A new humanized antibody is effective against pathogenic fungi in vitro

Cell lines

in our studies we used C. auris (strain DMS 21092) as fungal model, Candida albicans (ATCC 10231), clinical isolates of Candida glabrata, Aspergillus fumigatus and Fusarium solani, cervical carcinoma HeLa cells (ATCC) and CHO cells (ExpiCHO-S Cells, Thermo Fisher Scientific).

Humanization process

An analysis of murine 2G8 VH and VL against IgBlastTool⁴³ murine databases (IMGT mouse V genes, IMGT mouse D genes, IMGT mouse J genes) brought to the substitution of the non-homologous amino acids to decrease the value of the instability index calculated with ExPasy⁴⁴. The modified murine VH and VL underwent CDR-grafting following two methodologies: the first analysed the whole VH and VL sequences, the second just the Framework Regions (FRW). The latter method produced the VH and VL sequences with the lowest instability index and for this reason they were compared with the most similar ones in PDB, and the non-homologous amino acids were substituted. Some amino acids were back mutated to the murine ones.

The other two protocols use the MG-score and the humanization approach introduced in²⁷, resorting to statistical modelling of the variable regions of human antibody sequences: VH and VL regions were aligned according to the AHo scheme and juxtaposed, to yield a unique VH-VL sequence of length 298 amino acids (including gaps); then, the two protocols just differ in the choice of the CDR regions to be kept fixed: in one case, they coincided with residues chosen for the CDR-grafting protocol above, while in the second case, CDR residues were those corresponding to the Kabat numbering scheme. In Fig. 11 we show a schematic representation of both the CDR-grafting and MGM humanization process. This approach, that includes residue-residue correlations both within and between heavy and light chains, strongly depends on which residues are kept fixed, as belonging to CDR regions: starting from the murine sequence, the method implies mutating one residue at a time, choosing the mutation (at any site and to any amino-acid) that yields the greatest increase in the MG-score. This method does not guarantee to reach the highest score (i.e. the most humanlike sequence with the given CDRs), but it finds the local maximum closest to the initial murine sequence, with the smallest number of mutations (i.e. the shortest path to it). The trajectories in sequence space thus produced contain several sequences that are above the human-murine threshold²⁷ and thus constitute potential hits. These sequences were further analysed with CamSol Intrinsic Server^28,45, to check explicitly their tendency towards aggregation. Since CamSol uses local sequence information for prediction, the latter could be affected by the presence of the artificial contiguity between the C-term of the VH chain and the N-term of the VL chain in the alignment used in the design step above: for this reason, we analysed separately the VH and VL chains with CamSol, and also the scFv constructs, with a common linker between the two chains. This allows to pick, along the Kabat trajectory, separately the VH and VL sequences that have the least tendency towards aggregation, among those that score above the murine/human threshold, and also, the scFv that best copes with the requirements of having a good MG-score and a good solubility; the same was done for the other protocol, with the IMGT trajectory.

Having observed²⁷ that the correlations between VH and VL are quite smaller than correlations within VH and VL regions, we consider, for further experimental testing, the VH-VL combinations corresponding to the scFv sequences, and also the constructs obtained by combining in all possible ways the 2 VH and VL sequences thus obtained, together with those obtained with the CDR-grafting protocol. In the end, excluding repetitions, a total of 28 combinations of VH and VL were used to develop new full-length antibodies.

Antibody production

Recombinant antibodies were constructed in vitro using recombinant DNA technology. Antibody heavy and light variable fragments were synthesized based on sequences information obtained after humanization. They were then cloned into vectors for the expression in eukaryotic system and containing the human IgG1 heavy chain or the kappa light chain constant regions to create full length heavy and light chain constructs. Full length IgG1 heavy and kappa light chain genes were co-transfected in a high-throughput microscale system. 3 ml of CHO-S cells culture was transfected using the EXPICHO Expression system (Thermo Scientific™) according to the information supplied by the manufacturer. After 1 week, supernatants were collected for further analysis. The antibody production on a medium scale was carried out with the same expression system.

Antibody purification

Supernatants containing antibodies were centrifuged, filtered 0.22 μm to remove cells and debris and batch purified by affinity chromatography using TOYOPEARL AF-rProtein A HC-650F resin (Tosoh Bioscience LLC).

Antibodies were eluted with Buffer Citrate 0.1 M pH 3, neutralized with an alkaline solution and dialyzed in PBS1X with slide-A-Lyzer (Thermo Scientific™). Purity of the antibody was checked by SDS-PAGE analysis that was performed under reducing/non-reducing conditions according to the standard method. Endotoxin levels were evaluated by Pierce™ LAL Chromogenic Endotoxin Quantitation Kit (Thermo Scientific™) and the presence of aggregates analysed through HPLC-SEC analysis.

Mannan, chitin and β-1,6-glucans extraction

From an overnight inoculum of C. albicans, the cells were extracted with 3% NaOH for 1 h at 80 °C. The suspension was centrifuged; the alkali-extracted supernatant was used for the precipitation of mannan through Fehling’s reaction while the pellet was digested overnight with Kitalase lytic enzyme (FUJIFILM Wako Pure Chemical Corporation) containing β-1,3-glucanase but not chitinase or β-1,6-glucanase. The digested product was centrifuged: the supernatant composed mainly by the remaining β-1,6-glucans was dialysed against 10 mM Tris–HCl (pH 7.5) to remove digested β-1,3-glucanas while the precipitate representing chitin was solubilized. Phenol–sulfuric acid method was used for the polysaccharides quantification⁴⁶.

Competitive binding in immunofluorescence

12 µg/ml of H5K1 was added to different concentrations of laminarin in PBS 3% BSA in order to reach the following laminarin:H5K1 ratios: 40:1, 16:1, 4:1, 1:1 and 0:1. The solutions were left react at 37 °C for one hour then were put in contact with 3.0 × 10⁶ C. auris cells coming from an overnight inoculum. After 1 h the cells were centrifuged, washed, and marked for 1 h with anti-human IgG FITC antibody (Abcam, ab97224) 1:150 in PBS + 3% BSA. The cells were washed before and after fixing with paraformaldehyde (Carlo Erba) 4% for 1 h at 4 °C and finally they were resuspended in 100 µl of PBS. The samples were analysed through immunofluorescence microscope.

ELISA assay

96-well plates were coated with 50 μg/ml laminarin (Sigma-Aldrich, L9634) or extracted mannan, chitin and β-1,6-glucans in 0.05 M carbonate buffer pH 9.6 overnight at 4 °C. Nonspecific interactions were blocked with 100 µl/well blocking solution, 3% (w/v) BSA in PBS-Tween 20 (8 g/l NaCl, 0.2 g/l KH₂PO₄, 2.9 g/l Na₂HPO₄·12H₂O, 0.2 g/l KCl, 0.05% (v/v) Tween 20, pH 7.4) at 37 °C for 1 h. The plates were then incubated with decreasing concentrations of humanized mAb 2G8 (from 3.12 to 0.006 μg/ml, each concentration was tested in triplicate) in blocking solution for 2 h at 37 °C. At the same temperature and for the same time 100 µl Goat anti-human-HRP (Meridian Life Science, Inc., G5G16-0482) diluted 1:500 in blocking solution were poured in each well. After every single passage, the plates were washed 5 times with PBS-Tween 20. To reveal the binding, 100 µl of 5 mg-ABTS tablet (Roche Diagnostics) dissolved in 12 ml of sodium citrate 0.05 M, pH 3 and supplemented with 1:1000 dilution hydrogen peroxide (Carlo Erba) were added and after 15, 30, 45 and 60 min the absorbance at 405 nm was measured with a Microplate Reader (Bio-Rad). For the whole measuring time the plates were left in the dark. IC50 analysis of ELISA test was performed with Prism. The test was performed in triplicate.

Competitive binding in ELISA

Antigen-coated and blocked microplate was prepared as reported above in Materials and Methods ELISA assay. 0.06 µg/ml of H5K1 (concentration of IC₅₀) were left incubating for 1 h at 37 °C with serial dilutions of laminarin (from 400 to 0.000045 µg/ml) in blocking solution. Then, the serial dilution solution of laminarin and antibody were added to the plate and left for 1 h at 37 °C. The plate was washed and 100 µl/well of Goat anti-human-HRP (Meridian Life Science, Inc., G5G16-0482) diluted 1:500 in blocking solution was added to the plate and incubated for 1 h at 37 °C. After washing the binding was revealed with 100 µl/well of 5 mg-ABTS tablet (Roche Diagnostics) dissolved in 12 ml of sodium citrate 0.05 M, pH 3 and supplemented with 1:1000 dilution hydrogen peroxide (Carlo Erba). The absorbance was read at 405 nm after 15, 30, 45 and 60 min.

Flow cytometry and immunofluorescence

From an overnight inoculum, microorganism cells or conidia were washed with RPMI + MOPS (0.165 M, pH 7) and 3.0 × 10⁶ cells were pelleted, resuspended in Phosphate buffered saline (PBS) containing 3% (w/v) BSA (Bovine Serum Albumin Sigma Aldrich) and put in contact with 12 µg/ml of the H5K1 for 1 h at room temperature. The cells and the conidia were washed with PBS and marked with anti-human IgG FITC antibody (Abcam, ab97224) 1:150 in PBS + 3% BSA for 1 h. The cells and conidia were washed and fixed with paraformaldehyde (Carlo Erba) 4% in PBS 1 h at 4 °C. After fixing and washing with PBS, the pellet was resuspended in 400 µl of PBS and splitted in two tubes. The samples of Candida auris were analysed respectively using flow cytometry and immunofluorescence while the others just in immunofluorescence⁴⁷. A Flow cytometer (FACScanto II, BDBioscences, Erembodegem, Belgium), equipped with three lasers (488 nm, 633 nm, 405 nm) was employed to collect and quantitate FITC fluorescence from different samples. Both autofluorescence and fluorescence derived from aspecific binding of FITC-conjugated secondary Ab were quantitated by flow cytometry.

Growth inhibition assay

The growth inhibitory activity of H5K1 was tested as reported by Magliani⁴⁸, with some modifications. In brief, 150–250 cells of C. auris in 10 µl of PBS were incubated with 100 µl of hmAb at 250, 100 and 50 µg/ml (each concentration was tested in triplicate) and incubated for 18 h at 37 °C. The inhibition was evaluated by seeding the yeast on Potato Dextrose Agar (PDA) ((Sigma-Aldrich) plates. The plates were incubated at 37 °C for 48 h and the inhibition was calculated by count of the CFU. The assay was performed in triplicate in three different days^25,40.

Adhesion assay

In order to investigate a potential protective effect of H5K1 in preventing fungal adhesion to human cells, C. auris cells were left adhering to a monolayer of HeLa cells together with the humanized mAb 2G8. 1.0 × 10⁴ cervical cancer cells HeLa were resuspended in RPMI + *10% FBS (pH 7) and plated in a 96-well plate for 2 h at 37 °C + 5% CO₂. After incubation the cells not yet attached at the bottom of the wells were washed away with RPMI + MOPS (0.165 M, pH 7) and 1.0 × 10⁴ cells of C. auris resuspended in RPMI + MOPS (0.165 M, pH 7) were put on to reach a 1:1 ratio HeLa:yeast cells. Together with the yeast, 12 µg/ml of the humanized mAb 2G8 were added. PBS pH 7 was used in the control. The plate was left at 37 °C for 1 h and after washing 5 times, HeLa cells were lysed with PBS 0.1% Triton X-100 pH 7 for 15 min. RT. The suspension was plated in PDA plate and incubated at 37 °C for 48 h. This experiment has been repeated two times in triplicate^40,49.

Monocytes preparation

Cultures of macrophages were prepared from leukocyte buffy coats obtained from healthy donors. Briefly, the peripheral blood mononuclear cells were isolated by Lymphoprep density gradient medium (Stemcell Technologies) and monocytes were separated from lymphocytes by adherence to plastic dishes. Monocytes were cultured in RPMI containing 10% foetal calf serum and 1% antibiotics at 37 °C in a 5% CO₂ atmosphere for 10–12 days, at which time the monocytes had matured into macrophages and formed a monolayer.

Phagocytosis assay

From an overnight inoculum, C. auris cells were washed with RPMI + MOPS (0.165 M, pH 7) and labelled with 1 mg/ml Fluorescein-5-isothiocyanate (FITC, Sigma-Aldrich) in PBS for 10 min RT. After washing 250 µg/ml of H5K1 were left interacting for 1 h at 37 °C. Meanwhile macrophages were treated with 100 nM LysoTracker Red (Thermo Fisher Scientific, Waltham, MA, USA) for 45 min and washed with PBS. Yeasts and macrophages were counted in order to have a final ratio C. auris cells:macrophages 1:1. They were left 15, 30 and 180 min at 37 °C and then washed and fixed with paraformaldehyde (Carlo Erba) 4% in PBS 20 min at 4 °C. After washing again, the cells were scraped gently and acquired by flow cytometry. 15,000 cell events were acquired for each sample⁵⁰.

MIC assays

To evaluate the susceptibility of C. auris to CAS and AMB and their combination with H5K1, we followed the EUCAST antifungal MIC microdilution method⁵¹. Antifungal drugs were purchased from Sigma-Aldrich. CAS and AMB concentration ranges analysed were reported in EUCAST document. Those concentrations were tested alone and in combination with 0.25, 2.5, 25 and 250 µg/ml of humanized antibody. The wells with 100 µl of 2X final antifungal drugs concentration in RPMI 2% G medium were inoculated with 100 µl (1–5 × 10⁵ CFU/ml) of yeast suspension of C. auris. The monoclonal antibody was added to the samples of yeast suspension to test the drug-antibody combination. The plates were incubated at 37 °C for 24 and 48 h and read after incubation at 405 nm with a Microplate Reader (Bio-Rad). With the term MIC50 we consider the lowest concentration that inhibits the 50% of the growth compared to drug-free control and with MIC90, the lowest concentration that inhibits the 90% of the growth compared to drug-free control. The assays were performed three times in triplicate.

Time-kill curve assays

Based on MIC results a restricted range of drug concentration was used (4, 2, 1, 0.5, 0.25, 0.125 and 0.0625 µg/ml for caspofungin and 4, 2, 1, 0.5, 0.25, 0.125, 0.0625 and 0.03125 µg/ml for amphotericin B). In each well of a 96-well plate, 100 µl of 2X final antifungal drugs concentration in RPMI 2% G medium and 100 µl containing 1–5 × 10⁵ CFU/ml of C. auris suspension were put together. 250 µg/ml of humanized antibody was added to see the effect of the combination. 5 mM phosphate buffer was used as control. The plates were incubated at 37 °C for 0, 2, 4, 6, 8, 24 and 48 h. The samples were spread (tenfold dilutions were used when necessary) in PDA plates and left at 37 °C for 48 h before the CFU counting. Time-kills for strain-drug combination were performed a single time.

Statistical analysis

Data were plotted and analysed by GraphPad Prism 8 software. The statistical significance in adhesion and phagocytosis assays were assessed through a two-tailed Student’s t test while in growth inhibition results, the statistical significance was assessed by One-way Anova test.