Heptabladed β-propeller lectins PLL2 and PHL from Photorhabdus spp. recognize O-methylated sugars and influence the host immune system

Abstract

O-methylation is an unusual sugar modification with a function that is not fully understood. Given its occurrence and recognition by lectins involved in the immune response, methylated sugars were proposed to represent a conserved pathogen-associated molecular pattern. We describe the interaction of O-methylated saccharides with two β-propeller lectins, the newly described PLL2 from the entomopathogenic bacterium Photorhabdus laumondii, and its homologue PHL from the related human pathogen Photorhabdus asymbiotica. The crystal structures of PLL2 and PHL revealed up to 10 out of 14 potential binding sites per protein subunit to be occupied with O-methylated structures. The avidity effect strengthens the interaction by 4 orders of magnitude. PLL2 and PHL also interfere with the early immune response by modulating the production of reactive oxygen species and phenoloxidase activity. Since bacteria from Photorhabdus spp. have a complex life cycle involving pathogenicity towards different hosts, the involvement of PLL2 and PHL might contribute to the pathogen overcoming insect and human immune system defences in the early stages of infection.

Databases

Structural data are available in PDB database under the accession numbers 6RG2, 6RGG, 6RFZ, 6RG1, 6RGU, 6RGW, 6RGJ, and 6RGR.

Abbreviations

3OMG

3-O-methyl-d-glucose

fMLF

N-formyl-l-methionyl-l-leucyl-l-phenylalanine

IJs

infective juveniles

ITC

isothermal titration calorimetry

MGMR

3,6-O-Me₂-d-Glcβ1-4(2,3-O-Me₂)-l-Rhaα

MGMR-sp2

3,6-O-Me₂-d-Glcβ1-4(2,3-O-Me₂)-l-Rhaα-O-(p-C₆H₄)-O-CH₂CH₂NH₂

PAMPs

pathogen-associated molecular patterns

PMA

phorbol 12-myristate 13-acetate

PRP

pattern recognition protein

SPR

surface plasmon resonance

Introduction

A host immune reaction is dependent on pathogen recognition. It is mediated by interactions of pattern recognition proteins (PRPs) of the host with evolutionary conserved epitopes of a pathogen (pathogen-associated molecular patterns, PAMPs) [[1, 2]]. A typical example of PAMPs are glycans, including O-methylated sugars, which cover the surface of every cell. O-methylated glycans were reported to be present in bacteria, worms, fungi, plants and molluscs, but not in arthropods and mammals [[3]]. Proteins recognizing sugar moieties, lectins, are an example of PRP. Recently, the interaction of lectin Lb-Lec2 from the mushroom Laccaria bicolor with O-methylated carbohydrates was described [[4]]. Lb-Tec2 belongs to the tectonin β-propeller lectin family, consisting of evolutionary conserved immune system effectors. Tectonins were reported to bind bacterial LPS and aggregate bacterial cells. In addition, the Lb-Tec2 lectin expresses toxicity against the nematode Caenorhabditis elegans [[5]].

Photorhabdus is a Gram-negative bioluminescent Gamma-proteobacterial genus that traditionally belonged to the family Enterobacteriaceae. The phylogeny of the genus was recently intensively revisited. The whole genus Photorhabdus was assigned to the newly proposed family Morganellaceae based on phylogenetic analyses [[6]]. Many of its subspecies were elevated to the level of species [[7]]. All Photorhabdus species form highly entomopathogenic complexes with nematode symbiont Heterorhabditis spp.

The complex life cycle of Photorhabdus spp. involving a mutualistic and pathogenic phase towards different hosts is well described [[8-11]]. Bacteria live attached to the intestine of infective juveniles (IJs), a specialized larval stage of nematodes. IJs seek out a suitable host and invade it. After the invasion, Photorhabdus is regurgitated from the IJ’s intestine into the insect’s haemocoel, where it produces numerous toxins and other virulence factors that kill the host within 48 h [[12, 13]]. Furthermore, Photorhabdus grows on host tissues and serves as nutrition for the developing nematodes. IJs mature inside the cadaver and undergo several cycles of sexual reproduction. When the host tissues are depleted, newly developed IJs are reassociated with Photorhabdus and search for a new host [[14]]. A complex of Photorhabdus and its symbiont is produced and commercially used to control pest insects [[15, 16]]. Moreover, Photorhabdus asymbiotica and P. australis are able to directly infect humans [[17, 18]].

Lectins in general are involved in various processes, including immune response and cell-to-cell communication [[19, 20]]. Lectins produced by entomopathogens with a complex life cycle like the genus Photorhabdus could play a role in three basic processes including the mediation of pathogenicity, mutualism or communication within a bacterial population. The bacterium Spiroplasma ciri uses a lectin–glycoconjugate interaction to invade insect cells [[21]]. The marine nematode Laxus oneistus uses the lectin Mermaid to acquire a bacterial symbiont [[22]]. Recently, we proved the interaction of PLL produced by P. laumondii, formerly known as Photorhabdus luminescens subsplaumondii, and PHL from P. asymbiotica, both with host immune cells and nematode symbionts [[23, 24]]. The involvement of Photorhabdus lectins in the inhibition of immune response is one of the most interesting findings.

Upon recognizing a pathogen, constitutive immune effectors are activated. This is followed by the production of induced molecules such as antimicrobial peptides a few hours later [[25]]. In our previous research, we have shown that the early immune response is substantial in the host protection against nematobacterial infection [[26]]. The pathogen recognition in insects is generally manifested by the activation of coagulation and a phenoloxidase (PO) system; in the latter, the enzyme PO is present in its inactive form referred to as prophenoloxidase (pPO), which is activated by proteolytic cleavage in the last step of the activation cascade and subsequently catalyses melanization [[27]]. The activation of the PO system is accompanied by the production of cytotoxic molecules that participate in the killing of a pathogen [[28, 29]]. Similarly, the vertebrate innate immune system reacts to recognized pathogens by an oxidative burst in phagocytes and the production of highly effective reactive oxygen species (ROS) [[30]].

In this article, we focused on the interaction between O-methylated saccharides and two homologous lectins – PHL from P. asymbiotica and PLL2 from P. laumondii. The proteins were prepared in a recombinant form, and their interaction with O-methylated saccharides was characterized on a molecular level. Crystal structures of PHL and PLL2 in complex with O-methylated sugars confirmed clear recognition of this type of ligands. Since PLL2 lectin was not described yet, we also performed its basic characterization including the effect on the insect's early immune response and compared obtained results for PLL2 with its close homologue PHL.

Results

PLL2 basic characterization

The protein PLL2 was identified by bioinformatic analyses of the Photorhabdus laumondii subsp. laumondii genome as a close homologue of the lectins PLL [[23]] and PHL [[24]]. The sequence consists of 371 amino acids (including initial methionine). PLL2 has 65% and 81% sequence identity with PLL and PHL, respectively. The pll2 synthetic gene was cloned into Escherichia coli, the produced recombinant protein was purified on d-mannose-agarose resin by affinity chromatography and eluted isocratically with a typical yield of 60 mg·L⁻¹ of medium (Fig. 1A). The protein forms a single band on the SDS/PAGE gel with an apparent molecular mass of ~ 40 kDa (Fig. 1B). The protein identity was verified using MALDI MS/MS and intact mass analyses (Fig. 1C,D). The oligomeric state in solution was examined using the analytical ultracentrifugation (AUC) method of sedimentation velocity. The AUC sedimentation coefficient of PLL2 was 4.89 S, which corresponds to a dimeric state of the protein (Fig. 2).

Fig. 1

Protein quality and identity. (A) Representative record of PLL2 purification using affinity chromatography on d-mannose-agarose resin and (B) corresponding SDS/PAGE. Experiments were performed five times. Soluble fractions, protein marker III (AppliChem). Unbound fractions are highlighted by red rectangle, and fractions with PLL2 used for further analyses are highlighted by violet rectangle. (C) Sequence coverage of PLL2 from MALDI MS/MS analyses. Confirmed sequence is in bold and red, uncertain regions are red, and undetected parts are black. (D) PLL2 intact mass analysis showed molecular mass 40 115 Da, which corresponds to the full-length sequence including initial methionine.

Fig. 2

Sedimentation velocity experiment. (A) Sedimentation curves of PLL2 (0.15 mg·mL⁻¹) and fitted profiles were obtained from continuous C(s) analysis using Sedfit. (A, top) Scans were recorded every 4 min, every third scan is shown. Residual plot (A, bottom) shows difference between experimental data and fitted curves. (B) Continuous size distribution of sedimenting species resulted in peak with sedimentation coefficient of 4.89 S (S_20,w = 4.88 S as calculated in Sednterp). Figures were created using GUSSI 1.0.8 software [[31]]. The same distribution curves were obtained also for two different protein concentrations (0.1 and 0.25 mg·mL⁻¹).

A screening of PLL2 carbohydrate specificity was performed using a glycan array microchip containing 381 different sugar epitopes of mammalian or bacterial origin (Fig. 3). The protein interacted preferentially with α-l-fucose followed by the disaccharide Glcβ1-4GalNAcα and the unusual O-methylated disaccharide 3,6-O-Me₂-d-Glcβ1-4(2,3-O-Me₂)-l-Rhaα (MGMR), present in the Mycobacterium leprae glycolipid PGL-1 [[32, 33]]. Other fucosylated saccharides were recognized much more weakly. A similar binding of α-l-fucoside and MGMR was reported for PLL [[23]] and PHL [[24]].

Fig. 3

Box and whiskers plot for glycan array screening with PLL2 labelled with DyLight 488. Top 15 ligands are presented. Saccharides that gave signals with signal-to-noise ratio > 3 are labelled. Linker formulas are as follows: sp2, -O-(p-C₆H₄)-O-CH₂CH₂NH₂; sp3, -O-(CH₂)₃NH₂. Bottom and top of boxes are first and third quartiles, band inside boxes represents median; ends of whiskers are minimum and maximum values of data, and small squares inside boxes represent mean. RFU, relative fluorescence units. Complete glycan array results and raw data are given in supplemental Tables S1 and S2, respectively.

Isothermal titration calorimetry (ITC) was employed to gain more detailed information about PLL2–sugar interactions. Experiments were carried out with six standard monosaccharides chosen according to the glycan array results. The binding stoichiometry of the interacting ligands could not be properly determined due to the low affinity of the interaction. Therefore, the stoichiometry was fixed during the fitting procedure. The calculated dissociation constants are in the low millimolar range. The highest affinity was determined towards methyl-α-l-fucoside (Me-α-l-Fuc) and free d-galactose. l-fucose and d-glucose were weaker binders, while no significant binding was observed with N-acetyl-galactosamine and l-rhamnose (Table 1 and Fig. 4).

Table 1. ITC experiments. Calculated dissociation constants (K_D) of PLL2 (0.1 mm) and PHL (0.1 mm) titrated with selected carbohydrate ligands (20 mm) measured by ITC experiments. Stoichiometry (N) was fixed during the fitting procedure. In the case of d-Glc and 3OMG, the stoichiometry was fixed to the value known from the x-ray crystal structure. In the case of other ligands, the stoichiometry was fixed to value 2, 4, 6 and 8, respectively, due to the unknown number of binding sites. × = interaction not measured.

Ligand	KD (mm)
	PLL2				PHL
	N = 2	N = 4	N = 6	N = 8	N = 8
d-Gal	0.52 ± 0.02	0.37 ± 0.03	0.28 ± 0.04	0.25 ± 0.06	×
Me-α-l-Fuc	0.61 ± 0.01	0.43 ± 0.02	0.29 ± 0.03	0.23 ± 0.04	×
l-Fuc	2.81 ± 0.07	2.57 ± 0.08	2.35 ± 0.08	2.1 ± 0.1	×
d-Glc		3.56 ± 0.21			×
3OMG				11.2 ± 0.8	14.1 ± 0.9
GalNAc	No interaction				×
d-Rha	No interaction				×

Fig. 4

ITC curves of PLL2 (100 μm) and PHL (100 μm) titrated with 20 mm ligands. All experiments were replicated three times, representative image is shown for each experiment. Titration of PLL2 with d-galactose (A), methyl-α-l-fucoside (B), l-fucose (C), d-glucose (D), 3OMG (E), N-acetyl-galactosamine (G) and l-rhamnose (H). Titration of PHL with 3OMG (F). Baseline is highlighted in red. Individual monosaccharides are colour-coded: blue = d-galactose, red = methyl-α-l-fucoside, orange = l-fucose, dark green = d-glucose, light green = 3OMG, cyan = N-acetyl-d-galactosamine, violet = l-rhamnose.

Interaction with O-methylated saccharides

The interaction of lectins PLL2 and PHL with O-methylated saccharides was investigated by ITC and surface plasmon resonance (SPR). ITC showed a very weak interaction of both lectins towards 3-O-methyl-d-glucose (3OMG) in the low millimolar range (Table 1 and Fig. 4).

The interaction with α-l-fucoside and MGMR was further analysed using SPR (Fig. 5). Each saccharide was immobilized on the chip surface, and the interaction with PLL2 and its homologue PHL was analysed. The apparent K_D of PLL2 towards α-l-fucoside was calculated to be 117.7 µm, which is two orders of magnitude lower than the published value of PHL [[24]]. Two channels with different densities of MGMR were prepared to evaluate the avidity effect of binding. While the apparent affinity towards the ligand immobilized in low density was in the micromolar range, the apparent affinity towards the high-density immobilized ligand was four orders of magnitude higher, suggesting a strong avidity effect (Table 2).

Fig. 5

SPR curves of PLL2. All experiments were replicated three times, representative image is shown for each experiment. (A) PLL2 binding to immobilized O-methylated disaccharide MGMR. Apparent K_D was calculated using single-cycle kinetics approach. (B) PLL2 binding to immobilized α-l-fucoside. Different protein concentrations in twofold dilution series are colour-coded from dark red = 50 µm to violet = 0.195 µm. (C) Inhibitory effect of monosaccharides on PLL2 binding to α-l-fucoside channel and (D) MGMR channel. Individual monosaccharides are colour-coded: red = methyl-α-l-fucoside, orange = l-fucose, light green = 3OMG, dark green = d-glucose, violet = l-rhamnose and blue = d-galactose.

Table 2. Apparent K_D for PLL2 and PHL binding to immobilized MGMR or α-l-fucoside. Standard deviations are calculated from three individual measurements.

Ligand	Apparent KD
	PLL2	PHL
α-l-fucoside	117.7 ± 4.8 µm	9.6 ± 0.7 µm
MGMR (low density)	25.4 ± 5.2 µm	4.1 ± 0.4 µm
MGMR (high density)	6.1 ± 0.2 nm	0.95 ± 0.04 nm

Further, competitive inhibition with selected saccharides was tested. The determined IC₅₀ values are given in Table 3. For both proteins tested, methyl-α-l-fucoside was the best inhibitor with IC₅₀ in the high micromolar range, followed by l-fucose, being a 10 times weaker inhibitor. A clear difference between the inhibitory effect of d-glucose and 3OMG was observed. 3OMG was a relatively weak inhibitor with an IC₅₀ in the low millimolar range, while d-glucose was incapable of inhibiting the binding, even at the highest concentration tested (400 mm).

Table 3. Inhibitory effect of selected saccharides on PLL2 and PHL binding to immobilized ligand α-l-fucoside or 3,6-O-Me₂-d-Glcβ1-4(2,3-O-Me₂)-l-Rha (MGMR), respectively. IC₅₀ was determined from a chart of inhibitor dilution series vs RU response.

	IC50 [mm]
Ligand	PLL2		PHL
Inhibitor	MGMR	α-l-fucoside	MGMR	α-l-fucoside
l-Fuc	1.2	0.88	4.0	0.36^a,a, ^a,a
Me-α-l-Fuc	0.12	0.07	0.44	0.10
3OMG	5.2	12.7	4.8	7.1
d-Gal	> 100	> 100	> 100	> 100
l-Rha	> 400	332	> 400	147
d-Glc	> 400	> 400	> 400	> 400

• ^a From Ref. [[24]].

X-ray structure of lectin–carbohydrate complexes

In order to determine binding sites specific for methylated sugar molecules, both the PHL and PLL2 crystals were soaked with O-methylated sugar ligands – 3OMG and MGMR-sp2 and their nonmethylated components (d-glucose and l-rhamnose), respectively. In all complexes, the electron density was clear enough to assign the position of at least one ligand molecule per protein subunit however, the binding site preferences differed significantly for methylated and nonmethylated saccharides.

The overall structure of PHL was already determined earlier [[24]]. PLL2 crystallized as a homodimer in the P2₁ or I2 space groups (Table 4). Each subunit forms a seven-bladed beta-propeller with a tunnel in the centre. The N and C termini close the tunnel on one side (top), while both protein subunits interact via the opposite ‘open’ side of the propeller (bottom) (Fig. 6A,B). The overall similarity between PHL and PLL2 dimers was high. The backbone RMSD between the PHL and PLL2 complexes with the same ligand varied from 0.25 to 0.42 Å. RMSD of the protein backbone of the all solved PHL and PLL2 complexes ranged from 0.10 to 0.15 Å and 0.11 to 0.34 Å, respectively. Similar to PHL, the PLL2 dimer harbours two sets of binding sites arranged in four stacked rings (Fig. 6C,D) making it the second experimentally proven ‘bangle’ lectin, as was defined previously for PHL [[24]]. The binding sites are situated in between blades. All seven sites closer to the ‘top’ side of the propeller consist of a hydrophobic pocket deepening into the protein surrounded by mainly polar residues. The ‘bottom’ ring binding sites are formed by a different set of mainly polar amino acids without a hydrophobic pocket. As the previously proposed labelling of binding sites as fucose- and galactose-binding is not meaningful for the complexes studied here, we decided to label the ‘top’ ring sites as hydrophobic (H) and the ‘bottom’ ring ones as polar (P) sites.

Table 4. Data collection and refinement statistics for PLL2 and PHL complexes. Values in parentheses correspond to the highest resolution shell.

Complex	PLL2/3OMG	PLL2/MGMR-sp2	PLL2/Glc	PLL2/Rha	PHL/3OMG	PHL/MGMR-sp2	PHL/Glc	PHL/Rha
PDB ID	6RG2	6RGG	6RFZ	6RG1	6RGU	6RGW	6RGJ	6RGR
Beamline	BESSY, 14.2	PETRA, P13	BESSY, 14.2	BESSY, 14.2	BESSY, 14.2	BESSY, 14.2	BESSY, 14.3	BESSY, 14.3
Wavelength (Ǻ)	0.9184	0.9763	0.9184	0.9184	0.9184	0.9184	0.8950	0.8950
Space group	P21	P21	I2	I2	P312	P312	P312	P312
Unit-cell parameters
a, b, c (Å)	51.7, 56.7, 108.48	68.3, 89.4, 68.4	65.5, 71.1, 86.2	65.1, 71.0, 85.8	80.8, 80.8, 113.2	80.9, 80.9, 113.6	80.6, 80.6, 113.5	80.5, 80.5, 113.4
α, β, γ (°)	90, 95.3, 90	90, 101.4, 90	90, 104.0, 90	90, 104.6, 90	90, 90, 120	90, 90, 120	90, 90, 120	90, 90, 120
Resolution range (Ǻ)	44.84–1.12 (1.15–1.12)	67.03–2.20 (2.32–2.20)	47.39–1.80 (1.90–1.80)	47.11–1.70 (1.79–1.70)	44.02–1.75 (1.84–1.75)	44.13–1.75 (1.84–1.75)	44.03–1.80 (1.90–1.80)	43.99–1.89 (1.99–1.89)
Reflections measured	882 177 (120 183)	155 472 (22 456)	134 222 (19 902)	155 449 (23 066)	488 247 (65 470)	488 595 (70 377)	494 599 (68 724)	422 103 (59 852)
Unique reflections	224 157 (16 606)	40 910 (5 878)	35 262 (5 132)	41 015 (5 990)	43 779 (6 310)	44 002 (6 338)	40 112 (5670)	34 618 (4937)
Completeness (%)	98.6 (98.9)	99.8 (99.8)	99.0 (98.8)	98.6 (98.8)	100.00 (100.0)	100.00 (100.00)	99.7 (98.2)	99.8 (98.6)
CC1/2 (%)	99.8 (74.0)	99.3 (54.1)	90.3 (45.2)	97.5 (51.9)	99.9 (69.3)	99.9 (72.3)	99.8 (89.5)	99.9 (74.2)
R merge	0.072 (0.768)	0.096 (0.922)	0.288 (1.045)	0.133 (0.962)	0.093 (1.207)	0.073 (1.128)	0.144 (0.795)	0.113 (1.317)
Multiplicity	3.7 (3.5)	3.8 (3.8)	3.8 (3.9)	3.8 (3.9)	11.2 (10.4)	11.1 (11.1)	12.3 (12.1)	12.2 (12.1)
<I/σ (I)>	8.1 (1.6)	5.1 (1.1)	3.4 (1.5)	4.4 (1.3)	14.0 (1.8)	17.0 (2.2)	18.5 (4.5)	16.1 (2.0)
Reflections used	224 157	38 568	33 419	38 688	41 534	41 752	38 051	32 866
Reflections used for Rfree	11 786	2 033	1 672	2 004	2 206	2 209	2 021	1712
R factor (%)	13.64	19.92	21.04	18.15	16.43	18.30	16.58	19.97
Rfree (%)	16.10	22.42	23.50	20.95	19.82	20.95	19.54	22.88
Rmsd bond lengths (Ǻ)	0.009	0.007	0.003	0.005	0.012	0.005	0.010	0.006
Rmsd bond angles (°)	1.58	1.43	1.19	1.22	1.725	1.296	1.570	1.309
No. of water molecules	1059	16	361	372	282	266	355	172
No. of non-H atoms (total)	6877	5353	3089	3060	3097	3052	3106	2847
Ramachandran plot
Most favourable regions (%)	97.1	95.4	96.8	96.8	97.4	97.1	96.8	96.2
Allowed regions (%)	2.9	4.5	3.2	3.2	2.6	2.9	3.2	3.8
Outliers (%)	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0

Fig. 6

PHL and PLL2 overall structure of MGMR-sp2 complexes. (A) Side view of PLL2 dimer (magenta) and PHL dimer (green, symmetry-related molecule white). Saccharide ligands are shown in blue (PLL2), yellow (PHL) and white (PHL symmetry molecule). (B) Top view of PHL and PLL2 chain A with individual binding sites and N and C termini labelled. (C) Sequence alignment of PHL (NCBI reference sequence: WP_012776886.1) and PLL2 (NCBI reference sequence: WP_011145109.1) calculated using Clustal Omega [[34]]. Individual beta strands are marked with full arrows. Residues involved in polar binding sites are highlighted in green, and residues involved in hydrophobic binding sites are highlighted in blue. Different amino acids are highlighted in orange, and different amino acids involved in binding sites are highlighted in red. (D) Surface of PLL2 molecule with highlighted differences from PHL. Residues involved in polar binding sites are highlighted in green, residues involved in hydrophobic binding sites are highlighted in blue. Residues different from PHL are highlighted in orange (outside binding sites) or in red (involved in binding site).

The accessibility of particular binding sites was affected by the inter-subunit contacts in the crystal. For PLL2, sites 4H and 7H of chain A and sites 3H and 7H of chain B were partially blocked by crystal contacts. For PHL, the site 2H is occupied by Met60 from a neighbouring molecule and the site 5H is affected by steric hindrance (Fig. 7A). Carbohydrate molecules were not identified in any of these sites.

Fig. 7

Close-up of selected binding sites of PHL and PLL2. (A) PHL 2H binding site occupied by Met60 of neighbouring protein molecule. (B–G) Selected binding sites with coordinated ligand. F_obs-F_calc electron density around ligands is displayed at 2.5σ as dark green mesh and superposed with final ligand positions after refinement. Residues involved in ligand binding are coloured bright green. Individual panel depicts: PHL in complex with MGMR-sp2 sites 1H (B) and 3H (C), PLL2 in complex with MGMR-sp2 sites 2H (D) and 6H (E). PHL in complex with d-glucose site 4P (F). PHL in complex with l-rhamnose site 4P (G). (H) Superposition of PLL2 site 6P in complex with d-glucose (dark yellow) and 3OMG (cyan). Anomeric oxygens are coloured black. Anomeric oxygen of d-glucose points towards solvent (dark yellow arrow), anomeric oxygen of 3OMG points inside binding pocket (cyan arrow). (I) Superposition of PHL site 4P in complex with d-glucose (dark yellow) and l-rhamnose (magenta), different tilt of ligands and loop Q251-S260 reorientation is shown. Residues not resolved in PHL/Rha complex electron density are represented by dashed line. Residues involved in ligand binding are coloured green.

In complexes with MGMR-sp2, the ligand was found only in H-type sites (Table 5). In the case of PHL, only the 3,6-O-Me₂-Glc moiety was recognized, with O6-methyl group always preferred. O4 and O6 of the 3,6-O-Me₂-Glc moiety were coordinated by polar residues. The whole complex was further stabilized by a CH-π interaction between the saccharide and the Trp/Tyr residues. The electron density for assigning of the 2,3-O-Me₂-Rha moiety was clear enough only in sites 1H and 3H (Fig. 7B,C). In both cases, the l-rhamnose moiety showed only a minimal contact with the protein. In the case of PLL2, the 3,6-O-Me₂-Glc moiety was preferentially recognized. At sites 1H (both chains) and 4H (chain B), the 6O-methyl is bound in the hydrophobic pocket with further ligand stabilization by O4 and O6 hydrogen bonds and a CH-π interaction with C6. At site 5H (chain A), the 3O-methyl group is recognized with the ligand further stabilized by O3, O4 and O6 hydrogen bonds. The same binding mode (via 3O-methyl group) was observed for sites 6H (chain A) and 2H (chain B). However, in these two cases, the saccharide was stabilized in the crystal complex by additional binding of the 2O-methoxy group of the 2,3-O-Me₂-Rha moiety by neighbouring molecule sites 5H (chain B’) and 2H (chain A’), respectively (Fig. 7D,E).

Table 5. Site occupancy of PHL–sugar and PLL2–sugar complexes. Sites occupied with ligands are coloured in green.

	PHL				PLL2
	3OMG	MGMR-sp2	Glc	Rha	3OMG	MGMR-sp2	Glc	Rha
Hydrophobic binding site
1H
2H	^a	^a	^a	^a
3H
4H
5H
6H
7H
Polar binding site
1P
2P
3P
4P
5P
6P
7P

• ^a Occupancy with Met60 from symmetrically related PHL molecule, compounds of precipitant and cryoprotecting solution not listed.

In the complexes with 3OMG, the sugar ligand was identified in both types of sites. In both lectins, 3OMG has the same binding mode. In the H-type sites, the 3-O-methyl group is coordinated inside the hydrophobic pocket and the anomeric oxygen points to the solvent. This orientation allows for further continuation of the sugar chain with more complex saccharides, such as those naturally present in the mycobacterial cell wall. The accommodation of the O-methyl group by the hydrophobic binding site pocket is accompanied by hydrogen bonding to O3 and O4 (PHL sites 1H, 6H, 7H), O3, O4 and O6 (PLL2 sites 1H, 4H, 7H) or O1, O2 and O3 (PLL2 sites 2H, 3H, 6H, and PHL sites 3H, 4H). On the other hand, in the P-type sites, the 3-O-methyl group points outwards the binding site and does not contribute to binding. The ligand was found in the binding sites 2P, 4P and 6P in both lectins. In these sites, the saccharide is coordinated by a net of hydrogen bonds to O1, O2, O3 and O5 (PHL binding sites) or O1, O2, O3, O5 and O6 (PLL2 binding sites). The ligand is also stabilized via CH-π interactions of the ring CH groups.

The nonmethylated monosaccharides d-glucose and l-rhamnose were only bound in the P-type sites however with a different orientation from 3OMG and from each other. d-glucose occupied the same sites as 3OMG (2P, 4P, 6P) interacting mainly through hydrogen bonds of O2, O3, O4 and via Trp stacking. O1 points to the solvent in contrast to 3OMG, which has O1 oriented inside the binding pocket (Fig. 7H). d-glucose occupied also a site 5P, where it was bound through the H-bond of O1, O2, O3 and O5. l-rhamnose was only bound at 6P (PLL2) and 4P and 6P (PHL) sites. The ligand was coordinated via O1, O2, O3 and O5, and its position was tilted by ~ 60° with respect to the d-glucose molecule bound to the same binding sites (Fig. 4F,G). At the PHL site 4P, this resulted in a distortion of the neighbouring protein loop of blade V (Fig. 4I).

Binding to insect haemocytes

The interaction of PLL2 with host tissues was studied in the haemolymph of Galleria mellonella larvae, from which we prepared the layer of adhered haemocytes. Laser scanning confocal microscopy confirmed the binding of PLL2 labelled with DyLight 488 to the surface of adhered haemocytes (Fig. 8A). The staining of actin filaments in haemocytes treated with PLL2 did not show any signs of cytoskeletal rearrangement. The binding of PLL2 was not uniform among haemocytes and differed in intensity, as is especially visible in the merged figure (Fig. 8C). However, the flow cytometry determined that PLL2 binds to the vast majority of haemocytes instead of targeting specific populations (Fig. 8D,E).

Fig. 8

Interaction of PLL2 with haemocytes of G. mellonella. All experiments were replicated three times, representative image is shown for each experiment. PLL2 was labelled with DyLight 488 (A) and incubated with adhered haemocytes. Staining of actin filaments with Alexa Fluor® 594 phalloidin was used to visualize inner structure of haemocytes (B). Overlay image shows binding of DyLight 488-labelled PLL2 onto insect haemocyte surface (C). Scale bar equals 50 µm. Proportion of haemocytes bound with DyLight 488-labelled PLL2 was measured by flow cytometry and determined by histogram analysis. Positive (I1) and negative (I2) regions for DyLight 488-labelled PLL2 were set for G. mellonella haemolymph in the absence of PLL2 (D). All haemocytes were present in positive region after treatment with PLL2-DyLight 488 (E).

Activity of phenoloxidase (PO)

Melanization catalysed by PO in the G. mellonella cell-free haemolymph was significantly increased in the presence of PLL2, whereas the control protein BSA had no effect (Fig. 9). The tested protein did not influence the total PO activity measured after the proteolytic activation of pPO. Moreover, melanization in the insect haemolymph was about one order of magnitude higher after pPO activation than the PO activity measured in the presence of PLL2, which confirms the protein only activates a part of the PO system present in the plasmatic fraction of the G. mellonella haemolymph.

Fig. 9

Effect of PLL2 on PO and total PO activity (pPO) in cell-free haemolymph of G. mellonella. Haemolymph was treated with PLL2, and melanization catalysed by PO was measured for 30 min at 492 nm with 3,4-dihydroxy-dl-phenylalanine as substrate. To determine total PO activity, proenzyme was converted to active enzyme by α-chymotrypsin treatment before addition of PLL2 and substrate. 20 mm Tris/HCl, 150 mm NaCl, pH 7.5, buffer and BSA were used as controls. Melanization rate was expressed as integral of absorbance per µL of haemolymph. Data are presented as mean of four experiments ± SD; significant differences P < 0.05 are indicated by different letters above columns (Tukey's test).

The increase in melanization induced by PLL2 can be inhibited by pretreatment with l-fucose, Me-α-l-Fuc or d-glucose (Fig. 10). A similar effect was observed with the related protein PHL, where inhibition by l-fucose and Me-α-l-Fuc was also observed [[24]]. In contrast to PLL2, which was also inhibited by d-glucose, PHL-induced melanization was inhibited by 3OMG. The saccharides used for inhibition did not have any effect on PO activity in insect haemolymph when evaluated in the absence of tested proteins.

Fig. 10

Inhibitory effect of selected saccharides on PLL2- (A) and PHL- (B) promoted melanization in haemolymph of G. mellonella. Haemolymph was incubated with PBS, BSA, PLL2 or PHL at concentration of 4 mg·mL⁻¹ (active or inhibited by specified saccharides at concentration of 0.2 m) and subsequently mixed with substrate 3,4-dihydroxy-dl-phenylalanine. Melanization catalysed by PO was measured as increase in absorbance at 492 nm for 30 min and expressed as integral of absorbance per µL of haemolymph. Data are presented as mean of three experiments ± SD; significant differences, P < 0.05, are indicated by different letters above the columns (Tukey's test).

Constitutive and activated ROS production

To further test the recognition of PLL2 by the immune system, the protein was mixed with human blood and the oxidative burst of immune cells was measured luminometrically. We observed a significant increase in ROS production in blood when PLL2 was present (Fig. 11A), whereas the controls with buffer and BSA only showed a minimal level of ROS being produced.

Fig. 11

Production of ROS in human blood treated with PLL2. (A) Human blood was mixed with PLL2, and production of ROS was measured luminometrically for 2 h at 37 °C in CPS. (B) To determine effect of PLL2 on activated neutrophils, reaction was measured in the presence of activators zymosan A (grey columns), fMLF (light green columns) and PMA (dark green columns). 20 mm Tris/HCl and 150 mm NaCl, pH 7.5, buffer was used as control. Results are expressed as integral of reaction and compared with respective control. Data are presented as mean of four experiments ± SD; significant differences, P < 0.05, are indicated by different letters above the columns; ns = not significant result (Tukey's test).

Three activators of ROS production acting through different signalling pathways were used to test the effect of PLL2 in activated blood as a simulation of ongoing infection (Fig. 11B). Treatment with PLL2 caused a significant decrease in ROS production caused by zymosan A, which activates the Toll-like receptors 2 (TLR-2). The same behaviour was reported previously for PHL [[24]]. Conversely, when N-formyl-l-methionyl-l-leucyl-l-phenylalanine (fMLF) was used as the activator, which activates ROS production by binding to specific G protein-coupled receptors on the surface of neutrophils, we did not observe any significant change in ROS production in blood treated concurrently with PLL2. No effect of PLL2 was observed also after the activation of blood with PMA, the activator of protein kinase C in the intracellular pathway leading to ROS production.

Discussion

Thanks to their carbohydrate-binding abilities, lectins play an important part in mediating attachment to biological surfaces [[35]]. In the complex life cycle of Photorhabdus bacteria, there are three basic levels at which lectins could take part: (a) the interaction with host tissues, (b) the interaction with nematode symbionts and (c) the interactions within the bacterial population. The role of PHL and PLL2 in the pathogenicity of Photorhabdus bacteria is the most interesting from an immunological point of view, whereas their role in mutualism would, for instance, impact the production of nematobacterial complexes used in the biocontrol of insect pests [[15, 16]].

The focus of this study is on O-methylated saccharides, which are newly described targets for the innate immune system [[5]]. In the context of organisms involved in the Photorhabdus life cycle, sugar O-methylation was described in bacteria and worms, but not yet in insects or humans [[3, 5]]. As both organisms that form the pathogenic complex (Photorhabdus spp. and Heterorhabditis spp.) might be capable of producing O-methylated sugars, in contrast to the defending insect or human hosts, O-methylated saccharides might be a suitable target for the immune system of the host to identify a foreign invader. Such molecules are labelled as PAMPs [[1, 2]].

We proved that the studied Photorhabdus lectins can bind to O-methylated saccharides and the O-methyl group plays and significant role in the interaction. The apparent K_D calculated for both lectins in SPR experiments with immobilized MGMR increased by four orders of magnitude in the high-density channel compared with the low-density channel. This proves that the avidity effect plays a strong role in the binding interactions of these multiple-binding-site proteins. While a low-density surface coverage is more suitable for one-to-one binding site analysis, a high density of the surface-bound ligand may represent the natural tissue surface in a more realistic way.

The role of O-methyl groups on sugar recognition by studied lectins was assessed in a set of experiments with both methylated saccharides (MGMR, 3OMG) and the corresponding nonmethylated sugar units (d-glucose, l-rhamnose). Generally, the main groups participating in protein–sugar interactions are considered to be sugar OH groups. However, multiple findings demonstrated the importance of nonpolar interactions, such as CH-π interactions between aromatic amino acids and the apolar part of carbohydrate molecules [[36-39]], and therefore, the presence of O-methyl group might have a strong influence on the binding properties.

In the SPR inhibition assay, d-glucose is not able to inhibit the binding of studied lectins. On the contrary, 3OMG exhibited an IC₅₀ in the millimolar range, demonstrating the huge effect of the 3-O-methyl group in the ligand inhibition potency. Interestingly, the inhibition was not only observed for binding to immobilized MGMR, but also for the channel with immobilized α-l-fucoside. Similarly, l-fucose and Me-α-l-Fuc interfered with lectins binding to the immobilized MGMR. This demonstrated that both types of ligands compete for the same binding sites.

For a detailed understanding of the binding site organization, we determined the X-ray structures of protein–sugar complexes for the aforementioned set of ligands. For both proteins, the nonmethylated saccharides (d-glucose, l-rhamnose) were found to recognize only polar binding sites, similar to d-galactose in the previously published PHL/d-Gal complex [[24]]. The orientation of l-rhamnose, requiring a protein loop reorientation in site 4P of PHL, together with binding through anomeric oxygen leads to the conclusion that l-rhamnose is unlikely to be the natural ligand of PLL2 or PHL. On the other hand, d-glucose is predominantly bound with the O1 oxygen oriented towards the solvent. This allows the interaction with terminal d-glucose saccharides as is demonstrated by the interaction of PLL2 with the Glcβ1-4GalNAcα disaccharide seen in the glycan array. Hence, d-glucose derivatives might be the natural target of this lectin, which is yet to be identified.

MGMR-sp2 was identified exclusively in the hydrophobic binding sites of PLL2 and PHL. The same type of binding sites is occupied by fucose moieties in the previously studied PHL complexes [[24, 40]]. This showed a general preference of the H-type sites towards hydrophobic groups and also explained the inhibitory effect of fucose on PLL2/PHL binding to the MGMR modified surface. Interestingly, 3OMG was coordinated in both H- and P-type binding sites; however, its binding mode was different in both cases. Following the recognition pattern identified for other saccharides, 3OMG is bound in the P sites via the free hydroxyl groups, while the O-methyl group is recognized by the H sites. To the best of our knowledge, 3OMG on its own was not detected among naturally occurring saccharides, and therefore, there is little biological relevance for its binding by PHL/PLL2. However, it proves the general ability of these lectins to recognize various O-methylated saccharides and moreover demonstrates a potential design for synthetic inhibitors targeting lectins of dual specificity.

β-Propeller lectins from Photorhabdus spp. share certain binding features with a well-known group of lectins called tectonins [[4, 5]]. Both are mainly β-propeller-type oligomeric proteins with several binding sites per protein subunit situated in between the blades. The H-type sites from PHL/PLL2 and the tectonin binding sites share a similar binding motif, where two aromatic residues stabilize the ligand that is further coordinated via a net of hydrogen bonds [[4, 41]]. For the Lb-Tec2 tectonin from Laccaria bicolor, the ability to recognize O-methylated saccharides was shown recently [[4]]. The function of Lb-Tec2 is proposed to be in the recognition of the pathogens. Lectins PHL and PLL2 originating in pathogenic bacteria are unlikely to have the same function, and this leads to the hypothesis that these lectins could shield sugar epitopes at the surface of bacterial and/or nematode symbiont, hiding them from the host immune system. Also, from the structural point of view, there are various differences between tectonins and PHL or PLL2. While tectonins form six-bladed beta-propellers with one set of binding sites similar to the Aleuria aurantia AAL family [[42-44]], the lectins from Photorhabdus consist of seven-bladed propellers with two sets of binding sites with different binding preferences. The oligomeric arrangement of both lectin families is also nonequal. Both subunits in the PLL2 and PHL dimers are well aligned with a common pseudosymmetry axis, while the position of subunits from the tectonin FEL dimer [[41]] is shifted. The tetrameric member of the Photorhabdus family, PLL, consists of two dimers positioned side to side [[23]], while tetrameric tectonin Lb-Tec2 forms the shape of a tetrahedron [[4]]. Due to these differences, we can assume that these two lectin families are not closely evolutionarily related, and the common arrangement of binding sites is instead a result of convergence, proving the general suitability of the given arrangement for sugar ligand binding.

Consistent with the pathogenic origin of PLL2 and PHL, proteins were recognized by the insect immune system, which responded by activating the PO system [[24]]. This process generally manifests itself as increased melanization, and it is accompanied by the production of cytotoxic molecules such as quinolic and indolic intermediates in many insect species [[28, 45]]. In contrast, the bacteria of the genus Photorhabdus and Xenorhabdus produce a wide variety of virulence factors, including the inhibitors of PO [[46, 47]]. The harmless effect of the melanization response on Photorhabdus bacteria and their nematode symbionts is further supported by insect mutants defective in the PO system, which do not suffer from increased susceptibility to infection by the nematobacterial complex Heterorhabditis–Photorhabdus [[26, 48]]. Instead of the host protective effect of induced melanization, it was suggested that the pathogen-associated products promoting the host immune response could be used by the pathogen to exploit host defences and successfully establish the infection [[24, 49]]. In this process, the resistance of the pathogen to the induced response reaction is of immense importance.

Activation of the PO system caused by PLL2 or PHL is specifically inhibited by saccharides. l-fucose and Me-α-l-Fuc inhibited the activity of both proteins and 3OMG inhibited the activation by PHL, while d-glucose had no effect. These results are consistent with the SPR inhibition assay. In contrast, the strong PLL2 inhibitor 3OMG of its own had no effect on PO activity stimulation, and moreover, d-glucose was able to inhibit the biological activity of PLL2 while exhibiting no inhibitory effect in SPR. The situation in the complex biological systems is clearly more complicated and cannot be fully substituted by a single biophysical method such as SPR. More detailed studies at the cellular level will be needed to fully understand these processes.

In vertebrates, specifically in the human blood, the Photorhabdus lectins were shown to interfere with antimicrobial response by manipulating a level of ROS. PLL2 inhibited the ROS production elicited by zymosan A in human blood similar to PHL [[24]]. It is noteworthy that the employment of ROS in the immune response is typical for vertebrates and for many insect species [[50, 51]], yet the signalling pathways leading to ROS production are better characterized in the vertebrate innate immune system. Although PLL2 is recognized by human immune cells, which react with an increased ROS production, in the presence of potent immune activators such as zymosan A, the effect of the lectin manifests itself as an inhibition of the induced response. PAMPs such as zymosan A or lipopolysaccharides are recognized by TLRs present on the surface of immune cells, and after stimulation, they activate signalling pathways, leading to the activation of NADPH oxidase and ROS production [[52, 53]]. The signalling pathway acting through formyl peptide receptor or via the activation of protein kinase C was not influenced by PLL2 at all. Current results suggest that PLL2 interferes with the TLR-2 present on neutrophils and not with other downstream components of the transduction pathway leading to ROS production, and further research can identify the precise mechanism of this interaction.

As was shown above, homologous lectins PLL2 and PHL share many properties; however, despite 81% identity, they display significant differences in several attributes. Both lectins prefer l-fucose and its derivatives as a binding partner, and they recognize O-methylated saccharides and interfere with the host immune system in a similar manner. On the other hand, the affinity of PLL2 towards saccharide ligands is generally weaker than of PHL, with the most prominent 10-fold difference in K_D(app) towards immobilized l-fucose. On the cellular level, the lectin behaviour is even more distinct. Both lectins are capable to recognize insect haemocytes, but only PHL showed haemagglutination activity with human erythrocytes [[24]]. PLL2 did not show any haemagglutination activity under the same conditions or at the double protein concentration. This nicely corresponds with their role of insect and human pathogen P. asymbiotica (PHL) or strict insect pathogen not able to infect humans P. laumondii (PLL2). The observed differences might be caused by different amino acids localized in the close vicinity of ligand-binding residues (Fig. 6C). Most of these residues are localized on the surface of the protein (Fig. 6D), which is likely to influence the protein interaction with the complex surface of the cells.

In summary, we showed that the PLL2 and PHL lectins from the pathogenic bacterial genus Photorhabdus recognize O-methylated sugars, which were recently described as PAMPs, thus potentially competing with host immune system effectors over these ligands. Moreover, these lectins interfere with the host immune system by modulating ROS production and PO activity. All the presented data suggest that both lectins can play an important role in overcoming the host defence in the early stages of infection.

Materials and methods

Materials

l-fucose and protein molecular Marker III were purchased from AppliChem (Darmstadt, Germany);d-glucose, d-galactose, N-acetyl-d-galactosamine and methyl-α-l-fucoside from Carbosynth (Compton, UK); and IPTG and LB broth from ForMedium (Norfolk, UK). d-mannose, d-mannose-agarose, 3OMG, 3,4-dihydroxy-dl-phenylalanine, α-chymotrypsin, zymosan A from Saccharomyces cerevisiae, N-Formyl-Met-Leu-Phe (fMLF), phorbol 12-myristate 13-acetate (PMA), luminol, paraformaldehyde and Triton™ X-100 were purchased from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany); Alexa Fluor® 594 phalloidin and DyLight 488 were purchased from Thermo Fisher Scientific (Waltham, MA, USA); and l-rhamnose was purchased from MERCK (Kenilworth, NJ, USA). α-l-fucose-sp-biotin was purchased from Lectinity (Moscow, Russia). Methylated disaccharide 3,6-O-Me₂-d-Glcβ1-4(2,3-O-Me₂)-l-Rhaα-O-(p-C₆H₄)-O-CH₂CH₂NH₂(MGMR-sp2) was synthesized as described previously [[54]].

PLL2 identification and cloning

The hypothetical protein Plu0734 was identified with bioinformatic analyses of the genome of Photorhabdus laumondii subsplaumondii TTO1 (Taxonomy ID: 243265, NCBI reference sequence: BX470251.1) [[13]] using the NCBI BLAST tool [[55]]. The sequence of the PLL lectin from Photorhabdus laumondii (UniProt ID: Q7N8J0) was used as a template for the analyses. A synthetic gene of plu0734 was prepared by Life Technologies (Thermo Fisher Scientific) with the codon optimized for expression in E. coli, HindIII and NdeI restriction endonuclease sites were added to the sides of the sequence. The gene was cloned into the pET25b (Novagen, Darmstadt, Germany) vector using HindIII and NdeI restriction enzymes (NEB, Ipswich, MA, USA). The created vector pET25b_plu0734 was transformed into E. coli DH5α and subsequently into the expression strain E. coli Tuner(DE3) (Novagen). The presence of the pET25b_plu0734 vector in transformed cells was confirmed by the presence of ampicillin resistance. The sequence of pET25b_plu0734 was confirmed by sequencing the reisolated vector.

PLL2 production

Stock cells E. coli Tuner(DE3) pET25b_plu0734 kept at −80 °C were inoculated in standard LB broth with the addition of 100 µg·mL⁻¹ ampicillin and were cultivated at 37 °C to an OD₆₀₀ of 0.5. Protein production was induced using 0.5 mm IPTG, and cells were cultivated for another 16 h at 18 °C. Cells were harvested by centrifugation at 12 000 g for 10 min. The cell pellet was resuspended in buffer A (20 mm Tris/HCl, 300 mm NaCl, pH 7.5) and stored at −20 °C until further use.

PLL2 purification

Cells were disrupted by sonication (VCX 500; Sonics & Materials, Inc., Newton, CT, USA). The cell lysate was separated by centrifugation at 21 000g at 4 °C for 1 h and filtrated through a filter with pore size 0.22 µm (Carl Roth, Karlsruhe, Germany). The cell lysate was then loaded onto d-mannose-agarose resin equilibrated with buffer A. The PLL2 protein was purified by affinity chromatography by the AKTA FPLC system (GE Healthcare, Buckinghamshire, UK) and eluted isocratically. The purity of eluted fractions was analysed using SDS/PAGE (12% gels stained with Coomassie Brilliant Blue R-250). The purified protein was dialysed against the working buffer (20 mm Tris/HCl, 150 mm NaCl, pH 7.5) so the buffer would be suitable for further studies.

PHL production and purification

PHL was produced, purified and lyophilized as described previously [[24]] Prior to experiments, the lyophilized protein was dissolved in the working buffer and used for further studies.

Glycan microarray

PLL2 was fluorescently labelled with DyLight 488 according to the manufacturer's instructions. Unbound dye was extensively dialysed out of the protein solution. The microarray chip (Semiotic, Moscow, Russia) containing 381 saccharide ligands, each with 6 replicates, was treated with PBS (137 mm NaCl, 2.7 mm KCl, 8 mm Na₂HPO₄, 1.47 mm KH₂PO₄, pH 7.4) with added 0.1 % Tween-20 for 15 min. The labelled protein (200 µg·mL⁻¹) was then applied to the chip surface and incubated in a humidity chamber at 37 °C for 1 h. After the incubation, the unbound protein was washed out with PBS containing 0.05 % Tween-20 and deionized water. The chip was dried by an airflow and scanned using an InnoScan1100 AL (Innopsys, Carbonne, France) with a 488-nm laser. The data were analysed with the software mapix8.2.2 (Innopsys) and an online glycan chip converter (Semiotic).

Surface plasmon resonance (SPR)

Surface plasmon resonance experiments were carried out in a BIAcore T200 instrument (GE Healthcare) using a CM5 chip with a carboxymethyldextran surface (GE Healthcare). Sugar moieties were immobilized on the chip surface using the standard amine coupling procedure according to the manufacturer's instructions. The chip surface was activated with an N-ethyl-N-(3-dimethylaminopropyl)carbodiimide/N-hydroxysuccinimide solution using HBS buffer (10 mm Hepes, 150 mm NaCl, 0.05% Tween-20, pH 7.5). MGMR-sp2 was covalently immobilized onto channel 2. channels 3 and 4 were covered with streptavidin to a final response of 1000–1300 RU. All channels were blocked with 1 m ethanolamine to block all unreacted groups. Two chips differing in their MGMR-sp2 density on channel 2 were prepared. A low-density chip with a final response of 70RU was prepared using the loading buffer (10 mm sodium acetate, pH 4). A high-density chip with a final response of 136 RU was prepared using the 10 mm Hepes, pH 7.5, loading buffer. Channel 1 untreated with any saccharide served as a control for channel 2. Biotinylated α-l-fucose was loaded into channel 4 and immobilized via streptavidin–biotin interaction to the final response of 30 RU. The same immobilization level was used in a recent study of PHL [[24]]. Channel 3 was left without a saccharide to serve as a control for channel 4.

Protein PLL2 or PHL, respectively, was injected into the chip in a working buffer (20 mm Tris/HCl, 150 mm NaCl, pH 7.5) with the addition of 0.05% Tween-20 at a flow rate of 30 μL·min⁻¹ at 25 °C. The protein contact time was 1 min for the α-l-fucoside channel or 2 min for the MGMR channel. Fifty micromolar NaOH for 30 s (flow 30 μL·min⁻¹) was used for chip regeneration, in the MGMR channel after every protein injection, and in the α-l-fucoside channel once per 12 measurements (a complete cycle for one inhibitor). To determine the apparent K_D towards the immobilized α-l-fucose, we used a protein concentration range of 0.195–50 μm. The data were evaluated with the Biacore T200 evaluation software using a steady-state approach. To determine the apparent K_D towards the MGMR on the high-density chip, we used a protein concentration range of 1.5–25 nm and the data were evaluated with the Biacore T200 evaluation software using a single-cycle kinetics approach. For SPR inhibition experiments, 10 PLL2 or 25 μm PHL, respectively, was used for the α-l-fucoside channel, and 50 nm PLL2 or 10 nm PHL, respectively, for the MGMR channel. Proteins of the given concentration were mixed with a concentration series of tested inhibitors (0.01–0.4 m). Proteins without an inhibitor were used as a control. The response of the lectin bound to the sugar surface at equilibrium was plotted against the concentration of inhibitor in order to determine IC₅₀. The response in the appropriate blank channel was subtracted prior to all evaluations.

Isothermal titration calorimetry (ITC)

Isothermal titration calorimetry experiments were performed using an AutoITC calorimeter (Malvern Panalytical, Malvern, UK). Measurements were carried out at 25 ± 0.1 °C in the working buffer (20 mm Tris/HCl, 150 mm NaCl, pH 7.5). Prior to the analyses, PLL2 was extensively dialysed against the working buffer and the dialysate was used to dissolve ligands. PHL used for the analyses was dissolved in the working buffer from the lyophilisate. The proteins (0.1 mm) in the cell was titrated with 2-μL injections of the ligand (20 mm). Titrations of buffer into buffer, ligand into buffer and buffer into the protein solution were performed as control experiments resulting in insignificant signals. Experiments for each ligand were carried out in triplicates. Integrated heat effects were evaluated by nonlinear regression using a single-site binding model in origin 7 (Microcal; Malvern Panalytical) [[56]]. Triplicate measurements were fitted simultaneously using a global-fit approach.

Analytical Ultracentrifugation (AUC)

The oligomeric state of PLL2 in solution was investigated by AUC using a ProteomeLab XL-A analytical centrifuge (Beckman Coulter, Brea, CA, USA) equipped with an An-60 Ti rotor. Prior to analyses, the protein was dialysed against the working buffer (20 mm Tris/HCl, 150 mm NaCl, pH 7.5) and the dialysate was used as an optical reference. Experiments were performed at different protein concentrations (0.1–0.25 mg·mL⁻¹). Sedimentation velocity experiments were conducted in titanium double-sector centrepiece cells (Nanolytics Instruments, Potsdam, Germany) loaded with 380 µL of the protein and 380 µL of the reference solution. Data were collected using absorbance optics at 20 °C at a rotor speed of 50 000 r.p.m. Scans were performed at 4-min intervals at 280 nm and 0.003-cm spatial resolution in a continuous scan mode. The partial specific volume of the protein together with solvent density and viscosity was calculated from the amino acid sequence and the buffer composition, respectively, using the software Sednterp (http://jphilo.mailway.com/index.htm). The sedimentation profiles were analysed with the program Sedfit 15.01. A continuous size distribution model for noninteracting discrete species was used to provide a distribution of sedimentation coefficients [[57]].

Crystallization and data collection

The PHL and PLL2 proteins were concentrated to 12.5 or 10.0 mg·mL⁻¹, respectively, using an ultrafiltration unit with a 10-kDa cut-off membrane (Vivaspin 20; Sartorius, Göttingen, Germany). The crystals were obtained using the sitting-drop vapour diffusion method. For PHL, the initial conditions were optimized as described previously [[24]], giving crystals with a precipitant containing 3.8–4.0 m NaCl, 100 mm HEPES, pH 7.5. For PLL2, the initial conditions were obtained from crystallization screen trials using commercial screens by Molecular Dimensions (Sheffield, UK) and Qiagen (Hilden, Germany) and subsequently optimized, giving crystals upon mixing the protein solution with 8–12.5% poly(ethylene glycol) 4000, 100–150 mm sodium acetate, pH 4.6. The crystals were soaked in an appropriate saccharide solution (Table 6), cryoprotected using 40% poly(ethylene glycol) 400 and frozen in liquid nitrogen. The diffraction data were collected at the BESSY II electron storage ring (Berlin-Adlershof, Germany) [[58]] or PETRA III electron storage ring (DESY, Hamburg, Germany), respectively.

Table 6. Crystallization and soaking conditions for lectin–saccharide complexes.

Complex	Precipitant used	Soaking solution/time
PLL2/3OMG	11% poly(ethylene glycol) 4000, 150 mm NaOAc, pH 4.6	50 mm 3OMG /10 min
PLL2/MGMR-sp2	12.5% poly(ethylene glycol) 4000, 100 mm NaOAc, pH 4.6	35 mm MGMR-sp2/5 min
PLL2/Glc	8% poly(ethylene glycol) 4000, 150 mm NaOAc, pH 4.6	50 mm d-Glc/15 min
PLL2/Rha	8% poly(ethylene glycol) 4000, 150 mm NaOAc, pH 4.6	50 mm l-Rha/15 min
PHL/3OMG	4 m NaCl, 100 mm HEPES, pH 7.5	10 mm 3OMG /10 min
PHL/MGMR-sp2	4 m NaCl, 100 mm HEPES, pH 7.5	10 mm MGMR-sp2/8 min
PHL/Glc	3.9 m NaCl, 100 mm HEPES, pH 7.5	200 mm d-Glc/2.5 h
PHL/Rha	3.8 m NaCl, 100 mm HEPES, pH 7.5	200 mm l-Rha/2.5 h

Structure determination

Diffraction images were processed using the programs XDS [[59]] or XDSAPP [[60]] and converted to structure factors using the program Scala from the package CCP4 v.7.0 [[61]], with 5% of the data reserved for the R_free calculation. The structures of complexes were solved by molecular replacement with MOLREP [[62]]. As the initial coordinates, the PHL structure (PDB ID: 5MXE) was used for replacement of the PHL complexes, while the structure of the PLL chain A (PDB ID: 5C9O) was used for replacement of sugar-free PLL2, and the resulting PLL2 structure subsequently served as a model for other PLL2 complexes. Refinement of the molecules was performed using REFMAC5 [[63]] alternated with manual model building in Coot v.0.8 [[64]]. Coordinates for methyl derivatives of sugar ligands were established using JLigand [[65]], included in the CCP4 software package. Ligands were placed manually using Coot, in case of multiple orientations within a binding site, their occupancy was assigned by iterative refinement using F_o-F_c differential map. Water molecules were added by coot and checked manually. The addition of alternative conformations, where necessary, resulted in final structures that were validated using the (http://molprobity.biochem.duke.edu) validation server and were deposited in the PDB database as entries 6RG2, 6RGG, 6RFZ, 6RG1, 6RGU, 6RGW, 6RGJ, and 6RGR. Molecular drawings were prepared using pymol (Schrödinger, Inc., New York, NY, USA).

Interaction with insect haemocytes

Haemolymph was collected from the sixth larval instar of G. mellonella and directly mixed on the coverslip with a drop of PBS containing phenylthiourea (1 mg·mL⁻¹; pH 7.4) to prevent melanization. Haemocytes were left to adhere for 10 min before DyLight 488-labelled PLL2 (1.6 mg·mL⁻¹) was added and incubated for 10 min. Unbound PLL2 was washed out with PBS, and subsequently, haemocytes were fixed with 3.7% paraformaldehyde, permeabilized using 0.1% Triton™ X-100 and stained with Alexa Fluor® 594 phalloidin. Samples were mounted in Vectashield® and observed using an SP8 laser scanning confocal microscope (Leica, Wetzlar, Germany).

Flow cytometry of insect haemocytes

The flow cytometric analysis of the interaction of PLL2 with insect haemocytes was performed using a BriCyte E6 flow cytometer (Mindray, Nanshan, China) with a laser excitation wavelength of 488 nm and operated by mrflow software (Mindray, Nanshan, China). A CountBright™ bead standard provided by Thermo Fisher Scientific was monitored periodically to ensure the reproducibility of scatter and fluorescence measurements. Standard measurements were found to be within all tolerances set by the manufacturer. Haemolymph (10 µL) was collected from G. mellonella larvae directly into 980 µL of cold PBS (pH 7.4), and subsequently, 10 µL of PBS or DyLight 488-labelled PLL2 (3 mg·mL⁻¹ in PBS) was added to obtain 1 mL of sample. After 10 min of incubation in the dark at room temperature, the samples were gently mixed and immediately examined by flow cytometry. The background fluorescence was set in the absence of labelled lectin, whereas the FITC-positive region corresponds to the haemocytes labelled with DyLight 488. A total of 10 000 events were collected from each sample, and cell percentages were obtained with the flow cytometry software used. The flow cytometric analysis was performed in triplicates, and an example outcome is provided in the results.

Phenoloxidase (PO) activity in insect haemolymph

The determination of PO activity in cell-free haemolymph plasma was modified according to the previously published protocol [[66]]. Haemolymph of G. mellonella was collected in tubes containing ice-cold phosphate buffer (10 mm; pH 7.0) and snap-frozen at −80 °C in 14× dilution. Haemolymph samples were melted on ice prior to measurement, gently mixed by pipetting and centrifuged (15 min; 12000 g; 4 °C) to remove haemocyte debris. The supernatant was used to determine PO activity. All subsequent steps were done on ice, unless stated otherwise. Seventy microlitre of haemolymph plasma was transferred to a 96-well plate and mixed with 5 μL of α-chymotrypsin (5 mg·mL⁻¹ in ultrapure water) or 5 μL of ultrapure water. Chymotrypsin-treated plasma was incubated 10 min at room temperature to enable pPO activation and the measurement of total PO activity. Samples with ultrapure water without α-chymotrypsin were used to determine PO activity and were kept on ice. Subsequently, 25 μL of PLL2 (4 mg·mL⁻¹) in working buffer (20 mm Tris/HCl, 150 mm NaCl, pH 7.5) and an equivalent volume of working buffer or BSA (4 mg·mL⁻¹ in working buffer) as controls were added to the wells. The melanization reaction was started by adding 50 μL of substrate 3,4-dihydroxy-dl-phenylalanine (3 mg·mL⁻¹ in phosphate buffer; pH 7.0) and measured with a plate reader Sense (Hidex, Turku, Finland) as the increase in absorbance at 492 nm at one-minute intervals. PO activity was determined as the melanization rate in 30 min of reaction and expressed as the integral of the reaction curve.

The effect of PLL2 and PHL on PO activity in the whole haemolymph was tested as published previously [[24]]. For these experiments, the freezing and centrifugation step was omitted, and PO activity was measured immediately after bleeding larvae in haemolymph containing haemocytes. To perform the inhibition study, PLL2 and PHL (4 mg·mL⁻¹) were preincubated with 0.2 m l-fucose, Me-α-l-Fuc, d-glucose or 3OMG, respectively, for 15 min at room temperature prior to the measurement of melanization. All PO assays were repeated four times, and for each replicate, different batches of tested lectins and haemolymph were used.

Reactive oxygen species (ROS) production in human blood

Samples of human blood were collected from healthy donors into tubes without anticoagulant and anonymized prior to experiments. The effect of PLL2 on ROS production in human blood was measured within 1 h of its collection, as in a previous study of PHL [[24]]. Briefly, 2 μL of fresh human blood was diluted in 173 μL of HBSS (0.137 m NaCl, 5.4 mm KCl, 0.44 mm KH₂PO₄, 0.25 mm Na₂HPO₄, 4.2 mm NaHCO₃, 1.0 mm MgSO₄, 1.3 mm CaCl₂, 5.55 mm glucose, pH 7.4) and mixed with 25 μL of PLL2 (4 mg·mL⁻¹) in working buffer (20 mm Tris/HCl, 150 mm NaCl, pH 7.5). In the control measurements, 25 μL of working buffer or BSA (4 mg·mL⁻¹) was used instead of PLL2. After 10 min, incubation at 37 °C, 25 μL of 10 mm luminol and 25 μL of activator zymosan A (2.5 mg·mL⁻¹ in HBSS), fMLF (10 µg·mL⁻¹ in HBSS) or PMA (10 µg·mL⁻¹ in HBSS) were added to activate ROS production. The activators were replaced with 25 μL of HBSS when constitutive ROS production was determined. Luminescence was measured in counts per second (CPS) with a luminometer Chameleon V (Hidex) for 2 h at 37 °C. The integrals of ROS production in PLL2-treated blood were compared with the respective controls. The effect of PLL2 on the constitutive and activated ROS production was determined in seven and four independent measurements, respectively.

Statistical analysis

Data were analysed in graphpad prism 7 (GraphPad Software, San Diego, CA, USA). To test normality and homogeneity of data, Shapiro–Wilk and Brown–Forsythe tests were used, respectively. The effect of PLL2 treatment was compared with the buffer and protein (BSA) controls using one-way ANOVA with post hoc Tukey's test. The results of ROS production in human blood were analysed using repeated-measures ANOVA, which compared controls and PLL2-treated blood with respect to the donors. Differences were considered statistically significant for P values < 0.05.

Ethics statement

Anonymized human blood of blood groups A, B, O treated with sodium citrate was purchased from Transfusion and Tissue Department, The University Hospital Brno, Czech Republic. IRB approval for the use of blood samples was not required, as confirmed by the Ethics Committee of the Masaryk University.

Acknowledgements

We would like to thank the BESSY II electron storage ring (Berlin-Adlershof, Germany) and DESY, a member of the Helmholtz Association HGF (Hamburg, Germany), for access to the synchrotron data collection facility and allocation of synchrotron radiation beamtime. We also gratefully acknowledge the following Core Facilities at CEITEC, Masaryk University, Brno, Czech Republic, for their support with obtaining the scientific data presented in this paper: the Biomolecular Interaction and Crystallization Core Facility, especially Jan Komárek, Jitka Ždánská and Jana Kosourová for their help with lectin characterization; the Proteomics Core Facility for MS analyses; and Nanobiotechnology Core facility for help with glycan array and Core facility CELLIM for the access to the confocal microscope. This work was supported by the Czech Science Foundation (project 18-18964S) and by the Ministry of Education, Youth and Sports of the Czech Republic under the project CEITEC 2020 (LQ1601). CIISB research infrastructure project LM2018127 and Czech-BioImaging large research infrastructure project LM2015062 funded by MEYS CR are also gratefully acknowledged for the financial support of the measurement at CEITEC core facilities. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflict of interest

The authors declare no conflicts of interest.

Author contributions

MW conceived the study and contributed to project administration and funding acquisition. EF, JH, PD and GP performed formal analysis and underwent investigation. PH contributed to methodology. PH and MW underwent supervision, performed validation, and wrote, reviewed and edited the manuscript. JH, EF and PD performed visualization and wrote the original draft.