Structural and biochemical insights into the catalytic mechanisms of two insect chitin deacetylases of the carbohydrate esterase 4 family

Insect chitin deacetylases (CDAs) catalyze the removal of acetyl groups from chitin and modify this polymer during its synthesis and reorganization. CDAs are essential for insect survival and therefore represent promising targets for insecticide development. However, the structural and biochemical characteristics of insect CDAs have remained elusive. Here, we report the crystal structures of two insect CDAs from the silk moth Bombyx mori: BmCDA1, which may function in cuticle modification, and BmCDA8, which may act in modifying peritrophic membranes in the midgut. Both enzymes belong to the carbohydrate esterase 4 (CE4) family. Comparing their overall structures at 1.98–2.4 Å resolution with those from well-studied microbial CDAs, we found that two unique loop regions in BmCDA1 and BmCDA8 contribute to the distinct architecture of their substrate-binding clefts. These comparisons revealed that both BmCDA1 and BmCDA8 possess a much longer and wider substrate-binding cleft with a very open active site in the center than the microbial CDAs, including VcCDA from Vibrio cholerae and ArCE4A from Arthrobacter species AW19M34-1. Biochemical analyses indicated that BmCDA8 is an active enzyme that requires its substrates to occupy subsites 0, +1, and +2 for catalysis. In contrast, BmCDA1 also required accessory proteins for catalysis. To the best of our knowledge, our work is the first to unveil the structural and biochemical features of insect proteins belonging to the CE4 family.

To date, six crystal structures of CDAs from fungi and bacteria have been determined, including ClCDA from Colletotrichum lindemuthianum (14), AnCDA from Aspergillus nidulans (15), ArCE4A from Arthrobacter species AW19M34-1 (16), two chitooligosaccharide deacetylases VcCDA from Vibrio cholerae (17), VpCDA from Vibrio parahemeolyticus (18), and one putative CDA (EcCDA) from Encephalitozoon cuniculi (19). Except the structure of EcCDA, which is not capable of deacetylating chitin, the other structures exhibit conserved critical residues in the active site, indicating that all CDAs use the same metalassisted general acid/base catalytic mechanism conserved across CE4 enzymes (14 -18, 20 -28). However, the large discrepancies in the shape of the substrate-binding site among these enzymes lead to varied substrate preferences and deacetylation modes. The crystal structures of VcCDA and ArCE4A represent the only two known CDA structures resolved in the presence of their oligosaccharide substrates (16,17). The VcCDA-substrate complex structures provide evidence that six critical loops adopt conformational changes to effectively trap chitooligosaccharides in the substrate-binding pocket. VpCDA has a substrate-binding pocket nearly identical to that of VcCDA and is highly active toward (GlcNAc) 2 (18,29). In contrast, there are marked differences in the loops surrounding the substrate-binding site in ArCE4A, ClCDA, and AnCDA (16,17). All of them adopt shorter loops to form a relatively open substrate-binding cleft and are active toward both chitooligosac-charide and chito-polysaccharide substrates. Sequence alignment indicates that insect CDAs do not share the specific loops observed in other CDAs (30,31). Unfortunately, the structures of insect CDAs have long been pursued without success.
Insects possess a greater number of CDAs than any other organism. An insect may possess as many as five groups of genes encoding CDAs (32), which may function in the epidermis, tracheal tubes, and midgut (33)(34)(35)(36). Previous research suggested that chitosan was present in the precise locations where the flexibility of chitin fibers was required (37). Two cda gene mutants of Drosophila melanogaster embryos resulted in elongated and tortuous tracheal tubes (34,35). RNAi of all nine cda genes from Tribolium castaneum resulted in abnormal phenotypes in the tracheal tubes and cuticle and in joint defects, followed by molting failure or even death (33). RNAi of LmCDA2 from Locusta migratoria changed the chitin organization in the procuticle from a helicoidal to a unidirectional orientation (38). However, many efforts have failed to demonstrate the deacetylation activities of insect CDAs toward chitinous substrates in vitro (30,39). Data about the biochemical characteristics and structure-function relationship of insect CDAs remain scarce.
In this study, two CDAs from Bombyx mori, BmCDA1 and BmCDA8, were crystallized and resolved, providing the first insight into the structural characteristics, activity profiles, and deacetylation mode of insect CDAs. The structures revealed that insect CDAs possess several unique structural features that distinguish them from other known CE4 enzymes. This work will also assist the development of specific agrochemicals for pest control.

Overall structure of BmCDA1-CAD
BmCDA1 is composed of an N-terminal signal region (residues 1-23), a chitin-binding domain (residues 24 -122), a low-density lipoprotein receptor domain (residues 123-161), and a catalytic domain (CAD, residues 162-539) (Fig.  1A). Our experiments showed that full-length recombinant BmCDA1 was not stable and underwent autocleavage when incubated with the crystallization reagent. Thus, the truncated form of BmCDA1 (BmCDA1-CAD, residues 162-539) was cloned, expressed, and purified for crystallization. Diffraction data were collected to 1.98 and 2.4 Å on native and The key residues in the five motifs are shown in stick representation using the same color scheme. Structure-based sequence alignment of five CDAs showing the conserved motifs 1-5 (right panel). The highly conserved residues are indicated.

Structure and activity of two insect chitin deacetylases
SeMet protein crystals, respectively, and the structure was solved using the SeMet single-wavelength anomalous diffraction technique.
BmCDA1-CAD was crystallized in the trigonal space group I4 with one molecule in the asymmetric unit ( Table 1). The overall structure of BmCDA1-CAD consists of two regions: a CE4-conserved NodB homology subdomain and C-terminal loops (Fig. 1A). The NodB homology domain is a (␤/␣) 7 barrel (residues 162-471) composed of seven parallel ␤-strands arranged in a barrel that is surrounded by six ␣-helices. Notably, the (␤/␣) 7 barrel contains one loop insertion (residues 331-402) between ␤ 5 and ␣ 5 . The C-terminal loops (residues 472-539) consist of ␣ 8 , a pair of antiparallel ␤-strands (␤ 9 and ␤ 10 ), and several loops. These structural elements, the loop insertion of the (␤/␣) 7 barrel and the C-terminal loops, appear to be unique because they are not present in any of the other CE4 structures determined to date.

Active site and substrate-binding cleft of BmCDA1-CAD
The active site of BmCDA1-CAD is located at the top center of the (␤/␣) 7 barrel and contains a metal-binding triad conserved across the CE4 family, namely, a zinc ion coordinated by Motif 5 (YFH) is also less conserved because motif 5 of most CDAs contains LXH. The canonical Leu contributes to form a hydrophobic patch (21). Taken together, the differences in motif 4 and motif 5 confer BmCDA1 a more open and wider active pocket.
A long substrate-binding cleft was observed on the surface of BmCDA1-CAD (Fig. 2C), with only one solvent-exposed hydrophobic residue (Tyr 242 ) that might aid in substrate binding. The substrate-binding cleft is shallow and open when compared with the other microbial CDAs because of the lack of several loops that give microbial CDAs their characteristically deep and narrow substrate-binding cleft (Fig. S1).

Crystal structure of BmCDA8
BmCDA8 (residues 19 -381) lacking the N-terminal signal peptide was expressed, purified, and crystallized. Diffraction data were collected to 2.4 Å. The structure was resolved by molecular replacement with BmCDA1-CAD as the search model. BmCDA8 was crystallized in the trigonal space group P3 2 21 with one molecule in the asymmetric unit (Table 1). Residues 19 -22 were not included in the final structure because of a lack of interpretable electron density. BmCDA8 showed 37% sequence identity with BmCDA1-CAD. The overall architecture of BmCDA8 was similar to that of BmCDA1-CAD ( Fig.  2A), corresponding to an root-mean-square deviation (RMSD) of 1.31 Å for 345 equivalent C ␣ atoms.

Structure and activity of two insect chitin deacetylases
Similar to BmCDA1, BmCDA8 (residues 23-381) consists of two regions: a CE4-conserved NodB homology subdomain and C-terminal loops. The active site of BmCDA8 is located at the top center of the (␤/␣) 7 barrel ( Fig. 2A). The conserved metalbinding triad is Asp 63 -His 117 -His 121 coordinated with a zinc ion, and the active site is shaped by motifs 1-5 (Figs. 1B and 2A).
However, two differences between BmCDA8 and BmCDA1-CAD were observed. One obvious difference is the replacement of Ala 392 (BmCDA1) by Ser 241 (BmCDA8) in motif 4. The other difference is the substrate-binding cleft. Unlike BmCDA1-CAD, BmCDA8 contains a narrower and deeper substratebinding cleft that passes through the active site where the cat-alytic reaction occurs (Fig. 2, B and C). As shown in Fig. 2B, this cleft has an extended structure with two open ends. One end is shaped by the loop insertion. In detail, surface-exposed Phe 238 , Phe 247 , and Phe 248 form a hydrophobic "claw" at the top. The other end is shaped by the C-terminal domain that provides two aromatic residues, Phe 347 and Tyr 361 .

Enzymatic activity and the deacetylation mode of BmCDA8
The activity assay indicated that BmCDA8 instead of BmCDA1 is active. To investigate the catalytic characteristics of BmCDA8, the enzymatic activity was determined using various kinds of chitinous substrates, including (GlcNAc) 1-6 , and the polymeric In BmCDA8, the surfaces are colored slate (active site), deep blue (residues from the C-terminal loops), hot pink (residues from the loop insertion), and white (other regions). The surface-exposed aromatic residues that line in BmCDA8 and BmCDA1 are shown in stick representation in black. The structures of the substrate-binding clefts are each viewed from two angles rotated 90°along the vertical axes.

Structure and activity of two insect chitin deacetylases
substrates ethylene glycol chitin and colloidal chitin. Among oligomeric chitinous substrates, BmCDA8 showed no activity toward (GlcNAc) 1-2 but did exhibit activities toward (GlcNAc) [3][4][5][6] . As for (GlcNAc) 3-6 , the k cat /K m values of BmCDA8 increased as the degree of polymerization increased. In contrast to its relatively low affinities toward chitooligosaccharides, BmCDA8 showed higher affinities toward the polymeric substrates ethylene glycol chitin and colloidal chitin ( Table 2).
The deacetylation mode of BmCDA8 was investigated by a two-step analysis of the deacetylated products of (GlcNAc) 3 (Fig. 3A). In the first step, electrospray ionization (ESI)-MS analysis was performed to determine the number of deacetylated GlcNAc residues. In the second step, ESI-MS analysis was performed again to determine the deacetylation sites. Before the ESI-MS analysis, the deacetylated products were pretreated with the enzyme Ostrinia furnacalis hexosamini-dase1 (OfHex1), which specifically cleaves ␤-1,4 -linked GlcNAcs instead of GlcNs from the nonreducing ends of the chitooligosaccharides. Thus, only chitooligosaccharides with a GlcN at the nonreducing end remained for the second ESI-MS analysis. For the substrate (GlcNAc) 3 , the first-step ESI-MS analysis showed that the products included (GlcNAc) 3 and a monodeacetylated product with a mass loss of 42.0, the mass of a CH 3 CO-H group. The monodeacetylated product was not further hydrolyzed by OfHex1 in the second step, indicating that the product was GlcN-GlcNAc-GlcNAc (Fig. 3B). Notably, neither GlcN nor monodeacetylated products of (GlcNAc) 2 were present, indicating that deacetylation occurred at the first GlcNAc at the nonreducing end. Because the biochemical data indicated that BmCDA8 was not able to deacetylate GlcNAc or (GlcNAc) 2 (Table 2), we deduced that BmCDA8 activity requires substrates to occupy subsites 0, ϩ1, and ϩ2 (Fig. 3C), where 0 is the catalytic site, and the plus sign refers to the reducing ends, according to the nomenclature commonly used for CE4 enzymes (40 -42).
To further confirm the requirement of the subsites for catalysis, mutagenesis of Gln 125 at ϩ1 subsite and Ser 241 at ϩ2 subsite was performed. The two substrate-binding sites were predicted by molecular dynamics stimulations (supporting information, Fig. S4). The activity of the mutants toward (GlcNAc) 3 was determined. As shown in Fig. 3D, both Q125A and S241A showed markedly lower specific activity than the WT protein. The mutation of Ser 241 caused much more serious impairment of activity than the mutation of Gln 125 , even though Gln 125 interacts with both subsites ϩ1 and ϩ2. This difference suggests that subsite ϩ2 might be more crucial for (GlcNAc) 3 binding.

Enzymatic activity of BmCDA1 requires accessary proteins
Our study showed that the deacetylation activity of BmCDA1 was undetectable toward various chitinous substrates even with prolonged incubation time to 120 h or with higher enzyme concentrations at 200 M. To understand the activation mechanism of BmCDA1, we added molting fluid (MF), which is a protein mixture secreted by insect epidermal cells that facilitates old cuticle shedding, into the reaction mixture. Strikingly, the activity of BmCDA1 toward ethylene glycol chitin and colloidal chitin was boosted in the presence of MF when compared with that of BmCDA1 alone and the catalytic residue-mutated form D205S (Fig. 4A). To further understand what proteins are involved in the activation of BmCDA1, the cuticular chitinbinding protein CPAP-3A1, which is a homolog of obstructor A in Drosophila that physically interacts with Serpentine (the CDA1 homolog in Drosophila) (43), was mixed with an equimolar amount of BmCDA1. The enzymatic activity assay showed a significant increase in the deacetylation activity of BmCDA1 in the presence of CPAP3-A1 (Fig. 4B). The in vitro pulldown assay illustrated that CPAP3-A1 can pull down BmCDA1 (Fig. S2). However, CPAP3-D, which belongs to the same CPAP family as CPAP3-A1 and shows the highest binding affinity to deacetylated chitin among CPAP family members (44), could not activate nor pull down BmCDA1 (data not shown). Taken together, these data indicated that BmCDA1 requires specific accessary proteins to achieve activity.

Discussion
This study on BmCDA1 and BmCDA8 provides the first structural and biochemical comparisons of insect CDAs. As revealed by chemical and spectroscopic analyses, chitin in insects is deacetylated at a degree of 5-25% (45). The deacetylation activity of insect CDAs appears to be necessary because abrogation of its activity by gene knockdown results in chitinous laminar organization disorder or even lethality (38,46). Surprisingly, most studies indicate that insect CDAs are ineffective enzymes in in vitro testing assays.
Insect CDAs seem to be designed less active. The deacetylated degree of the insect chitin matrix (5-25% chitosan) was relatively low when compared with that of the fungal cell wall (ϳ75% chitosan in Mucor rouxii) (47). MrCDA from M. rouxii was highly active toward chitinous substrates (4). The specific activity of MrCDA toward (GlcNAc) 5 was 467 M/min/M (4), which was 65.7-fold higher than that of BmCDA8. The weak activity of insect CDAs might help to maintain the low degrees of chitin deacetylation observed in vivo. Structural alignment of BmCDA1/BmCDA8 with other CE4 enzymes including peptidoglycan deacetylases, poly-␤-1,6-N-acety-D-glucosamine deacetylases, acetylxylan de-O-acetylases, chitin deacetylases, and chitin oligosaccharide deacetylases shows strong conservation of the active site (Fig. S3). The similarity of the active sites suggests that insect CDAs use the same catalytic mechanism. However, the different architectures of the substrate-binding clefts of the CE4 CDAs confer different catalytic

Structure and activity of two insect chitin deacetylases
properties to these enzymes. The narrow entrance of the substrate-binding pocket of VcCDA is composed of six loops close to each other, contributing a specific and high catalytic efficiency toward chitooligosaccharides (17). The openness and shortness of the substrate-binding cleft of ArCE4A represents a common feature of other CE4 enzymes, conferring them with higher rates for various substrates (16). Unlike VcCDA and ArCE4A, the only two known structures complexed with chitooligosaccharide substrates, both BmCDA1 and BmCDA8 possess a much longer, wider and more open substrate-binding cleft (Fig. 2, B and C). The lack of steric constraints within the substrate-binding clefts may reduce the effectiveness of trapping substrates once bound, perhaps explaining the weak activity observed for BmCDA1 and BmCDA8. The longer substrate-binding cleft formed by unique loops appears to be an intentional feature of insect CDAs to fit chitin fibers in vivo. To illuminate the functions of the unique loops in insect CDAs, we have constructed nine truncates of BmCDA8, each of which lacks partial or whole loops (supporting information, Fig. S5). Unfortunately, we failed to obtain any recombinant proteins. The function of loops unique in insect CDAs requires further investigation. The activation of BmCDA1 in the presence of molting fluid suggests that the activity of insect CDAs could be regulated by an unknown mechanism. As a component of molting fluid, the chitin-binding protein CPAP3-A1 instead of CPAP3-D1 was capable of activating BmCDA1, suggesting that the activation mechanism is complex. A possibly similar case has been shown for PgaB, a poly-␤-1,6-N-acety-D-glucosamine deacetylase from Escherichia coli. PgaB is composed of two domains: an N-terminal deacetylase domain and a C-terminal GH18/ GH20-like domain, the association of which is proposed to create a substrate-binding cleft during de-N-acetylation (49). In Figure 3. Structural analysis of deacetylated products of (GlcNAc) 3 . A, scheme of the two-step analysis method. B, ESI-TOF MS spectra of the deacetylated products of (GlcNAc) 3 . C, schematic representations of the binding mode of (GlcNAc) 3 to BmCDA8. E, GlcNAc residue; •, GlcN residue. D, specific activities of the WT protein and mutants of BmCDA8 toward (GlcNAc) 3 .

Structure and activity of two insect chitin deacetylases
Drosophila, the CPAP3-A1 homolog obstructor A chitin-binding protein interacts with the deacetylation domain protein Serp (CDA1) (43). We further verified the interaction of BmCDA1 and CPAP3-A1 using an in vitro pulldown assay. One may deduce the activation of BmCDA1 requires an accessary factor, e.g. CPAP3-A1. Future structural studies of the complex will provide information about the activation mechanism.
Taken together, the structural and biochemical data provide insights into the novel characteristics of insect CDAs. The lack of available and clear information regarding insect CE4 enzymes highlights the importance of the addition of the BmCDA1-CAD and BmCDA8 structure to the CE4 structural database, adding new knowledge about the CE4 family.

Gene cloning and expression plasmid construction
Total RNA was extracted from five B. mori specimens at the fifth instar (day 5) using RNAiso TM Plus (TaKaRa, Japan) according to the manufacturer's protocol. The cDNA was synthesized using the PrimeScript TM RT reagent kit (TaKaRa, Japan). The gene encoding BmCDA1-CAD and BmCDA8 was amplified from the cDNA with the primers listed in Table S1. A His 6 affinity tag was introduced at the C-terminal. The product was ligated into the EcoRI and XhoI restriction sites of the pPIC9 expression vector using the In-Fusion kit (Clontech). The resulting expression plasmids were subsequently linearized with the restriction enzyme SalI to allow integration into the chromosomal DNA of Pichia pastoris GS115 (Invitrogen).

Expression and purification
Recombinant P. pastoris was first grown in buffered complex medium containing glycerol (BMGY; Invitrogen) at 301 K to an optical cell density of 4.0 at 600 nm. The cells were collected by centrifugation, resuspended in buffered methanol complex medium (BMMY; Invitrogen), and transferred into a 5-liter fermentation tank. The volume of cultures for production of recombinant proteins is 3 liters. The pH was controlled with a sterilized base solution of 1 M KOH. Protein production was induced by delivering methanol to the vessel at a constant feed rate. The fermentation proceeded for 92 h at 301 K. The culture supernatant was obtained by centrifugation. The supernatant was subjected to ammonium sulfate precipitation with 75% saturation at 277 K for 12 h. After centrifugation, the supernatant was removed, and the precipitate was resuspended in distilled water and then desalted in buffer A (20 mM sodium phosphate, 0.5 M sodium chloride, pH 7.4) using a HiTrap desalting column (5 ml; GE Healthcare). The resulting sample was then loaded into a HisTrap HP affinity column (5 ml; GE Healthcare) equilibrated in buffer A. The target protein was eluted with 20 mM sodium phosphate, 0.5 M NaCl, 250 mM imidazole (pH 7.4). The eluted protein was Ͼ95% pure, as analyzed by SDS-PAGE. The purified protein was then desalted in 20 mM Tris (pH 7.4) and 20 mM NaCl and concentrated to an appropriate concentration for the crystallization experiments. The yields for the recombinant proteins are 0.2 mg/liter (BmCDA8), 1 mg/liter (BmCDA1), and 5 mg/liter (BmCDA1-CAD). Mutations were introduced by PCR-based site-directed mutagenesis, and the mutated proteins were purified using the same protocol described above.

Expression of selenomethionine-containing BmCDA1-CAD
The same strain was used for the expression of SeMet-incorporated BmCDA1-CAD. The growth conditions before induction were the same as those described above. The cells were then washed three times with PBS and resuspended in buffered methanol medium with the following modifications. The medium contained 100 mM potassium phosphate at pH 6.0, 1.34% (w/v) yeast nitrogen base without amino acids (Invitrogen), 0.09 mg/ml adenine sulfate, 0.09 mg/ml uracil, 0.34 mg/ml thiamine, 0.3 mg/ml succinic acid, 0.01 mg/ml inositol, 0.09 mg/ml L-tryptophan, 0.09 mg/ml L-histidine, 0.09 mg/ml L-arginine, 0.09 mg/ml L-tyrosine, 0.09 mg/ml L-leucine, 0.09 mg/ml L-isoleucine, 0.09 mg/ml L-lysine, 0.15 mg/ml L-phenylalanine, 0.3 mg/ml L-glutamic acid, 0.3 mg/ml L-aspartic acid, 0.45 mg/ml L-valine, 0.6 mg/ml L-threonine, 1.2 mg/ml L-serine, 0.12 mg/ml L-cysteine, 0.3 mg/ml L-glutamine, 0.2 mg/ml L-proline, 0.2 mg/ml L-alanine, and 0.1 mg/ml selenomethionine. The induction of the expression and purification procedures were the same as those described above.

Enzymatic activity assays
The enzyme activity was determined based on the detection of acetate released by the action of BmCDA8. The assays were performed using the K-ACETRM kit (Megazyme International, Wicklow, Ireland) according to the manufacturer's instructions. A deacetylation activity screen was performed using different chitinous substrates, including both polymeric substrates (ethylene glycol chitin (Wako Pure Chemicals), colloidal chitin, and ␣-chitin (Sigma-Aldrich)) and oligomeric substrates (monomer to hexamer; Qingdao BZ Oligo Biotech Co., Ltd.). The conditions of the enzymatic assay on chitin were as follows: 100 l of a mixture containing 6 M enzyme and 0.5 mg of the polymeric substrate in 20 mM Tris and 20 mM NaCl (pH 7.4) was incubated at 30°C for 30 min. The reaction was terminated by boiling at 100°C for 1 min. The amount of acetic acid was determined using the K-ACETRM kit. For oligomeric substrates, Michaelis-Menten parameters were determined. The reaction components were incubated in a final volume of 100 l at 30°C for 2 h in the presence of 20 mM Tris (pH 7.4) and 20 mM NaCl, 6 M enzyme, and 1-10 mM (GlcNAc) 1-6 . Then enzyme reaction was stopped by boiling at 100°C for 1 min, and the concentration of released acetic acid was then measured by the K-ACETRM kit according to the manufacturer's protocol. Data analysis was performed with OriginPro 8.5 (OriginLab).

BmCDA1 activation reactions
To test the activation of BmCDA1 by molting fluid and CPAP3-A1, the recombinant BmCDA1 protein (100 g) was incubated with 100 g of molting fluid and CPAP3-A1, respectively, in a total volume of 50 l in buffer containing 20 mM Tris and 20 mM NaCl (pH 7.4) at 30°C for multiple time intervals as followed: 0, 2, 5, 10, and 24 h. Ethylene glycol chitin and colloidal chitin were used as substrates to assess the activation of BmCDA1 by molting fluid and CPAP3-A1. The molting fluid was extracted as mentioned by Qu et al. (50). CPAP3-A1 was recombinantly expressed and purified as previously described (44).

ESI-MS analysis
For this experiment, 10 mM (GlcNAc) 3 was treated with 10 M BmCDA8 in 20 mM Tris (pH 7.4) and 20 mM NaCl at 30°C for 48 h. The samples were boiled at 100°C for 2 min and then centrifuged at 17,000 ϫ g for 10 min. Purified OfHex1 (51) was added to the samples to a final concentration of 10 M. The resulting solution was incubated at 30°C for 48 h. NaN 3 was added to all the samples at a final concentration of 0.03% to prevent bacteria growth. The reactions were terminated by boiling for 2 min. The samples treated with and without OfHex1 were subjected to ESI-MS and recorded in both positive and negative mode using an Agilent 6224 TOF LC/MS system with a dual-nebulizer ESI source (Agilent Technology).

In vitro GST pulldown assays
GST and GST-CPAP3-A1 were constructed and expressed in E. coli strain BL21 (DE3). Pulldown assay was performed using BeyoGold TM GST-tag purification resin (Beyotime) according to the manufacturer's instructions. The GST proteins were incubated with 50 l of resins in 20 mM Tris (pH 8.0), 200 mM NaCl, and 0.5% (v/v) Nonidet P-40 for 2 h at 4°C. The resins were washed five times and then incubated with equal amounts of His-BmCDA1 overnight at 4°C. The beads were washed five times again, and the presence of His-BmCDA1 was detected by Western blotting using anti-BmCDA1 antibody.

Crystallization, data collection, and structure determination of BmCDA1-CAD
Crystals of native BmCDA1-CAD were grown by vapor-phase diffusion using the hanging-drop method with an equal volume (1 l) of protein (12 mg/ml in 20 mM Tris and 20 mM NaCl, pH 7.4) and reservoir solution (0.2 M sodium malonate and 20% PEG 3350 with a final pH 7.0) at 277 K. Crystals were harvested in rayon fiber loops and bathed in a solution containing reservoir solution and 25% (v/v) glycerol as a cryoprotectant prior to flash freezing in liquid nitrogen. Diffraction data were collected using BL18U of the National Facility for Protein Science Shanghai at the Shanghai Synchrotron Radiation Facility in China. The diffraction data were processed and integrated using the HKL-3000 package (52). Data processing statistics are given in Table 1.
The structure of BmCDA1-CAD could not be solved using any of the deposited CE4 structures as a search model in molecular replacement calculations. Phasing information was thus obtained using single-wavelength anomalous dispersion with selenium derivative data. The crystals of selenium-containing BmCDA1-CAD protein were harvested using the same procedure under the same conditions. Diffraction data were collected using BL17U at the Shanghai Synchrotron Radiation Facility in China (53). Detection of heavy atom sites, phasing, and density modification was performed with AutoSol (54) in Phenix (55). The initial structure was built using Autobuild (56) in Phenix. COOT (57) was used to make manual corrections to the model between further cycles of refinement using Phaser (58). The selenium-containing BmCDA1-CAD structure was used as the starting model for refinement of the native BmCDA1-CAD data. The native BmCDA1-CAD data were handled in the same way as the selenium-containing data. Final data and refinement statistics for the two structures are shown in Table 1.

Crystallization, data collection, and structure determination of BmCDA8
The crystallization conditions of BmCDA8 were screened by means of hanging-drop vapor diffusion in 96-well VDX plates at 277 K. The drop consisted of equal volumes (1 l) of protein (10 mg/ml) and reservoir solution. The crystallization condition (0.02 M calcium chloride dihydrate, 0.1 M sodium acetate trihydrate, pH 4.6, 30% v/v (Ϯ)-2-methyl-2,4-pentanediol) was obtained from Crystal Screen 2 (Hampton Research). Crystals of BmCDA8 were harvested in rayon fiber loops and flash frozen in liquid nitrogen. The diffraction data were collected on Beamline BL18U1 of the National Facility for Protein Science Shanghai at the Shanghai Synchrotron Radiation Facility in China. The data were processed and scaled using HKL3000 (52). Data analysis was performed using Phenix (55). The structure of BmCDA8 was resolved by molecular replacement with Phaser (58) using native BmCDA1-CAD structure as the model. The structure figures were created using the molecular visualization software PyMOL. Secondary structure topology cartoons were analyzed by Pro-origami (48). The structures were validated using the website https://validate-rcsb-1.wwpdb.org/. 3

MD simulation
Three independent MD simulations were performed using NAMD 2.10b2 with the force field charmm36. The systems were solvated with TIP3P water molecules, sodium ions were added to 0.15 M in water, and chloride ions were added to neutralize the system. An isothermal isobaric ensemble was employed at 310 K and 1 atm with a Langevin thermostat. Each system was equilibrated for 10 ns (the total energy was stable) with position restraints. Then 50-ns MD simulations without restraints were performed for each system. The RMSD value was calculated from protein C␣ atoms superimposed on the starting structure.