Purified (99%) glucosylceramide (GlcCer) from Pleurotus citrinopileatus (Tamigotake) was purchased form Nacalai Tesque (Japan). Soybean and porcine GlcCer were purchased from Avanti Polar Lipids (Alabama, United States). Fusarium oxysporum GlcCer was obtained from Professor Eliana Barreto Bergter of the Universidade Federal do Rio de Janeiro (Brazil). For immunization, GlcCer from P. citrinopileatus was dissolved in a methanol:chloroform:water mixture (16:16:5, v/v/v) and spotted on a TLC (Thin Layer Chromatography) silica glass plate (Sigma–Aldrich). Silica with adsorbed GlcCer was scraped from the plate and suspended in phosphate buffer (1 mg/ml). The other GlcCer samples were dissolved in a chloroform:methanol mixture (2:1, v/v).

For immunization, a mixture was prepared from 1E+07 Botrytis cinerea (B05.10) spores (see below) and 3.5 mg homogenized B. cinerea mycelium. For the latter, B. cinerea mycelium was grown in culture flasks containing half strength PDB (12 g/l Potato Dextrose Broth medium (Lab M, United Kingdom)) for 3 days at room temperature. The mycelium was recovered using a filter covered with Miracloth and resuspended in phosphate buffer.

Botrytis cinerea (R16, B05.10, kindly provided by Rudi Aerts, KU Leuven, Belgium), Verticillium dahliae (MUCL19210), F. culmorum (MUCL30162), F. graminearum (MUCL30161) and Alternaria brassicicola (MUCL20297) were used to test the antifungal activity of VHHs. Cultivation and spore harvesting was performed as described previously (Broekaert et al., 1990). Spores were collected from V. dahliae and B. cinerea grown on half strength PDB (12 g/l Potato Dextrose Broth medium (LABM, United Kingdom), 15 g/l select agar) and A. brassicicola, F. culmorum, and F. graminearum on 6CA-medium (cereal agar; 20 g/l baby cereals (Nestlé), 15 g/l select agar).

Recombinant Raphanus sativus AFP2 (RsAFP2) was produced in P. pastoris and purified from the supernatant as previously described (Vriens et al., 2016).

Tomato (Solanum lycopersicum) cv. Castlemart was grown at 22°C in a greenhouse with 75% relative humidity in a 16-h light/8-h dark regime.

Two llamas were subcutaneously immunized twice with 100 μg GlcCer and four times with 50 μg GlcCer together with incomplete Freund’s adjuvant (Sigma–Aldrich) following a two-weekly time schedule. Subsequently these llamas were also boosted three times with a mixture of 1E+07 B. cinerea spores and 3.5 mg mycelium following a two-weekly time schedule. All llamas remained healthy throughout the immunization process and blood samples were taken 7 days after the last GlcCer immunizations and B. cinerea boost, respectively. All vaccination experiments are executed according to EU animal welfare legislation and after approval of the local ethics (CNREEA : C2EA – 14). All animals are registered, manipulated by authorized staff, and an experienced veterinarian.

Figure 1 gives a schematic overview of the library construction. Peripheral blood mononuclear cells were prepared from 0.4 L blood of the immunized llamas using Ficoll-Hypaque according to the manufacturer’s instructions (GE Healthcare). Total RNA was extracted from these cells using the RNeasy Maxi Kit (Qiagen) and used as starting material for cDNA synthesis using the Superscript III First-strand cDNA kit (Invitrogen). First strand cDNA synthesis is followed by RT-PCR using the forward primers Lib1 (5′-GGCTGAGCTGGGTGGTCCTGG-3′) and Lib2 (5′-GGCTGAGTTTGGTGGTCCTGG-3′) and the reverse primer Lib3 (5′-GGTACGTGCTGTTGAACTGTTCC-3′) (de Haard et al., 2014). The VHH encoding gene fragments are excised from agarose gel and amplified by PCR using the following specific primers: Lib5 (5′-AAATGAGGAGACGGTGACCTGGGT) and Lib7 (5′-CATTTGAGTTGGCCTAGCCGGCCATGGCACAGGTGCAGCTGCAGGAGTCTGGGGG-3′). The gel-purified PCR-products were digested with SfiI and Eco91I and ligated into the phagemid vector pASF20. pASF20 is an expression vector that is derived from pUC119 which contains the lacZ promotor, a synthetic leader sequence, a multiple cloning site, a coliphage pIII protein coding sequence, a resistance gene for ampicillin, and an M13 phage origin for single strand production. In frame with the VHH coding sequence, the vector codes for a C-terminal (His)6 peptide tag and c-myc peptide tag. The ligation products were transformed into TG1 E. coli competent cells. Four libraries each with a clonal diversity equal to or greater than 1E+08 were obtained.

Phages were prepared according to standard methods described in Kay et al. (1996) and phage display selection was performed according to Verheesen and Laeremans (2012).

Fungal glucosylceramides from P. citrinopileatus and F. oxysporum were used as targets to isolate the phage expressing fGlcCer-binding VHHs. fGlcCer from P. citrinopileatus, dissolved in methanol:chloroform:water mixture (16:16:5, v/v/v) and the fGlcCer from F. oxysporum dissolved in chloroform:methanol (2:1, v/v) were diluted in methanol and immobilized on polystyrene Nunc Maxisorp^® 96-well plates by overnight evaporation at room temperature. Wells with coated fGlcCer (10 μg for P. citrinopileatus GlcCer and 1 μg for F. oxysporum GlcCer in 100 μl) were washed three times with phosphate buffer and blocked with 1% fish gelatin (Sigma–Aldrich). Aliquots (100 μl) of the blocked phage premix were added to the wells and incubated for 2 h at room temperature on a shaking platform. The wells were washed 15 times with phosphate buffer and bound phages were eluted with trypsin (0.1 mg/ml). Two consecutive rounds of selections were performed, one campaign using P. citrinopileatus GlcCer in the first round and F. oxysporum GlcCer in the second round and one campaign using F. oxysporum GlcCer in both rounds. Individual clones were picked from second round selections for further characterization by sequence analysis (LGC group) and primary binding assays. Periplasmic extracts were prepared from 96-well plate produced cultures. Nunc Maxisorp^® plates were coated with 1 μg F. oxysporum GlcCer as described above. The plates were blocked with 1% fish gelatin and 10-fold diluted VHH-containing periplasmic extracts were incubated for 1 h at room temperature in 1% BSA/1% Gelatin. After the plates were washed they were incubated with mouse anti-histidine antibodies and anti-mouse alkaline phosphatase antibodies (Sigma–Aldrich). Finally, the enzyme activity was determined with the alkaline phosphatase substrate (Sigma–Aldrich). The optical density was measured at 405 nm.

For further characterization, VHHs were produced in E. coli in culture flasks according to standard procedures (Marks et al., 1991). His-tagged VHHs were purified from the periplasmic extracts using the ÄKTAxpress using a combination of immobilized Nickel IMAC and desalting columns, according to the manufacturer’s instructions (GE Healthcare Life Sciences).

Nunc Maxisorp^® plates were coated with 1 μg F. oxysporum, 10 or 1 μg P. citrinopileatus and 1 μg soybean or porcine GlcCer as described above. The plates were blocked with 1% fish gelatin and purified VHHs (1 μg/ml) were incubated for 1 h at room temperature in 1% BSA/1% Gelatin. After the plates were washed they were incubated with mouse anti-histidine antibodies and anti-mouse alkaline phosphatase antibodies (Sigma–Aldrich). Finally, the enzyme activity was determined with the alkaline phosphatase substrate (Sigma–Aldrich). The optical density was measured at 405 nm.

The antifungal activity of the VHHs against a range of plant pathogenic fungi was analyzed following the protocol previously described (Osborn et al., 1995) with some modifications. Briefly, a twofold dilution series of the VHHs in sterile water was prepared in 96-well plates, after which 10 μl of VHH was mixed with 90 μl of half strength PDB containing 5E+04 spores/ml of the fungus. 96-well plates were incubated at 21°C in the dark for 48 or 72 h after which the mycelium growth was visually determined as compared to the control (i.e., water) treatment, using an inverted microscope (Motic AE21).

Detached leaves of 6-week old tomato plants were sprayed using an airbrush (VidaXL, nozzle diameter of 0.35 mm) until runoff with 60 μM VHH 41D01 resuspended in 0.2 × PBS (phosphate buffered saline) Tween20 (0.01%) or with a mock solution [0.2 × PBS Tween20 (0.01%)]. After 24 h, 10 leaves of five different plants were each inoculated with four 5-μl droplets of 1E+05 spores/ml of B. cinerea B05.10 and stored in plastic chambers on wet paper at high humidity (>99%). Lesion diameters were measured 3, 4, and 5 days post inoculation. Data were statistically analyzed with Two-way ANOVA and Bonferroni post-test. The experiment was repeated three times with similar results.

The aim of this study was to generate VHHs against fGlcCer and to select those VHHs that exhibit binding to fGlcCer. Figure 1 provides a schematic overview of the library construction. To select for fGlcCer-binding VHHs via phage display, two llamas were immunized with six boosts of commercially available fGlcCer from the fungus P. citrinopileatus and three additional boosts of a mixture of homogenized mycelium and spores of the fungus B. cinerea. All llamas remained healthy throughout the immunization process. Four phage libraries each with a clonal diversity higher than 1E+08 were obtained and used in the selection procedure (Table 1). For this, fGlcCer from P. citrinopileatus or F. oxysporum were coated and incubated with phages. Non-binding and fast-dissociating phages were washed away and specific phages remained bound to the fGlcCer. After two rounds of selections a total of 12 master plates containing 1128 individual clones were selected for further analysis by sequencing and primary binding assays (Table 1). Primary binding assays were performed via an ELISA-based binding assay with VHH-containing periplasmic extracts and GlcCer from P. citrinopileatus. Interestingly, 130 VHH-containing periplasmic extracts showed to bind fGlcCer with higher OD 405 nm values than the unrelated VHH A and VHH B or blank controls (Supplementary Figure S1), indicating that the enrichment was successful. Sequence analysis revealed 79 unique sequences from the identified set of anti-GlcCer VHH. To select the strongest binding and most specific fGlcCer interactors for further study we also took into account the binding ratio - i.e., the ratio of the absorbance of the interaction experiment with and without fGlcCer. Clones with the highest ratio were selected. Figure 2 represents the 20 most potent VHHs with a binding ratio higher than 4.

To further characterize the selected VHHs, a specificity ELISA was performed. For this, the selected VHHs were produced in E. coli and purified from the periplasmic extracts. Typical yields were 0.5–10 mg purified VHH per liter culture. VHHs 54D03 and 53F05 were not obtained in sufficient quantities and therefore not pursued further. To test binding specificity of the purified VHHs, 1 μg/ml VHH was incubated with coated GlcCer from P. citrinopileatus, F. oxysporum, Glycine max (soybean) and porcine. Although the binding capacity of the VHHs toward the fGlcCer varied strongly (Figure 3A), none of the VHHs showed binding capacity toward plant or mammalian GlcCer (Supplementary Figure S2).

Twelve VHHs were selected for further characterization and tested in vitro for antifungal activity against B. cinerea. The minimal inhibitory concentration at which 95% of the fungal growth was inhibited (IC₉₅) varied between 47 and 198 μg/ml (between 3 and 13 μM) (Figure 3A) with VHH 41D01 having the highest activity, and approximately as potent as the plant defensin RsAFP2 (Figure 3B). To test whether VHH 41D01 has broad antifungal activity we determined its inhibitory activity against the fungal plant pathogens F. culmorum, F. graminearum, V. dahliae, and A. brassicicola. VHH 41D01 also inhibited the growth of these fungi (IC₉₅ < 12 μM) (Figure 4) showing it has broad spectrum activity. RsAFP2 was taken along as positive control.

To further test the in vivo efficacy of VHH 41D01, we evaluated disease symptom development of detached tomato leaves inoculated with B. cinerea (B05.10) 24 h after spraying the leaves with 60 μM VHH 41D01 or a mock solution [0.2 × PBS Tween20 (0.01%)]. Compared to the control treatment, VHH 41D01 reduced disease symptoms induced by B. cinerea by approximately 50% (Figure 5).