Carbohydrate-binding ability of a recombinant protein containing the DM9 motif from Drosophila melanogaster

Abstract

Proteins containing DM9 motifs, which were originally identified in the Drosophila melanogaster genome, are widely distributed in various organisms and are assumed to be involved in their innate immune response. In this study, we produced a recombinant protein of CG13321 (rCG13321) from D. melanogaster, which consists of four DM9 motifs, in Escherichia coli cells. In affinity chromatography using a mannose-immobilized column, rCG13321 exhibited mannose-binding ability and was separated into high-affinity and low-affinity fractions, named HA and LA, respectively, based on its binding ability to the column. In addition to having a higher affinity for the column, HA exhibited self-oligomerization ability, suggesting slight differences in tertiary structure. Both LA and HA showed hemagglutinating activity and were able to agglutinate an oligomannose-containing dendrimer, indicating that they have multiple carbohydrate-binding sites. Glycan array analysis suggested that rCG13321 primarily recognizes d-mannose and d-rhamnose through hydrogen bonding with the 2-, 3- and 4-hydroxy groups. Isothermal titration calorimetry demonstrated that rCG13321 has a comparable affinity to typical lectins. These findings suggest that CG13321 functions as a carbohydrate-binding protein or lectin that recognizes mannose and related carbohydrate-containing molecules on the surface of foreign organisms as a pattern recognition molecule.

Graphical Abstract

Open in new tabDownload slide

Abbreviations

 
• CD

  circular dichroism

 
• CRD

  carbohydrate-recognition domain

 
• DLS

  dynamic light scattering

 
• ITC

  isothermal titration calorimetry

 
• PD

  polyamidoamine dendrimer

 
• TBS

  Tris-buffered saline

Lectins are carbohydrate-binding proteins that were initially discovered as hemagglutinins in plant seeds. They have been found to be widely distributed in various organisms and play important roles in molecular and cellular recognition processes, such as cell adhesion, immune responses and molecular transport (1,). Animal lectins are categorized into different families based on the structure of their carbohydrate recognition domains (CRDs) (2,). C-type lectins, which require Ca²⁺ ions to bind carbohydrates, are one of the major families and exhibit diverse functions, especially innate immune systems (3,). For instance, mannose-binding lectin (MBL) and related collectins present in mammalian sera can activate the immune system, such as the complement system and phagocytosis, by binding to mannose-containing carbohydrate chains on the surface of bacteria and viruses (4,). Furthermore, numerous C-type lectins expressed on the surface of immune cells function as pattern-recognition receptors (PRRs) (5).

In invertebrates, the involvement of lectins, including C-type lectins, in immune systems has also been increasingly reported recently. Invertebrate lectins are particularly important for their innate immune systems due to the absence of the acquired immune system (6–9,). We have isolated a mannose-specific lectin CGL1 from the bivalve Pacific oyster Crassostrea gigas as a novel lectin that shows no similarities with any known lectin families (10,). This lectin is composed of two identical subunits of approximately 15 kDa each. The amino acid sequence of the CGL1 protomer consists of tandemly repeated motifs of approximately 70 residues, which are homologous to DM9 repeats found in the fruit fly D. melanogaster genome (11,). The lectin fold with DM9 motifs has not been found in other known lectin families. It has also been reported that multiple homologs of CGL1 (CgDM9CPs) are expressed in C. gigas, and they are suggested to be involved in innate immunity based on their ability to bind several pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide, peptidoglycan, mannan and β-1,3-glucan (12–15). This strongly suggests their involvement in the innate immune system to defend against pathogens.

DM9 domain proteins have been found in some insects and flatworms, such as Ascaris (16,), Anopheles gambiae (17,  18,), Fasciola (19–21,), Trichobilharzia szidati (22,), Opisthorchis (23,  24,) and Plasmodium (17,  20,). They are suggested to be involved in interactions between parasites and hosts based on observations that indicate their enhanced expression and/or intracellular relocalization after invasion by the parasites. Among these proteins, mannose-binding activity has only been reported for the DM9 protein from the Platyhelminthes Fasciola gigantica, which also exhibited hemagglutination and bacterial agglutination activities (21). This suggests that DM9 domain proteins may play a role in interactions with pathogens and parasites through binding to their mannose-containing carbohydrate chains. However, it remains unclear whether DM9 domain proteins generally function as lectins to mediate such cellular recognition processes.

In this study, we produced the recombinant CG13321, one of the DM9 domain proteins in D. melanogaster, which exhibits relatively high similarity to CGL1. The recombinant CG13321 (rCG13321) was purified using affinity chromatography, and its carbohydrate-binding ability was examined through hemagglutination assay, oligomannose-containing dendrimer-binding assay, glycan array analysis and isothermal titration calorimetry (ITC) analysis. Notably, the expressed rCG13321 was separated into two fractions that differed in their affinity for the mannose-immobilized column and tendency to self-oligomerize. These findings may have potential implications for the biological function of CG13321.

Materials and Methods

 

Materials

Fruit fly (D. melanogaster) specimens were obtained from a local retailer. Oligonucleotides were purchased from Integrated DNA Technologies (Tokyo, Japan). The plasmid vector pET-3a was obtained from Merck. The NucleoSpin RNA II kit, the PrimeScript RT-PCR Kit and the In-Fusion HD Cloning Kit were purchased from Takara (Otsu, Japan). E. coli C43(DE3) was obtained from Lucigen (Middleton, WI, USA). The mannose-immobilized Cellufine (mannose-Cellufine) column was prepared by attaching mannose to the Cellufine gels (JNC Corp., Tokyo, Japan) using divinyl sulfone (Sigma-Aldrich), as described previously (25). Polyamidoamine dendrimer (PD) (generation 4.0; MW 14,214) was obtained from Sigma-Aldrich. Cy3 NHS ester was obtained from Abcam (Cambridge, UK). All other chemicals used were of analytical grade for biochemical purposes.

Cloning of cDNA encoding CG13321 and its expression using E. Coli cells

Total RNA was extracted from fruit flies (D. melanogaster) using the NucleoSpin RNA II kit. The cDNA encoding CG13321 was amplified using the PrimeScript RT-PCR Kit with two primers (forward: AAGGAGATATACATATGGGAGACTACACCTGGATTAGCA and reverse: TTAGCAGCCGGATCCTCAGCCCTTGACCAGGACCTCGTAG) containing 15 bases of 5′-overlap with NdeI and BamHI restriction sites of the pET-3a vector. The amplified DNA was inserted into the pET-3a vector at the NdeI and BamHI sites using the In-Fusion HD Cloning Kit, and the sequence was confirmed with an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems, Waltham, USA). rCG13321 was expressed in E. coli C43(DE3) cells, induced with 0.4 mM isopropylthiogalactoside. The recombinant protein was obtained in both soluble and insoluble fractions.

Purification of rCG13321

The soluble proteins from the E. coli lysate were directly applied to a mannose-Cellufine column (10,) that had been pre-equilibrated with TBS (Tris-buffered saline; 10 mM Tris–HCl, pH 7.6, 150 mM NaCl). The column was then washed with the same buffer. The protein that adsorbed to the column was then eluted using 100 mM mannose in the same buffer. Fractions containing purified rCG13321 were combined and dialyzed against TBS and stored at −80°C until needed, as rCG13321 tends to form precipitation when stored at around −20°C. For N-terminal sequencing, the proteins separated by SDS-PAGE were transferred onto the polyvinylidene difluoride (PVDF) membrane in the transfer buffer (48 mM Tris, 39 mM glycine, 0.1% [w/v] SDS, 20% [v/v] methanol) for 40 min at 160 mA. SDS-PAGE was conducted in the presence of 0.1% (v/v) 2-mercaptoethanol. Protein bands were stained with Ponceau S solution (0.1% [w/v] Ponceau S in 5% [v/v] acetic acid), and their N-terminal amino acid sequences were analyzed using a protein sequencer PPSQ-21 (Shimadzu, Kyoto, Japan). The chemical and physical parameters of the proteins were calculated using the ProtParam tool in the ExPASy Bioinformatics Resource Portal (www.expasy.org). Multiple sequence alignments were performed using Clustal W (26,). The tertiary structure of CG13321 was predicted by AlphaFold2 (27).

Circular dichroism (CD)

The CD spectra of the proteins in TBS (0.25 mg/ml) were recorded on a JASCO J-720 spectropolarimeter (Tokyo, Japan) using a 1-mm path length cell at room temperature.

Dynamic light scattering (DLS)

The hydrodynamic radius of the proteins was measured by dynamic light scattering in TBS at 25°C using a Zetasizer Nano ZS (Malvern Instruments). The value was calculated as the average of six measurements.

Hemagglutination assay

The hemagglutination assay was performed by mixing serial twofold dilutions of sample proteins in TBS (30 μL) with an equal volume of a 5% (v/v) suspension of rabbit erythrocytes in round-bottomed microtiter plate wells (96 wells). The extent of agglutination was visually determined after incubation for 1 h at room temperature.

Measurements of carbohydrate-binding activity using mannotetraose (Man₄)-PD

Man₄-PD was prepared by reacting the aldehyde group of Man₄ with the primary amino groups of PD through reductive amination, as previously described (28). Carbohydrate-binding activity was evaluated by measuring the increase in Rayleigh scattering resulting from the formation of complexes between the protein and Man₄-PD. After recording the initial scattering intensity at 420 nm of Man₄-PD (26 μg/ml, 1 ml) in TBS using the Model F-3010 Fluorescence Spectrophotometer (Hitachi, Japan) at 25°C, small volumes of the protein solution in the same buffer were serially added, and the changes in the scattering intensity were recorded. In the inhibition assay, Man₄-PD (26 μg/ml) and LA or HA (19 μg/ml) were incubated in TBS at 25°C for 10 min to form their complex. Monosaccharides were then added as competitive inhibitors, and the decreases in light scattering were measured. The scattering values were corrected for dilution by the addition of the solutions.

Glycan array analysis

To examine the binding specificities of the recombinant rCG13321 for various oligosaccharides, glycan array analysis was performed using the RayBio Glycan Array 100 kit (RayBiotech, Norcross, GA, USA), as previously described (29,). The protein (0.3–0.5 mg/ml in 10 mM sodium phosphate, pH 8.0, 150 mM NaCl) was incubated with Cy3 NHS ester (0.63 mg/ml) at room temperature for 1 h in the dark, and then dialyzed against TBS for one day to remove any remaining reagent. After blocking the surface of the glycan array slide with the included Sample Diluent for 30 min, Cy3-labeled lectin was added to the array, which contained 100 oligosaccharides immobilized by linkers. The array was subsequently washed with Wash Buffer I and Wash Buffer II from the kit, followed by water. Detection of bound lectin was carried out using an Agilent DNA microarray scanner G2565CA (Agilent, Santa Clara, CA, USA). Fluorescence intensity analysis was performed using ImageJ (30).

Isothermal titration calorimetry

Isothermal titration calorimetry (ITC) was used to analyze the interaction between rCG13321 and mannose at 25°C using an iTC200 instrument (MicroCal, GE Healthcare). Aliquots of mannose solution (10 mM) in TBS were injected at 2-min intervals into 0.2 ml of LA or HA solutions (1.0 mg/ml) in TBS. The obtained data were analyzed using ORIGIN software, assuming the presence of four carbohydrate-binding sites in a protein molecule. Control experiments were conducted to measure the heat related to ligand dilution, and the resulting values were subsequently subtracted from the ligand binding thermograms.

Results

 

Structural comparison of CGL1 and CG13321

Nine genes containing DM9 repeats are listed in FlyBase (http://flybase.org/) (Table 1). They encode proteins comprising either two or four DM9 repeats. Since the highest similarity with CGL1 was found with CG13321 (identity, 32.82; Clustal W score, 31.47), we attempted to express its recombinant protein using E. coli cells. Figure 1A shows the comparison between the domain structures of CGL1 and CG13321. CG13321 is composed of four DM9 motifs that share apparent sequence similarity with the N- and C-terminal halves of CGL1 (Fig. 1B). The positions corresponding to the residues that interact with mannose molecules in CGL1 (enclosed in red and blue boxes) are relatively well shared in CG13321. Given that a protomer of CGL1 is composed of two DM9 repeats, CG13321 is expected to consist of two globular domains. In fact, AlphaFold2 predicted a CG13321 model with two distinctive domains, each of which is mainly composed of β-sheets, similar to the CGL1 protomer (Fig. 2).

Expression and purification of rCG13321

The cDNA encoding CG13321 was amplified by PCR using two primers and total RNA from D. melanogaster as a template. Subsequently, the amplified cDNA was inserted into a pET-3a vector, and rCG13321 was expressed in E. coli C43(DE3) cells. After cell disruption, both soluble and insoluble forms of rCG13321 were obtained (Fig. 3A). The soluble proteins were then applied to a mannose-Cellufine column. As shown in Fig. 3B, during washing the column with the buffer, a tailing elution of rCG13321 was observed, while 100 mM mannose solution eluted a significant amount of adsorbed rCG13321. Both proteins were found to be almost pure, showing bands corresponding to the mass of CG13321 (31 kDa) on SDS-PAGE in the presence of 2-mercaptoethanol. Subsequently, both protein bands were electroblotted onto a PVDF membrane and analyzed for their N-terminal amino acid sequences. The sequences of the first ten residues were determined to be GDYTWISTNV, confirming that they are the recombinant CG13321 without N-terminal methionine residues. It is likely that these residues were cleaved by E. coli methionine aminopeptidase. These results revealed that rCG13321 has an affinity for mannose, although it was separated into two fractions with different affinities for the mannose-Cellufine column. Thus, the tailing and adsorbed fractions were designated as LA (low-affinity) and HA (high-affinity), respectively (Fig. 3B). Interestingly, DLS analysis indicated that the particle size of HA increased to 30 nm after dialysis against TBS for 24 h, while that of LA remained almost unchanged (Fig. 4). This indicates that HA has a tendency to gradually form oligomers in solution.

CD spectra of rCG13321

As shown in Fig. 5, the far-UV CD spectra of LA and HA exhibited similar patterns, characterized by negative peaks around 215 nm. This suggests that both LA and HA are predominantly composed of β-sheet-rich secondary structures, similar to CGL1 (10). While the intensity of the negative peak in LA was slightly lower than that in HA, these spectra indicate that the secondary structures remained largely unaltered during oligomerization. AlphaFold2 also predicted a structure abundant in β-strands (Fig. 2B), which is consistent with these CD spectra.

Carbohydrate-binding properties of rCG13321

Hemagglutination assays of LA and HA were conducted using rabbit erythrocytes. As shown in Fig. 6A, agglutination of the cells was observed at concentrations of 50 μg/ml of LA and 6.3 μg/ml of HA. These results suggest that rCG13321 can bind to the carbohydrate chains on rabbit erythrocytes at multiple carbohydrate-binding sites, leading to cross-linking of the cells. Fig. 6B demonstrates that hemagglutination by LA and HA was inhibited with 6.3 mM and 3.1 mM of mannose, respectively. However, it was not inhibited by 25 mM of glucose, galactose, N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), fructose, d-fucose and l-rhamnose. This indicates that the binding of LA and HA is specific to mannose among these monosaccharides. The oligomannose-binding ability of rCG13321 was examined using Man₄-PD, which contains mannotetraose linked to the terminal primary amino groups of PD (31). As shown in Fig. 7A, when LA and HA were mixed with Man₄-PD, an increase in light scattering at 420 nm was observed, indicating that these proteins formed complexes with Man₄-PD. To examine the affinities of LA and HA for various monosaccharides, a competitive inhibition assay was performed. As shown in Figs. 7B and C, the most effective inhibition was observed for mannose among the monosaccharides tested, while fructose exhibited only slight inhibition, particularly in the case of LA. These results confirm the high specificity of CG13321 for mannose among the monosaccharides.

The binding specificities of LA and HA for various oligosaccharides were examined using glycan array analysis (Fig. 8). Consistent with the other experimental results, LA showed an affinity for the mannose monomer. However, it was also found that LA can bind to d-rhamnose with higher affinity, along with relatively lower affinity for l-rhamnose, compared to mannose. LA appears to mainly recognize the 2-, 3- and 4-hydroxy groups shared by mannose and d-rhamnose, with the 6-hydroxy group of mannose not involved in the binding. On the other hand, HA also showed binding to mannose, l-rhamnose and d-rhamnose similar to LA, but their intensities were relatively low. This may be due to the loss of the protein caused by its precipitation during Cy3-labeling. The introduction of Cy3 promoted the oligomerization tendency of HA, probably because of its hydrophobic nature, which led to the precipitation of the HA protein. In addition to the binding of these carbohydrates, weak binding of glucuronic acid was also observed for HA. However, its structural similarities with the other carbohydrates are uncertain.

As shown in Fig. 9, the association constants (K_a), and enthalpy (ΔH) and entropy (ΔS) changes for the binding of mannose to LA and HA were measured by ITC at pH 7.5 and 25°C. The analysis was performed assuming the existence of four mannose-binding sites in a protein molecule due to the relatively low affinity of mannose for the proteins. The K_a value for LA was determined to be 1.37 × 10³ M⁻¹ with no significant errors in the fitting to a theoretical titration curve. However, during the measurements for HA, relatively large errors were often observed in the nearly saturated region (as indicated by the vertical arrows in Fig. 9B). Therefore, the evaluation was carried out by eliminating the points with large deviations. This resulted in a K_a value of 3.73 × 10³ M⁻¹ for HA. Such irregular heat changes were consistently observed in the measurements for HA, likely due to the oligomerization of HA promoted by binding with mannose.

Discussion

CGL1 is composed of two identical subunits, each of which is a CRD consisting of two DM9 motifs. Two carbohydrate binding sites are located at the interface between these motifs. Within each binding site, residues from both DM9 motifs participate in carbohydrate binding. The DM9 repeats were initially discovered as novel repeats with unknown functions in the D. melanogaster genome (11,). Proteins that contain DM9 motifs are primarily found in invertebrates and are believed to be involved in interactions with foreign organisms (17,  20,  32,  33). Nine genes in D. melanogaster are annotated as DM9 repeat-containing proteins in FlyBase (Table 1). They contain either two or four DM9 repeats in a single molecule. Among them, CG13321 exhibits the highest sequence similarity with CGL1, although it contains four DM9 motifs corresponding to two CRDs, which is consistent with the structure predicted by AlphaFold2 (Fig 2B). The CD spectra of rCG13321 (Fig. 5) suggest a secondary structure rich in β-sheets, similar to CGL1. However, their overall structures are quite different. While CGL1 has its two subunits in contact over a large surface area, CG13321 has its two domains distinctly separated with a narrowing in the central region, indicating the possibility of flexibility between the domains.

Affinity chromatography of rCG13321 using the mannose-Cellufine column demonstrated that rCG13321 possesses mannose-binding ability. However, the protein was separated into two fractions, LA and HA, based on their difference in binding ability for the column. Furthermore, HA was found to have a higher tendency to oligomerize in solution, as revealed by DLS measurements (Fig. 4). There may be a subtle conformational difference between HA and LA, leading to variations in their binding ability to the column and their tendency for oligomerization, although this conformational difference appears very small based on their nearly identical CD spectra (Fig. 5). While such a conformational difference may arise during the folding process in E. coli cells, it is also probable that some bacterial components may have affected these variations in the expressed proteins, given that rCG13321 possibly has a role in innate immunity.

Hemagglutination and Man₄-PD agglutination assays confirmed the presence of multiple mannose-binding sites in an rCG13321 molecule, consistent with its amino acid sequence and tertiary structure prediction (Fig. 2). Inhibition experiments for these assays also demonstrated that rCG13321 primarily binds to mannose among the tested monosaccharides. However, a slight affinity for fructose was detected in the Man₄-PD agglutination assay. On the other hand, glycan array analysis revealed that LA has a higher affinity for d-rhamnose compared to mannose. This suggests that while rCG13321 primarily recognizes the 2-, 3- and 4-hydroxy groups of mannose and d-rhamnose, the methyl group of d-rhamnose at the C-6 position further enhances the binding affinity. Figures 10A and B illustrate the binding modes of mannose in the two carbohydrate binding sites of CGL1. In these sites, the C-6 carbon atoms of mannose are in contact with the aromatic side chains of Phe126 and Phe55. This suggests that the binding of mannose is stabilized by CH/π interactions (34–36) in addition to hydrogen bonds with the 2-, 3- and 4-hydroxy groups (Fig. 10A) or the 1-, 2- and 3-hydroxy groups (Fig. 10B). In the case of CG13321, the predicted binding site structures suggest that the residues equivalent to these hydrogen-bonding residues (Glu, Lys and Asp) (enclosed by red circles) are highly conserved. Furthermore, the aromatic residues Tyr129, Trp58, Phe270 and Tyr199 are located in positions corresponding to Phe126 and Phe55 of CGL1, suggesting that these aromatic residues are involved in the binding of mannose through CH/π interactions. In contrast to LA, the binding signals of HA were too weak in the glycan array analysis to distinguish differences in the affinities for the oligosaccharides. This was due to the partial formation of precipitates of HA during Cy3-labeling. Therefore, the precise binding specificity of HA could not be determined. However, it is highly likely that HA shares similar binding specificity with LA, given the similarity in their binding profiles in the Man₄-PD assay.

In the present study, we found that rCG13321 has the ability to bind to mannose. However, it was separated into two fractions, LA and HA, which differ in their ability to bind to the mannose-Cellufine column and their tendency to form oligomers. The oligomerizing behavior of HA was also observed during glycan array analysis and ITC measurements, probably induced by carbohydrate binding or labeling with Cy3. The differences in the oligomerization tendency between LA and HA suggest that there are partial differences in their spatial structures. It is important to note that DM9 domain proteins in other organisms have been suggested to play roles in innate immune responses. For instance, after oral infection with the pathogenic bacterium Pseudomonas entomophila, a DM9 domain protein, CG16775, was up-regulated in D. melanogaster larvae (37,). A DM9 protein of the mosquito A. gambiae, PRS1, is induced in the epithelial cells of the salivary glands and the midgut upon invasion by the malaria parasite Plasmodium, and it was relocated to vesicle-like structures (32,). In fact, CG13321 and other DM9 domain proteins, CG10527 and CG3884, have been found to be associated with the Drosophila phagosome and are presumed to be involved in intracellular interactions, including immune responses (38). Although it is currently uncertain whether rCG13221 exists in two different forms, LA and HA, in Drosophila cells, it is possible that these forms may be associated with specific functions, such as the detection and exclusion of mannose-containing foreign microorganisms. Further investigation into the structure of CG13321 would provide valuable insights into the functions of this protein.

Funding

This work received support from the Organization for Marine Science and Technology, Nagasaki University, and the Nagasaki University Grant for Co-creation Research.

Supplementary Data

Supplementary data are available at JB Online.

Conflict of Interest

The authors declare that they have no conflicts of interest.

References