A New Nomenclature for Cry1Ab Proteins Reflecting 3-D Structure Differences

Cry1Ab proteins produced by the insecticidal bacterium Bacillus thuringiensis are mostly studied and applied, facing the challenge of insect resistance. The 3-D structure of the toxic core for all available 34 Cry1Ab proteins were constructed by the method of homology active against mosquito. The data obtained from the present in silico study provided new insights into structure-function relationship of Cry1Ab proteins.


INTRODUCTION
Bacillus thuringiensis, commonly known as Bt, is a Gram positive bacterium that occurs naturally in the soil around the world. For decades, bacteriologists have known that some strains of Bt kill certain insects and that the toxic substance responsible for the insects death is a protein generally referred to as parasporal crystal proteins (Cry proteins) [1]. These Cry proteins produced by Bt is widely used in biopesticide formulations and transgenic crops for insect control. The mode of action of Cry toxins is still a matter of investigation, generally, following ingestion by insects, they are activated by gut proteases and by binding to specific receptors on midgut epithelial cells [2]. Receptor binding induces the conformational change in the toxin necessary for membrane insertion, where it forms ion selective channels via oligomerization of toxin monomers, leading to cell lysis and finally to larval death [3,4].
Although Cry1Ab have been mostly studied and used, few efforts have been focused on the structural relationships among Cry1Ab members. To establish a new nomenclature which provide structural relationships among Cry proteins helping us to learn more about the relationship between structural differences and activities, the secondary structure and 3-D structure of all available 34 Cry1Ab proteins were constructed by the method of homology modeling, then different groups and subgroups were identified based on the secondary structure pattern and 3-D structure comparison.

3-D Structures of the Toxic Core of Available 34 Cry1Ab Proteins were Constructed by Homology Modeling
The structural models of the Cry1Ab toxic core were obtained comprising of 578 amino acids out of 1155 long primary structure, approximately. Sequence alignment showed an average of 88.5% identity between each Cry1Ab and Cry1Aa (PDB: 1CIY). All the 3-D structures of the toxic core of 34 available Cry1Ab proteins were constructed by homology modeling and evaluated by Ramachandran plot. Comparison of structures among the members of Cry toxin family revealed that Cry1Ab shares similar architecture with them and forming a wedge shape. The predicted structure of toxic core is comprised of three putative domains ( Fig. 1). A Ramachandran plot indicated that most (95 %) residues have φ and ψ angles in the core and allowed regions (Fig. 2).
In the absence of an experimentally determined structure, comparative or homology modeling can sometimes provide a useful 3-D model for a protein that is related to at least one known protein structure [26]. It is observed that a model tends to be reliable if identity percentage between the template and target protein is above 40%. Low degree of reliability arises when identity decreases below 20% [27]. All 34 Cry1Ab proteins shared more than 85 % identity between them and template protein (Cry1Aa), and the results of Ranachandran plot all surpass 95%, hence, the models constructed in this study are reliable.

The 34 Cry1Ab Proteins are Divided into Four Groups Based on the Secondary Structure Patterns of Toxic Core
The secondary structure pattern of the toxic core of the 34 Cry1Ab proteins was characterized based on the structural models (Table 1). Four different groups were identified and named as Cry1AbⅠ, Cry1AbⅡ, Cry1AbⅢ, and Cry1AbⅣ ( Cry1AbⅠ, Cry1AbⅡ and Cry1AbⅢ domainⅠ was composed of N-terminal 235 amino acid residues folded into a bundle of 8 amphipathic αhelices and 1 small β-strand. Cry1AbⅣ domainⅠ, however, was composed of N-terminal 234 amino acid residues. As with other Cry toxins, DomainⅡ of Cry1Ab consists of three Greek key β sheets arranged in β prism topology. It was composed of 205 amino acid residues, with 4 αhelices and 11 β strands in Cry1AbⅠand Cry1Ab Ⅳ, and 10 β strands in Cry1Ab Ⅲ . DomainⅡ of Cry1AbⅡ, however, consisted of 211 amino acid residues, with 5 α-helices and 8 β strands. Domain Ⅲ comprised residues 480-606 in Cry1AbⅠand Cry1Ab Ⅲ, 479-605 in Cry1Ab Ⅳ, and 489-607 in Cry1Ab Ⅱ. The amino acid residues of all Cry1Ab Domain Ⅲ are highly conserved.

The 31 Members of Cry1AbⅠ Ⅰ Ⅰ Ⅰ Group are Further Divided into Three Subgroups Based on 3-D Structural Comparison
The 3-D structural comparison among the 31 members of Cry1AbⅠindicated that 3-D structure of Cry1Ab31, Cry1Ab33 are different from the rest of Cry1AbⅠ ( Fig. 3 and Fig. 4). Thus, Cry1AbⅠcan be further divided into three subgroups, Cry1AbⅠ3 (Cry1Ab33 only), Cry1AbⅠ2 (Cry1Ab31 only), and Cry1AbⅠ1 (the rest of Cry1AbⅠ, Cry1Ab1 as the model) ( Table 2). The differences among the Cry1AbⅠ group were found in loops of domainⅡ. The loop β8-β9 in domainⅡof Cry1AbⅠ1 differ from that of Cry1AbⅠ2 resulted from residue 440 is Phe in Cry1AbⅠ1 vs Leu in Cry1AbⅠ2. The loop β4-β5 is different resulted from residue 369 is Arg in Cry1AbⅠ1 instead of Ser in Cry1AbⅠ3. Both of loops β4-β5 and β8-β9 are different between Cry1AbⅠ2 and Cry1AbⅠ3, which resulted from residue differences Arg369Ser and Leu440Phe. Cry1AbⅠ1 showed activities against Lepidoptera, include Noctuidae, Lymantridae, Sphingidae, Pyralidae, Pieridae, Plutellidae, Tortricidae, Chrysopidae and Lasiocampidae (Table 3).
Although the insecticidal activity data of Cry1AbⅠ2 and Cry1AbⅠ3 were not available, we speculated they had different insecticidal activites compared to Cry1AbⅠ1 because they differ in the receptor-binding loops in domainⅡ.
Cry1AbⅢ had the unique activity against mosquito (Diptera) while the other Cry1Ab were only active against Lepidoptera (Table 3). We speculated that Ala450Pro mutant of Cry1AbⅠ might gain the activity against mosquito because the mutant exhibited the same 3-D structure with Cry1AbⅢ.

Structural Differences between
There are 610 amino acids in the toxic core of Cry1AbⅠand 609 in Cry1AbⅣ. The additional residue, which was Trp in residue 182 in Cry1AbⅠand no corresponding residue in Cry1AbⅣ (Fig. 7), resulting in a shorter α6 in Cry1AbⅣ than that in Cry1AbⅠ (Fig. 8).

Structural
Differences between Cry1AbⅢ Ⅲ Ⅲ Ⅲ and Cry1AbⅣ Ⅳ Ⅳ Ⅳ There are 610 amino acids in the toxic core of Cry1AbⅢ and 609 in Cry1AbⅣ, the toxic core of Cry1AbⅢ was comprised of 13 α-helices, 21 βstands and turns, however, Cry1AbⅣ consisted of 13 α-helices, 22 β-stands and turns. Results of 3-D structure comparison revealed the absence of residue Trp in α6 and the additional of β9 (I446-R448) component in Cry1AbⅣ (Fig. 9). The amino acid sequence is different in 182, 450, 537, 545 and 568 according to sequence alignment of Cry1AbⅢ and Cry1AbⅣ (Fig. 10). Furthermore, the differences of residues 182 and 450 result in 3-D structural difference between them.
---lack of component, The components highlighted in red color represent the main differences between Cry1AbⅠ and other types proteins; those in green represent the main differences between Cry1AbⅡ and other types proteins; and those in blue represent the main differences between Cry1AbⅣ and other types proteins

Structural Differences between Cry1AbⅡ Ⅱ Ⅱ Ⅱ and the other three Groups
There are significant differences among Cry1AbⅡ and other groups from secondary structure and 3-D structure comparison (Fig. 11). A few of the components α1, α2, α6 and some loops differ in their locations in domainⅠ. The other differences among them in domainⅠ is in Cry1AbⅡ, the absence of β6, β7and β8 and the presence of additional α13 components in comparison with Cry1AbⅠand Cry1AbⅣ, whereas the absence of β6, β7and β9 and the presence of additional α13 components in comparison with Cry1AbⅢ, and a few of the components α11, α12, β9, β10, β11 and β12 differ in their locations in domainⅡ. Compared to other groups, Cry1AbⅡhave different locations of almost all components and the absence of β13 in domainⅢ. and 504-508 between Cry1AbⅡand Cry1AbⅣ, respectively (Fig. 12). a b c

Fig. 4. Structural comparison among the subgroups of Cry1AbI
The residues highlighted in red color represent helix; those in yellow represent stand and turn; and those in green represent coli and are generated using Pymol software. a the 3-D structural comparison between Cry1Ab1 and Cry1Ab31. b the 3-D structural comparison between Cry1Ab1 and Cry1Ab33. c the 3-D structural comparison between Cry1Ab31 and Cry1Ab33 a b

Fig. 5. Structural comparison between Cry1AbI and Cry1AbIII
The residues highighted in red color represent helix;those in yellow represent stand and turn;and those in green represent coli and are generated using Pymol software. a the secondary structural comparison between Cry1AbI and Cry1AbIII. b the 3-D structural comparison between Cry1AbI and Cry1AbIII. The difference between them is Cry1AbIII lack of one β-sheets in domainII.

DISCUSSION
As showed in this study, all 34 available Cry1Ab proteins, which were listed in the old nomenclature on the basis of their full-length amino acid sequences sharing at least 95% homologies, can be divided into four groups and subgroups based on toxic core structural differences. This change from a sequence-based to a structure-based nomenclature allows closely related proteins to be ranked together and provide researchers function information.
Domain Ⅰ of four Cry1Ab groups are similar (Table 1), which have 8 α-helices and is thought to be directly involved in membrane penetration and pore formation after binding to the specific receptor on surface of midgut [28]. In the pore formation model of Cry toxin action, binding to cadherin facilitates the proteolytic removal of domainⅠ α-1 promoting oligomer formation [29]. Nuria et al. showed the helix α-3 (the same to α-4 in this study) in domain Ⅰ that could form coiled-coli structures important for oligomerization [30]. In other reports the mutations Arg-93 and Ala-92 (located at the beginning of α-3) of Cry1Ab severely affected toxicity and correlated with loss of pore formation [31,32], and substitutions in residue Arg-99 also resulted in a complete loss of pore activity [33]. In addition, characterization of domainⅠ α-4 (the same to α-5 in this study) mutants revealed that in contrast to α-3 mutants described above, the point mutations in α-4 were able to form oligomeric structures [34]. Li et al. [8] suggested that the helical hairpin α4-α5 (the same to α5-α6 in this study) act as the initiator of the membrane related allosteric mechanism of penetration commonly known as umbrella model, and Thanate et al. [35] signified that the polarity at the α4-α5 loop residue Asn-166 was directly involved in ion permeation. All data revealed that domainⅠ was an essential component for poreformation so that the structure of four groups Cry1Ab of domain Ⅰ was conservative and exactly alike. It is possible that mutation aimed to an increase in these helices will improve the pore forming activity of Cry1Ab toxin.
The main differences among the four Cry1Ab groups in domainⅡ are the length and location of one of the two loops joining the apical βstands. Loop α9-α10 represent loop α8 in other papers and the other three receptor binding loops called loops 1, 2 and 3, respectively. In this study, loop α8, loop 1 (β2-β3) and loop 2 (β4-β5) are similar whereas loop 3 is different among the four groups. Loop  ) in Cry1Ab Ⅲ , respectively. Loops 2 and α8 of Cry1Ab are reported to have a binding affinity for M. sexta Bt receptor (BtR) [36], and loops 2 and 3 are reported to have a binding affinity for H. virescens BtR [37]. In addition, the mutations G439A and F440A significantly reduced toxicity toward M. sexta and H. virescens and in contrast, mutants S438A, S441A, N442A, and S443A were similar or only marginally less toxic to the insects compared to the wild-type toxin [38]. Both of Cry1AbⅠ and Cry1AbⅡ have activity against M. sexta and H. virescens (Table 2), so the difference of loop 3 between Cry1Ab Ⅰ and Cry1AbⅡ has no influence on activity. No data reveal Cry1Ab Ⅲ and Cry1Ab Ⅳ have activity against M. sexta and H. virescens, but the structure of Cry1AbⅣ is similar with Cry1AbⅠ, so it is possible that Cry1Ab Ⅳ have activity against M. sexta and H. virescens.
Domain Ⅲ , which consists of 2 β-sheets in a jellyroll conformation, has been implicated in determining specificity, then Cry1Ab Ⅱ have different structure of domainⅢ compare to other three groups. Swapping domain Ⅲ between toxins, such as Cry1Ab become more active against Spodoptera exigua when its domain Ⅲ was replaced by part of that of Cry1Ca [39], this result shows domain Ⅲ have influence on insecticidal activity. In addition, mutations in domainⅢ of Cry1Aa had an effect on both ion channel activity and membrane permeability [40]. DomainⅢ could play a role in protecting the toxin against further cleavage by gut proteases [41].

CONCLUSION
Based on the template of Cry1Aa, we have built 3-D structure models for all the current 34 Cry1Ab proteins. Based on the secondary structure pattern, four different groups were identified and named as Cry1AbⅠ, Cry1AbⅡ, Cry1Ab Ⅲ , and Cry1Ab Ⅳ . The three Cry1Ab proteins, Cry1Ab2, Cry1Ab7 and Cry1Ab28 were recognized as Cry1AbⅡ, Cry1AbⅢ, and Cry1Ab Ⅳ, respectively. The other 31 Cry1Ab proteins were grouped as Cry1Ab Ⅰ , and were further divided into three subgroups based on 3-D structural differences. The structural differences among different Cry1Ab groups and subgroups were analyzed in details. The insecticidal activities of different Cry1Ab groups and subgroups were also discussed. It was worthy to speculate that the only difference in 3-D structure, residues 447-449 form β-sheet in Cry1AbⅠ vs loop in Cry1AbⅢ, resulted in Cry1AbⅠ inactive vs Cry1AbⅢactive against mosquito. The results provided new insights into structure-function relationship of Cry1Ab proteins.