Crystal structures of two camelid nanobodies raised against GldL, a component of the type IX secretion system from Flavobacterium johnsoniae

Two nanobodies raised against the cytoplasmic domain of GldL were produced, their interaction with their target was characterized, and their crystal structures were solved.

Abstract

GldL is an inner-membrane protein that is essential for the function of the type IX secretion system (T9SS) in Flavobacterium johnsoniae. The complex that it forms with GldM is supposed to act as a new rotary motor involved in the gliding motility of the bacterium. In the context of structural studies of GldL to gain information on the assembly and function of the T9SS, two camelid nanobodies were selected, produced and purified. Their interaction with the cytoplasmic domain of GldL was characterized and their crystal structures were solved. These nanobodies will be used as crystallization chaperones to help in the crystallization of the cytoplasmic domain of GldL and could also help to solve the structure of the complex using molecular replacement.

1. Introduction  

The type IX secretion system (T9SS) is a sophisticated machinery that spans both the inner and outer membranes. T9SS is widespread in the Fibrobacteres–Chlorobi–Bacteroid­etes superphylum (CFB group; Veith et al., 2017 ▸) and can deliver various substrates through the cell envelope. These substrate proteins are secreted by a two-step mechanism: they are first exported into the periplasm by the Sec pathway after the cleavage of their N-terminal signal peptide, and their conserved C-terminal domain (CTD) then triggers their secretion outside the cell by the T9SS. The T9SS has mainly been studied in the two bacterial models Porphyromonas gingivalis and Flavobacterium johnsoniae. In P. gingivalis, a human oral pathogen involved in periodontal disease, the T9SS secretes virulence factors associated with pathogenicity. In F. johnsoniae, a nonpathogenic soil bacterium that moves rapidly on solid surfaces by gliding motility, the T9SS secretes adhesins involved in the attachment of the cell to the surface. To date, 18 genes have been identified as essential for T9SS function in P. gingivalis (Lasica et al., 2017 ▸) and 12 in F. johnsoniae (McBride & Zhu, 2013 ▸). Four of these genes, porK–porL–porM–porN in P. gingivalis and gldK–gldL–gldM–gldN in F. johnsoniae, are co-transcribed and are absolutely required for T9SS function in both bacteria (Vincent et al., 2017 ▸). It has been proposed that the PorK–PorL–PorM–PorN complex forms the T9SS core and that the GldL–GldM inner-membrane complex could act as a rotary motor that energizes the system (Sato et al., 2010 ▸; Vincent et al., 2017 ▸). We recently solved the structure of the periplasmic domain of GldM (Leone et al., 2018 ▸) and, in order to obtain further structural information on the potentially new rotary motor, we intended to solve the structure of the cytoplasmic domain of GldL. Faced with unsuccessful crystallization, we decided to use nanobodies as crystallization chaperones. Nanobodies correspond to the variable domain (VHH) of heavy-chain-only immunoglobulins from Camillidae and are valuable tools in structural biology (Desmyter et al., 2015 ▸). Indeed, their use was decisive in the structure resolution of the periplasmic domain of PorM (Duhoo et al., 2017 ▸; Leone et al., 2018 ▸). Using the nanobody platform of our laboratory, two nanobodies were raised against the cytoplasmic domain of GldL; the interaction with their target was characterized and their crystal structures were solved.

2. Materials and methods  

2.1. Cloning and production of the GldL cytoplasmic domain  

The sequence of the cytoplasmic domain of GldL corresponding to residues 59–215 (GldLc) was amplified from F. johnsoniae cDNA (ATCC17061, Leibniz Institute DSMZ) and was cloned into the pLIC03 vector (kindly provided by BioXtal; unpublished work) using the primers 5′-CCGAGAACCTGTACTTCCAATCAGAACCAGTTGAGGATGAATTAG and 5′-CGGAGCTCGAATTCGGATCCTTATTATCCTTTGTTACTCATTGCAG (sequences annealed onto the gldL gene are italicized). The pLIC03 vector is derived from the pET-28a+ expression vector, in which a cassette coding for a His₆ tag and a Tobacco etch virus (TEV) protease cleavage site followed by the sacB suicide gene flanked by BsaI restriction sites is introduced downstream of the ATG start codon.

GldLc was produced in T7 Escherichia coli cells (NEB) cultured in TB medium at 37°C until the OD_600 nm reached 0.6–0.8. 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) was then added to induce expression of the protein and the temperature was decreased to 17°C for an additional 18 h. The cells were harvested by centrifugation at 4000 rev min⁻¹ for 15 min at 4°C and were resuspended in 50 mM Tris pH 8.0, 300 mM NaCl, 10 mM imidazole (buffer A) supplemented with 0.1 mg ml⁻¹ lysozyme and 1 mM phenylmethylsulfonyl fluoride (PMSF). After freezing overnight at ∼80°C, the pellet was thawed and incubated with 20 µg ml⁻¹ DNAse I and MgSO₄ for 30 min at 4°C. The cells were lysed by sonication and the insoluble pellet was removed by centrifugation at 14 000g for 30 min at 4°C. GldLc was purified from the soluble fraction by metal ion-affinity chromatography using a 5 ml HisTrap Crude (GE Healthcare) Ni²⁺-chelating column equilibrated in buffer A. GldLc was eluted with buffer A supplemented with 500 mM imidazole. The pool containing the purified GldLc was incubated overnight with 1:10(w:w) TEV protease to cleave the His₆ tag, and was concomitantly dialyzed against buffer A to remove the imidazole. The flowthrough from a second nickel-affinity chromatography column that contained the cleaved GldLc was collected, and the protein was further purified by size-exclusion chromatography on a HiLoad 16/60 Superdex 75 prep-grade column (GE Healthcare) equilibrated in 10 mM HEPES pH 7.5, 200 mM NaCl.

2.2. Generation and production of llama nanobodies  

Immunization of one llama (Lama glama; from Ardèche Lamas, France) was achieved with five injections of 1 mg purified GldLc in 10 mM HEPES pH 7.5, 200 mM NaCl. The selection and screening of nanobodies was performed as described previously (Duhoo et al., 2017 ▸) using the Nabgen Technology platform (https://nabgen.org/). After sequence analysis, two different nanobodies (which are identified here using the PDB codes of their determined structures: 7bnw and 7bnp) were chosen for further characterization. The expression and purification of 7bnw and 7bnp were performed as described previously (Conrath et al., 2009 ▸). Briefly, the nanobodies were cloned in pHEN6 vector in fusion with the pelB sequence and a His₆ tag at the N- and C-termini, respectively, purified from the periplasmic fraction of a WK6 cell culture by nickel-affinity chromatography and further purified by size-exclusion chromatography in 10 mM HEPES pH 7.5, 200 mM NaCl.

2.3. Size-exclusion chromatography–multi-angle light scattering (SEC-MALS)  

For SEC-MALS analysis, purified 7bnw and 7bnp were concentrated to 24 and 60 mg ml⁻¹, respectively. Size-exclusion chromatography was carried out on an Ultimate 3000 HPLC system (Fisher Scientific) using a Superdex 200 Increase 10/300 GL column (GE Healthcare) equilibrated in 10 mM HEPES pH 7.5, 200 mM NaCl at a flow rate of 0.6 ml min⁻¹. Detection was performed using an eight-angle light-scattering detector (DAWN 8, Wyatt Technology) and a differential refractometer (Optilab, Wyatt Technology).

2.4. Bio-layer interferometry (BLI)  

BLI assays were performed as described previously (Duhoo et al., 2017 ▸) using an Octet Red96 (Sartorius). Biotinylated 7bnw and 7bnp were loaded onto streptavidin biosensor tips (Sartorius) at 5 µg ml⁻¹. The association of each nanobody with various concentration of GldLc (9.4, 32.5, 75, 150, 300 and 600 nM) was recorded for a period of 600 s, followed by dissociation in kinetic buffer (KB; Sartorius; PBS, 0.01% BSA, 0.002% Tween 20, 0.005% sodium azide) for 900 s. Complete dissociation of the complex was obtained by three cycles of tip regeneration/neutralization (5 s in 10 mM glycine pH 1.7 followed by 5 s in KB). Control experiments were run to check that there was no nonspecific interaction between the analyte (GldLc) and control biosensors (loaded with no protein and blocked with biocytin). The curve-fitting 1:1 binding model was used to measure the constants with the Octet Red system software (version 7.1).

2.5. Crystallization, data collection and processing  

Crystallization screens using purified 7bnw (24 mg ml⁻¹) and 7bnp (60 mg ml⁻¹) were performed by the sitting-drop vapour-diffusion method at 293 K in 96-well Swissci-3 plates using Stura Footprint (Molecular Dimensions), Wizard 1 and 2 (Rigaku) and Structure Screens 1 and 2 (Molecular Dimensions). The reservoirs of the Swissci-3 plates were filled using a Tecan pipetting robot, and the nanodrops were dispensed by a Mosquito robot (TTP Labtech) by mixing different volumes (100, 200 and 300 nl) of protein solution and 100 nl precipitant solution. 7bnw crystals were obtained in 100 mM sodium acetate pH 4.5, 200 mM calcium acetate, 30%(v/v) PEG 400 and 7bnp crystals were obtained in 100 mM Tris pH 8.5, 25%(w/v) PEG 3350. The crystals were mounted in cryo-loops (Hampton CrystalCap Magnetic) and soaked for few seconds in crystallization solution supplemented with 20%(v/v) ethylene glycol before being flash-cooled in a nitrogen-gas stream at 100 K using a cryocooling device (Oxford Cryosystems).

Diffraction data were collected on the PROXIMA-1 beamline at the SOLEIL synchrotron, Saint-Aubin, France and were processed with the XDS package (Kabsch, 2010 ▸). The structures of 7bnw and 7bnp were solved by molecular replacement using MOLREP (Vagin & Teplyakov, 2010 ▸). The PDB entries with the highest sequence similarity to 7bnw and 7bnp were used as starting models (PDB entries 1t2j and 1kxq, respectively; Desmyter et al., 2002 ▸; R. K. Gaur, R. Fischer & K. M. V. Hoffmann, unpublished work). Refinement was performed with autoBUSTER (Blanc et al., 2004 ▸), and the structures were corrected with Coot (Emsley et al., 2010 ▸). For 7bnw, the final refinement steps were performed with REFMAC5 (Murshudov et al., 2011 ▸) using the twin refinement option. Model validations were performed with MolProbity (Chen et al., 2010 ▸). Data-collection and refinement statistics are given in Table 1 ▸.

Table 1. Data-collection and refinement statistics for 7bnw and 7bnp .

Values in parentheses are for the highest resolution shell.

PDB code	7bnw	7bnp
Data collection
Space group	P65	C2
a, b, c (Å)	95.9, 95.9, 118.9	42.7, 45.5, 58.1
α, β, γ (°)	90, 90, 120	90, 108.9, 90
Resolution (Å)	44.5–2.59 (2.75–2.59)	30.2–1.67 (1.79–1.69)
Unique reflections	19199 (3027)	11641 (6051)
Multiplicity	6.5 (6.6)	3.1 (3.2)
Completeness (%)	99.8 (99.0)	95.0 (89.7)
〈I/σ(I)〉	9.0 (1.1)	15.2 (1.25)
R meas (%)	15.3 (180.0)	3.7 (93.0)
CC1/2	0.994 (0.46)	0.999 (0.64)
Refinement and model quality
Resolution (Å)	44.5–2.59	30.2–1.70
Reflections	18270	11209
R/R free (%)	19.5/24.3	21.4/22.6
No. of atoms
Protein	3633	883
Water	276	60
B factors (Å2)
Protein	57.2	44.5
Water	60.3	52.4
R.m.s.d.
Bond lengths (Å)	0.013	0.009
Angle (°)	1.78	0.95
Ramachandran plot (%)
Most favoured	91.5	96.3
Allowed regions	6.6	2.8
Outliers	1.9	0.9
Twin operator	h, −h − k, −l	—
Twin fraction	0.23	—

3. Results  

3.1. Nanobody generation and binding characterization  

The recombinant cytoplasmic domain of GldL (residues 59–215; GldLc) was produced in E. coli, purified to homogeneity and used to immunize a llama. Several strong GldLc binders were identified after three rounds of panning using phage display coupled to ELISA. Two nanobodies (here called 7bnw and 7bnp) were selected according to their high affinity for GldLc and their sequence differences in the variable regions (also called the complementarity-determining regions; CDRs; Fig 1 ▸). Their CDR1s and CDR2s have the same length, but the CDR3 in 7bnp is one residue longer. 7bnw and 7bnp were produced and purified, and their interaction with GldLc was characterized. 7bnw and 7bnp both bind to GldLc in the nanomolar range (Table 2 ▸).

Figure 1.

Sequence alignment of 7bnw and 7bnp. The sequences of 7bnw and 7bnp as expressed and crystallized are displayed. Secondary-structure elements from the 7bnw and 7bnp structures are displayed above and below the alignment, respectively. The CDR1, CDR2 and CDR3 sequences are boxed. Cysteine residues are denoted by asterisks. Residues involved in the 7bnp CDR3-specific interactions and the conserved cation–π interactions, as identified by the CaPTURE program (Gallivan & Dougherty, 1999 ▸), are denoted by black triangles and spheres, respectively. Kabat numbering is used (Kabat et al., 1991 ▸). The sequence alignment was generated by Multalin (Corpet, 1988 ▸) and processed by ESPript (Robert & Gouet, 2014 ▸).

Table 2. Kinetic and thermodynamic parameters of the interactions of 7bnw and 7bnp with the periplasmic domain of GldL (GldLc).

	K d (nM)	K ass (M −1 s−1)	K diss (s−1)	Full χ2/full R 2
GldLc–7bnw	8.1 ± 0.35	4.4 × 105	3.5 × 10−4	3.87/0.93
GldLc–7bnp	32 ± 0.36	8.5 × 104	2.7 × 10−3	7.44/0.98

3.2. Nanobody structures  

In addition to serving as crystallization chaperones, nanobodies can be used to solve the structure of the complex with the protein of interest by molecular replacement. We recently used this strategy to solve the structure of the periplasmic N-terminal domain of PorM. Thus, as no structural homolog is available for GldLc, we crystallized 7bnw and 7bnp and we solved their structures by molecular replacement.

The 7bnw crystals belonged to space group P6₅ and diffracted to 2.59 Å resolution. Four molecules are present in the asymmetric unit, but 7bnw is monomeric in solution at the crystallization concentration (Fig. 2 ▸). The three N-terminal residues, as well as the five C-terminal histidine residues from the His₆ tag, could not be modelled due to weak electron density. As expected, 7bnw adopts the classical immuno­globulin fold with two β-sheets composed of four and five antiparallel β-strands: β1–β3–β7–β8 and β4–β5–β6–β9–β10. The second and last β-strands, β2 and β11, lie along one edge of the sandwich, and these strands extend to form a parallel β interaction ensuring proper closure of the β-sandwich (Fig. 3 ▸). The four molecules in the asymmetric unit are very similar, with an r.m.s.d. ranging from 0.30 to 0.85 Å on C^α atoms; the differences arise from the CDRs, which display only slightly different conformations.

Figure 2.

Size-exclusion chromatography–multi-angle light scattering (SEC-MALS) analysis of 7bnw and 7bnp. Chromatograms of purified 7bnw (red) and 7bnp (blue) are shown. The measured molar masses are ∼14.7 and ∼13.8 kDa for 7bnw and 7bnp, respectively.

Figure 3.

Crystal structure of 7bnw. Left and middle, only molecule A is displayed for clarity; right, the four molecules present in the asymmetric unit are superimposed. The CDR1, CDR2 and CDR3 according to Kabat numbering (Kabat et al., 1991 ▸) are represented in blue, green and red, respectively. The N- and C-terminal extremities are labelled N and C, respectively. The figure was generated by PyMOL (Schrödinger).

The 7bnp crystals belonged to space group C2 and diffracted to 1.67 Å resolution, with one molecule in the asymmetric unit. The three N-terminal residues, as well the C-terminal His₆-tag residues and residues at the beginning of CDR1 (residues 25–30; Supplementary Fig. S1) could not be modelled due to weak electron density. Interestingly, 7bnp displays a domain-swapped structure: the C-terminal residues (residues 101–119) are exchanged with a symmetric molecule, which reconstitutes an immunoglobulin fold (Fig. 4 ▸). The hinge loop of this domain swapping corresponds to the CDR3 (residues 98–100e). Indeed, instead of folding back on itself, the CDR3 rather projects toward the second protomer to form an ‘arm-in-arm’ dimer. The extended helical conformation of the CDR3 is stabilized by two cation–π interactions on both sides of the α-helix of both protomers, which involve the 7bnp CDR3-specific Trp100a of one protomer and the conserved Arg95 from the other protomer. The end of the CDR3 is locked to the second protomer by reconstituting the conserved double cation–π interaction involving Trp103 of one protomer and Lys45 and Tyr91 of the second protomer, thus positioning the swapped C-terminus to reconstitute the β-sandwich of the second protomer. While in the classical nanobody fold the C-terminal β-strand β11 and the short N-terminal β-strand β2 are stabilized by β-sheet interactions, no such secondary elements are present in the N- and C-termini of 7bnp, reflecting a disturbed conformation of these regions that could weaken the interactions between the C-terminus and the core of the nanobody.

Figure 4.

Domain-swapped dimer in the 7bnp crystal. 7bnp is displayed in cyan, with the CDR3 in blue; the domain-swapped symmetry mate is displayed in orange, with the CDR3 in yellow, as a worm diagram for clarity. The residues involved in cation–π interactions are displayed in stick format; the residues on both sides of the part of the CDR1 that is not visible in the electron-density map (Ala24 and Asn31) are displayed in sphere format; the N- and C-­terminal extremities are labelled N and C, respectively. A close-up view of the domain-swapping hinge is shown. The CDR3 and the preceding residues (Ala94, Asn93, Cys92 and Tyr91) are displayed in stick format, and the corresponding electron-density map (contoured at 1σ) is displayed in blue. Superimposed 7bnw is displayed in light grey with the CDR3 in red. 7bnp and 7bnw are displayed as worm diagrams for clarity. The figure was generated by PyMOL (Schrödinger).

The domain swapping results in an intricate dimer with a buried interaction surface of 2670 Ų for each monomer (of a total surface of 8376 Ų), representing 32% of the molecular surface. SEC-MALS analysis carried out with 7bnp at the crystallization concentration (60 mg ml⁻¹) revealed a molar mass of ∼13.8 kDa, which corresponds to a monomer (Fig. 2 ▸). However, as the sample is diluted during the experiment, concentration-dependent dimerization cannot be ruled out. Remarkably, 7bnp elutes well before 7bnw, despite having a smaller measured molar mass (∼13.8 and ∼14.7 kDa for 7bnp and 7bnw, respectively; Fig. 2 ▸). As nanobodies are not supposed to interact with the SEC column matrix, this observation suggests that the 7bnp molecular shape is larger than a classical nanobody fold, which could be due to partial unfolding of the C-terminus of 7bnp favoured by a weak interaction network with the core of the protein.

4. Concluding remarks  

In the context of the structural studies that we are carrying out on the T9SS, we produced the cytoplasmic domain of GldL in order to solve its crystal structure. As no crystals could be obtained, we decided to generate nanobodies that could be used as crystallization chaperones and that could help to solve the structure of the complex with the target by molecular replacement. Two nanobodies, 7bnw and 7bnp, were selected and produced. Their nanomolar-range interactions with GldL were characterized by bio-layer interferometry (BLI) and their crystal structures were solved. They both adopt the classical immunoglobulin fold with two antiparallel β-sheets. Interestingly, 7bnp presents a domain-swapped dimeric structure, with 19 C-terminal swapped residues and with the CDR3 acting as the hinge loop. Similar domain swapping was previously observed for the nanobody VHH-R9 raised against the Reactive Red 6 dye hapten (Spinelli et al., 2004 ▸). VHH-R9 harbours the shortest CDR3 among camelid nanobodies (six residues), and the seven N-terminal residues (including the first β-strand β1) were proteolytically cleaved during crystallization. It was proposed that the VHH-R9 domain swapping was driven by the strain induced by the short CDR3 together with the lability of the C-terminus that is induced by the absence of the stabilizing N-terminal β-strand β1. Similarly, 7bnp harbours a potentially labile C-terminus with a propensity for unfolding, as suggested by the SEC profiles. On the other hand, the quite long and flexible CDR3 of 7bnp (13 residues, comprising four glycine residues) can be stabilized in an extended, protruding conformation through symmetric interactions with a symmetry-mate molecule, which could further favour swapping of the C-terminus.

The domain swapping in 7bnp, with the CDR3 as the hinge loop, raises the question as to whether 7bnp could still act as a crystallization chaperone for GldL. Indeed, the CDR3 is generally important for antigen binding, and its domain-swapping-induced conformation as observed in the crystal might be not compatible with GldL interaction. However, domain swapping is a slow kinetic process that probably takes place during crystallogenesis, when high concentration conditions promote molecular contacts. Given the nanomolar affinity between 7bnp and GldL as measured by BLI, we can expect that the complex could be purified at moderate concentration and further concentrated for crystallization trials without the formation of 7bnp dimers. Large-scale production of GldL–nanobody complexes for crystallization is under way.

Supplementary Material

PDB reference: llama nanobodies raised against GldL from Flavobacterium johnsoniae, 7bnp

PDB reference: 7bnw