to the rest of the protein (Figure 3a). Subunits A and B are equivalent (C_α rmsd = 0.59 Å)—so are subunits C and D (C_α rmsd = 0.62 Å); however, the two pairs are radically different from one another by the orientation of DI and subtly different by the orientation of individual helices and loops within DII. With the exception of DI the four monomers align well (average C_α rmsd of 0.7 Å). The four monomers are arranged in a peculiar pattern: the “nose” of the VIP3B2160 tetramer consists of three locally twofold‐symmetrical interacting DI domains arranged in a straight line: C–B–A–D, such that DIs of subunits C and D do not interact at all (Figure 3b). This arrangement is utterly different at the “body” end of the tetramer (domains DII) where subunits C and D come into close contact and A and B are completely separated (Figure 3c) so that the circle A–D–B–C–A with a cross‐contact C–D describes the pseudo‐fourfold topology of subunits within the “body” of the tetramer.

Figure 2.

(a) Side view of the VIP3B2160 tetramer. (b) Bottom view of the tetramer

Figure 3.

(a) Rotation of DI in VIP3B2160 monomers A and B (green) versus monomers C and D (magenta). (b) Linear arrangement of DI domains at the top of the tetramer. (c) Circular arrangement of DII domains at the bottom of the tetramer

In 2017, Palma and co‐workers11 published the first foray into structural biology of VIP3 insecticidal proteins—a transmission electron microscopy (TEM) reconstruction of VIP3A tetramer before and after trypsin treatment. Although the sequences of VIP3B2160 used in this work and of VIP3Ag4 used by Palma et al. are only 64% identical, the major structural features of the two proteins are clearly preserved—as can be seen from alignment (Figure S1) and by visual comparison of the TEM class averages (Fig. 6 of the Palma et al. work11) with a bottom‐side projection of VIP3B2160 structure (Figure 2b of the present article). Locations of the N‐termini as indicated by Ni‐NTA nanogold particle attachment are also in agreement between the TEM and crystallographic observations. Even the tryptic cleavage site of VIP3A (Figure S1) is a mere five residues upstream of that of VIP3B2160 which further underscores the fundamental structural and functional similarity of these two proteins. There also are some remarkable differences—the most significant one being the absence of the sharp “nose” observed for VIP3B2160. Given that the reconstruction of VIP3A tetramer was rendered at its full 380 kDa volume, the missing sections must somehow be distributed within the rendered volume ‐ perhaps because of substantial conformational rearrangement. Even though the tryptic core structure of VIP3B2160 in its present state does not merit publication (this structure will be deposited with RCSB separately), it is important to note that without the N‐terminal 21‐kDa segment the protein does not assemble into tetramers, instead preferring a dimeric form (not shown). Functional relevance of this form is dubious since the protein expressed as pure tryptic core (i.e., it never had the 21‐kDa fragment—as opposed to being expressed as full‐length protein, then trypsinized) was not biologically active in our hands. In contrast, both the full‐length VIP3B2160 and the in vitro trypsinized VIP3B2160 were highly active on several Lepidopteran species (data to be published separately).

The structure of VIP3B2160 provides a fertile ground for speculation regarding binding, activation, and pore formation functions of this fascinating protein. Given the structural similarity between DIV/DV and various carbohydrate‐binding modules it is very likely that VIP3B2160 binds to glycosylated receptors in the insect midgut membranes (and by extension VIP3‐family proteins are likely to have glycosylated receptors in general). If true, this is a direct parallel to the mode of action proposed for the three‐domain Cry insecticidal proteins and to the tentative conclusions reached by the authors of previously published VIP3 studies.6, 10, 28, 29, 30, 31 It also seems likely that protease activation is a universal feature of the VIP3 family, given the feature conservation present in the N‐terminal ~240 residues—including the sequence‐divergent region of putative proteolytic processing sandwiched between two highly conserved islands corresponding to the DII helical elements (Figure S1). It is not clear how the activation signal is transmitted from the recognition elements (DIV and DV) to the pore‐forming elements (currently unknown) ‐ however, it is noteworthy that DIII resembles Domain II of the 3dCry toxins (Figure 1c) and DII is feature‐similar to Domain I of the same (Figure 1b). Perhaps this is another example of convergent evolution that drives critical portions of the otherwise radically different 3dCry and VIP3 protein families toward a (poorly understood) common function. It is equally interesting to speculate whether the pore‐forming elements of VIP3 toxins reside in the 21‐kDa fragment (the tip of the “rocket”) or whether they derive from DII or DIII (the authors suspect that DII is the most likely origin of pore‐forming residues) and whether the pore is formed when the base or the tip of the tetramer is aligned toward the cell membrane.

In conclusion, the structure of full length VIP3B2160 answers several key questions posed by prior research into this fascinating family of insecticidal toxins, yet it generates many additional questions that need to be addressed. Given time the VIP3 molecular mode of action will be illuminated in greater detail—and we hope that the present structure is instrumental in furthering such understanding.

CONFLICT OF INTEREST

Authors would like to declare that this work has been performed as a part of a broader commercial interest into ongoing studies of insecticidal proteins.

STATEMENT OF BROAD INTEREST

Bacillus thuringiensis toxins are effective biological insecticides that fall into distinct superfamilies with diverse sequences, properties, and modes of action. Vegetative insecticidal proteins (VIPs) of the VIP3 subfamily have emerged as an important tool to counter emergent insect resistance toward the insecticidal proteins deployed in first‐generation transgenic traits. We report the structure of a representative VIP3 protein which reveals the unusual fold and arrangement of domains, which sheds light on the functional relevance of various VIP3 sequence features.

Supporting information

Data S1 Supporting Information.

Zheng M, Evdokimov AG, Moshiri F, Lowder C, Haas J. Crystal structure of a Vip3B family insecticidal protein reveals a new fold and a unique tetrameric assembly. Protein Science. 2020;29:824–829. 10.1002/pro.3803