*Article* **Semi-Automated Cell Panning for Efficient Isolation of FGFR3-Targeting Antibody**

**Byeongkwi Min 1,2 , Minyoung Yoo <sup>2</sup> , Hyeree Kim 1,3, Minjung Cho <sup>2</sup> , Do-Hyun Nam 1,2,4,\* and Yeup Yoon 1,2,3,5,\***


**Abstract:** Phage display technology is a widely used practical tool for isolating binding molecules against the desired targets in phage libraries. In the case of targeting the membrane protein with its natural conformation, conventional bio-panning has limitations on the efficient screening of the functionally relevant antibodies. To enrich the single-chain variable fragment (scFv) pools for recognizing the natural conformation of the membrane targets, the conventional bio-panning and screening process was modified to include the semi-automated cell panning protocol. Using FGFR3-overexpressing patient-derived cancer cells, biotin-X-DHPE was introduced and coupled to Streptavidin-coated magnetic beads for use in the solution-phage bio-panning procedure. The resulting clones of scFv were compared to the diversity of the binding region, especially on CDR-H3. The clones enriched further by cell-based panning procedure possessed a similar binding site and the CDR-H3 loop structure. The resulting antibodies inhibited cell growth and induced target degradation. This process may be a useful tool for screening biologically related antibodies that recognize natural conformational structure on cell membrane protein. Furthermore, cell-based panning has the potential to further expand to a high-throughput screening (HTS) system and automation process.

**Keywords:** phage display; cell-based panning; semi-automated cell panning; FGFR3-specific antibody

#### **1. Introduction**

Phage display, first described in 1985, is a practical tool for displaying proteins or peptides of interest in bacteriophage through fusion with viral envelope proteins [1–3]. Phage libraries are used to select and isolate binding molecules with high affinity for the target antigen with applications in monoclonal antibody (mAb) discovery, affinity maturation, and humanization [4,5]. Bio-panning for affinity selection has been used to isolate target protein-binding molecules from phage libraries [1,6,7]. The bio-panning procedure includes four major steps for phage selection: (i) Incubating and binding the phage library with the desired target; (ii) washing for non-binding and non-specific phage removal; (iii) eluting the specific phage binders; (iv) amplifying for eluted phages through Escherichia coli re-infection [8–11].

Conventional bio-panning has been based on various selection methods such as solidphase for immobilized purified antigen, solution-phase using biotinylated antigen, and whole cell panning (WCP) [2,12]. The solid-phase selection is a fairly straightforward technique; however, the antigen must be presented in the correct conformation. Otherwise,

**Citation:** Min, B.; Yoo, M.; Kim, H.; Cho, M.; Nam, D.-H.; Yoon, Y. Semi-Automated Cell Panning for Efficient Isolation of FGFR3-Targeting Antibody. *Int. J. Mol. Sci.* **2021**, *22*, 6240. https://doi.org/10.3390/ ijms22126240

Academic Editors: Annamaria Sandomenico and Menotti Ruvo

Received: 3 May 2021 Accepted: 7 June 2021 Published: 9 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the binders could recognize the epitopes that are naturally masked in the native form [12,13]. WCP was applied to isolated phage binders using cells intact with cell membrane proteins such as G-protein-coupled receptors, ligand-gate ion channels, receptor tyrosine kinases, and immunoglobulin-like receptors [3]. Although WCP may select non-specific phage binders to off-target cell surface protein, it could potentially enrich the phage binders recognizing the naturally exposed epitope on the cell surface [12,14,15].

We utilized the cell-based panning process in addition to the conventional bio-panning to enrich the phage binders specific for FGFR3 with the appropriate biological functions. The semi-automated cell panning method was optimized to maintain the advantages of the WCP by labeling a cell membrane-like substance (biotin-X-DHPE) and Streptavidincoated magnetic beads without damaging the membrane proteins of intact cells [16]. Using FGFR3-overexpressing patients-derived cells (glioblastoma, GBM), biotin-X-DHPE and Streptavidin-coated magnetic beads were coupled to use in the solution-phage bio-panning procedure. It was applied to an automatic mechanical system using a magnetic particle processor (Thermo Fisher Scientific, KingFisher™ Flex, Waltham, MA, USA) to efficiently perform and reduce the screening time for WCP [17–19].

Fibroblast growth factors (FGFs) and their receptor (FGFR) signaling have functional roles in the regulation of cell proliferation, differentiation, and apoptosis. FGFR3 consists of an extracellular domain containing three immunoglobulin-like (Ig-like) domains (D1–D3), a transmembrane domain, and two intracellular tyrosine kinase domains. FGFR3 has two main splice variants FGFR3-IIIb and -IIIc. Aberrantly activated and/or overexpressed FGFR3 has been implicated in various human malignancies [20–25]. Several therapeutic antibodies targeting FGFR3 have been used in clinical development, such as monoclonal antibodies (Vofatamab, Rainier Therapeutics, San Leandro, CA, USA) or antibody-drug conjugates (LY3076226, Eil Lilly) [26–28].

In this study, we successfully selected FGFR3 antibodies with relevant biological function from a synthetic human single-chain variable fragment (scFv) phage library through the introduction of semi-automated cell panning. Sequence analysis and Fv modeling of the selected scFvs were performed, and sequence and structural similarity of CDR-H3 were analyzed using multiple alignments. In individual scFv, the respective IgG antibodies were generated based on scFvs and demonstrated the specific binding properties and biological functions such as inhibition of cell viability and target degradation. The introduction of a semi-automated cell panning process may be an efficient tool for selecting antibodies with functionality and specificity for cell membrane proteins in the generation of antibodies and antibody engineering.

#### **2. Results**

#### *2.1. Introduction Strategy of Semi-Automated Cell Panning in Addition to Conventional Bio-Panning*

We immobilized live cells using the complex of biotin-X-DHPE and Streptavidincoated magnetic beads. This implied that the phospholipid-magnetic beads complexlabeled cells can be applied to magnetic beads-based bio-panning, as well as to the automated magnetic particle processor for semi-automated cell panning (Figure 1A). To efficiently enrich scFv binders, a semi-automated cell panning process was combined with conventional bio-panning. The scFv binders recognize the natural conformational structure expressed on the cell surface. First, 3–5 rounds of conventional bio-panning such as solid phase (e.g., immobilized antigen coating) or solution phase (e.g., magnetic beads-based with biotinylated antigen) were performed to enrich the phage pool bound to the purified protein. Next, the isolated phage pools underwent semi-automated cell panning to finally isolate scFv binders that recognize the cell surface 3D structure (Figure 1B).

binding characteristics using the bio-panning procedure (time of bio-panning: >8 weeks including repeated experiments). Conversely, in the case of semi-automated cell panning, clones that show biologically relevant properties and binding properties during the biopanning procedure are enriched. It minimizes repetition of the bio-panning cycle in the in vitro binding test step (e.g., affinity or cell binding assay by ELISA) because unwanted clones are not selected and hence, the process is less time-consuming. Therefore, the required time and the burden of economic costs efficiently decrease because the bio-

panning step is not re-performed (time of bio-panning: 3–4 weeks).

**Figure 1.** (**A**) Scheme for attachment of magnetic beads to live cells for applying bio-panning. Biotin-X-DHPE, a biotin tagged material, is a cell membrane-like substance composed of a hydrophilic head and a hydrophobic tail. Biotin-X-DHPE reacted with Streptavidin-coated magnetic beads to form a complex that is attached to the surface of living cells. (**B**) Bio-panning workflow combined with semi-automated cell panning. Binding to the rescued phage pool and target protein (Step 1–2). Wash to remove non-binding phage and then elute and amplify phage that binds to target protein (Step 3–5). Phage pool that recognizes naive target protein is isolated and screened by performing semi-automated cell panning on target-expressing cells (Step 6–8). Images created using BioRender. **Figure 1.** (**A**) Scheme for attachment of magnetic beads to live cells for applying bio-panning. Biotin-X-DHPE, a biotin tagged material, is a cell membrane-like substance composed of a hydrophilic head and a hydrophobic tail. Biotin-X-DHPE reacted with Streptavidin-coated magnetic beads to form a complex that is attached to the surface of living cells. (**B**) Bio-panning workflow combined with semi-automated cell panning. Binding to the rescued phage pool and target protein (Step 1–2). Wash to remove non-binding phage and then elute and amplify phage that binds to target protein (Step 3–5). Phage pool that recognizes naive target protein is isolated and screened by performing semi-automated cell panning on target-expressing cells (Step 6–8). Images created using BioRender.

*2.2. Optimal Immobilization Condition of Cell for Applying Semi-Automated Cell Panning*  To confirm whether biotin-X-DHPE affects the expression level or cell viability of the target protein on the cell surface after conjugation to the cell membrane, the conjugation of biotin-X-DHPE to cell membranes and attachment into living cells were assessed by FACS analysis. Biotin-X-DHPE was detected using Streptavidin-FITC (SA-FITC), FGFR3 was detected using anti-FGFR3-PE-conjugated control antibody, and histogram (shift, %) of FACS analysis was compared. The percent change in FITC intensity was approximately more than 80% after incubation with biotin-X-DHPE while that of PE was not changed, indicating that most of the cells were efficiently labeled with biotin-X-DHPE with no expressional change of FGFR3. Additionally, the FGFR3 expression level before and after To efficiently select clones that recognize both the purified target protein and cell surface membrane protein from a synthetic human scFv phage library, semi-automated cell panning was performed following conventional bio-panning [29]. In the case of conventional bio-panning such as bio-panning using purified protein or WCP, since clones are initially selected by a single method, it is difficult to select clones with desired binding characteristics using the bio-panning procedure (time of bio-panning: >8 weeks including repeated experiments). Conversely, in the case of semi-automated cell panning, clones that show biologically relevant properties and binding properties during the bio-panning procedure are enriched. It minimizes repetition of the bio-panning cycle in the in vitro binding test step (e.g., affinity or cell binding assay by ELISA) because unwanted clones are not selected and hence, the process is less time-consuming. Therefore, the required time and the burden of economic costs efficiently decrease because the bio-panning step is not re-performed (time of bio-panning: 3–4 weeks).

#### *2.2. Optimal Immobilization Condition of Cell for Applying Semi-Automated Cell Panning*

To confirm whether biotin-X-DHPE affects the expression level or cell viability of the target protein on the cell surface after conjugation to the cell membrane, the conjugation of biotin-X-DHPE to cell membranes and attachment into living cells were assessed by FACS analysis. Biotin-X-DHPE was detected using Streptavidin-FITC (SA-FITC), FGFR3 was

detected using anti-FGFR3-PE-conjugated control antibody, and histogram (shift, %) of FACS analysis was compared. The percent change in FITC intensity was approximately more than 80% after incubation with biotin-X-DHPE while that of PE was not changed, indicating that most of the cells were efficiently labeled with biotin-X-DHPE with no expressional change of FGFR3. Additionally, the FGFR3 expression level before and after treatment with biotin-X-DHPE remained unaltered, irrespective of the FGFR3 expression level on the cell surface (Figure 2A). On comparing the reaction time with the temperature at which biotin-X-DHPE is most effectively inserted into living cells, it was confirmed that more than 80% of biotin-X-DHPE was most attached to the cell surface under the reaction conditions of 30 min at room temperature using FACS analysis (Figure 2B). The labeling buffer was optimized by comparing various reaction buffers such as PBS (pH 7.4), 2 mM EDTA (0.1% BSA), and Pluronic F-68 buffer during cell labeling. Pluronic F-68 buffer was found to ensure more efficient cell labeling compared to other buffers under the same conditions (magnetic beads attached cells at more than 65%) (Figure 2C). Since it is a nonionic surfactant, it is considered useful for reducing the formation of bubbles that occur during agitation and incubation and for reducing the adhesion of cells to materials such as glass. considered useful for reducing the formation of bubbles that occur during agitation and incubation and for reducing the adhesion of cells to materials such as glass. There are direct or indirect methods for labeling cells with biotin-X-DHPE and Streptavidin-coated magnetic beads (Figure S1). In the direct method, biotin-X-DHPE and Streptavidin-coated magnetic beads are incubated prior to forming a complex, and then the complex is attached to the cell. In the indirect method, biotin-X-DHPE is first attached to the living cells and then further conjugated with the beads. When comparing the two methods, the direct method showed a significantly higher labeling efficiency than the indirect method. The direct method showed approximately 85% or more labeled yield, while the indirect method showed less than 20% yield (Figure 2D). Furthermore, it was confirmed through microscopic observation that a complex consisting of biotin-X-DHPE and Streptavidin-coated magnetic beads was attached to the living cells (Figure 2E). The bio-panning scheme was modified from the conventional semi-automation panning to accommodate the semi-automated cell panning procedure. The dedicated software protocol for the optimal semi-automated cell panning process was modified, and the appropriate detailed scheme is included in Figure S2.

treatment with biotin-X-DHPE remained unaltered, irrespective of the FGFR3 expression level on the cell surface (Figure 2A). On comparing the reaction time with the temperature at which biotin-X-DHPE is most effectively inserted into living cells, it was confirmed that more than 80% of biotin-X-DHPE was most attached to the cell surface under the reaction conditions of 30 min at room temperature using FACS analysis (Figure 2B). The labeling buffer was optimized by comparing various reaction buffers such as PBS (pH 7.4), 2 mM EDTA (0.1% BSA), and Pluronic F-68 buffer during cell labeling. Pluronic F-68 buffer was found to ensure more efficient cell labeling compared to other buffers under the same conditions (magnetic beads attached cells at more than 65%) (Figure 2C). Since it is a nonionic surfactant, it is

*Int. J. Mol. Sci.* **2021**, *22*, 6240 5 of 20

**Figure 2.** (**A**) Following biotin-X-DHPE labeling on FGFR3-overexpressing cells (PDC #1), FGFR3 expression levels on the cell surface were analyzed by FACS analysis. Biotin-X-DHPE was detected using Streptavidin-FITC, and FGFR3 was detected using PE-direct conjugated antibody. (**B**) The optimal reaction temperature and time for labeling the living cells with biotin-X-DHPE were analyzed by FACS analysis. (**C**) Optimal reaction buffer analysis for attaching magnetic beads on cells. (**D**) The capture yield of live cells attached with magnetic beads was compared using both direct and indirect method through cell counting and microscopic observation. \*\*\* *p* < **Figure 2.** (**A**) Following biotin-X-DHPE labeling on FGFR3-overexpressing cells (PDC #1), FGFR3 expression levels on the cell surface were analyzed by FACS analysis. Biotin-X-DHPE was detected using Streptavidin-FITC, and FGFR3 was detected using PE-direct conjugated antibody. (**B**) The optimal reaction temperature and time for labeling the living cells with biotin-X-DHPE were analyzed by FACS analysis. (**C**) Optimal reaction buffer analysis for attaching magnetic beads on cells. (**D**) The capture yield of live cells attached with magnetic beads was compared using both direct and indirect method through cell counting and microscopic observation. \*\*\* *p* < 0.001 paired *T* test. (**E**) Microscopic observation of live cells attached with magnetic beads (magnification 400×).

There are direct or indirect methods for labeling cells with biotin-X-DHPE and Streptavidin-coated magnetic beads (Figure S1). In the direct method, biotin-X-DHPE and Streptavidin-coated magnetic beads are incubated prior to forming a complex, and then the complex is attached to the cell. In the indirect method, biotin-X-DHPE is first attached to the living cells and then further conjugated with the beads. When comparing the two methods, the direct method showed a significantly higher labeling efficiency than the indirect method. The direct method showed approximately 85% or more labeled yield, while the indirect method showed less than 20% yield (Figure 2D). Furthermore, it was

confirmed through microscopic observation that a complex consisting of biotin-X-DHPE and Streptavidin-coated magnetic beads was attached to the living cells (Figure 2E). The bio-panning scheme was modified from the conventional semi-automation panning to accommodate the semi-automated cell panning procedure. The dedicated software protocol for the optimal semi-automated cell panning process was modified, and the appropriate detailed scheme is included in Figure S2.

#### *2.3. Isolation of FGFR3-Specific Clones through Introduction of Semi-Automated Cell Panning*

We performed two conventional bio-panning: (1) Solution phase selection with a biotinylated antigen using semi-automated bio-panning; (2) solid phase selection with immobilization antigen coating using purified human FGFR3-IIIc. In the solution phase, semi-automated bio-panning using biotinylated human FGFR3-IIIc was performed using a magnetic particle processor. Using an automated processor and the included driving software, biotinylated FGFR3-IIIc and Streptavidin-coated magnetic beads were combined and then incubated with the rescued phage pool. Next, to collect FGFR3-IIIc-specific phage pools, the solution was washed and eluted at the designated plate. By performing five rounds of bio-panning, the scFv binders specific for human FGFR3-IIIc were amplified in the solution phase. In the solid phase, the process of amplifying the FGFR3-specific scFv binders was performed four times by immobilizing human FGFR3-IIIc to the immuno-tube and then treating the phage pool. The output to input phage titer ratio was calculated for each bio-panning round to confirm the amplification of the FGFR3-specific scFv binder pool. The ratio was improved after five rounds or four rounds of bio-panning compared to that after the first round (Table 1). Using conventional panning, 376 and 658 clones were screened and then 8 and 11 clones were isolated in the solution and solid phase, respectively, (cut-off > 2, relative O.D) with different sequences via ELISA screening analysis (Table 2).



**Table 2.** Summary of sequenced clones obtained from conventional bio-panning and semi-automated bio-panning.


Semi-automated cell panning \*: magnetic beads-based semi-automated panning (solution phase) + semi-automated cell panning; semiautomated cell panning \*\*: immobilized antigen panning (solid phase) + semi-automated cell panning.

> To select clones, which recognize the surface FGFR3 protein of cells (PDC #1), semiautomated cell panning was performed for each phage pool that specifically binds to

FGFR3-IIIc protein purified by two conventional bio-panning method, and input and output phages were titrated (Table 1).

We compared the binding efficiency of each phage pool rescued from conventional bio-panning and additional semi-automated cell panning to FGFR3-overexpressing cells (PDC #1). Each output phage pool was collected through PEG precipitation and titration, and the same amount of phage particles was bound to FGFR3-overexpressing cells, and then the fluorescence intensity was compared using flow cytometry. The output phage pool in which semi-automated cell panning was further introduced had partially enhanced the selectivity for the FGFR3-overexpressing cells as compared with the output phage pool in which conventional bio-panning was performed (Figure 3A). *Int. J. Mol. Sci.* **2021**, *22*, 6240 8 of 20

**Figure 3.** (**A**) Phage pool binding to FGFR3-overexpressing cells in a flow cytometer; comparison of cell surface binding between amplified phage pools obtained using conventional bio-panning (blue histogram) only or with the introduction of cell panning (red histogram). Comparison of CDR-H3 sequence similarity of isolated antibodies through cell panning introduction. (**B**) Phylogenetic tree of similar sequences built on the basis of CDR-H3 (Kabat numbering) using CLUSTAL W multiple sequence alignment programs. Each phylogenetic tree was analyzed for 13 clones selected through conventional bio-panning only and 6 clones selected through introduction of cell panning. The A1D06, S2D05, S3A06, and S3B09 clones were grouped into "Clade A." (**C**) Comparison of CDR-H3 loop structure alignment using POSA analysis (interactive multiple protein structure alignment). The CDR-H3 of the A1D06 (yellow), S2D05 (green), S3A06 (cyan), and S3B09 (magenta) clones of Clade A were aligned based on the VH region in the Fv modeling annotation. CDR-H3 loop structure and length are different for each clone (gray, framework). **Figure 3.** (**A**) Phage pool binding to FGFR3-overexpressing cells in a flow cytometer; comparison of cell surface binding between amplified phage pools obtained using conventional bio-panning (blue histogram) only or with the introduction of cell panning (red histogram). Comparison of CDR-H3 sequence similarity of isolated antibodies through cell panning introduction. (**B**) Phylogenetic tree of similar sequences built on the basis of CDR-H3 (Kabat numbering) using CLUSTAL W multiple sequence alignment programs. Each phylogenetic tree was analyzed for 13 clones selected through conventional bio-panning only and 6 clones selected through introduction of cell panning. The A1D06, S2D05, S3A06, and S3B09 clones were grouped into "Clade A". (**C**) Comparison of CDR-H3 loop structure alignment using POSA analysis (interactive multiple protein structure alignment). The CDR-H3 of the A1D06 (yellow), S2D05 (green), S3A06 (cyan), and S3B09 (magenta) clones of Clade A were aligned based on the VH region in the Fv modeling annotation. CDR-H3 loop structure and length are different for each clone (gray, framework).

After the introduction of semi-automated cell panning procedure, we screened 188 and 372 individual clones and isolated 2 and 4 clones each (cut-off > 2, relative O.D) with different sequences via ELISA screening analysis (Table 2). The sequences of the six clones selected by introduction of semi-automated cell panning are included in the 19 clones of conventional bio-panning sequences. Clones with the characteristics of binding to the surface of FGFR3-overexpressing cells were enriched in the phage pool selectively amplified for purified FGFR3-IIIc by conventional bio-panning.

#### *2.4. Complementarity-Determining Regions of the Heavy Chain (CDR-H3) Sequence Analysis and Structure Homology Alignment Using Variable Fragment (Fv) Modeling*

The sequences of CDR-H3 (Kabat numbering) were compared in the six CDR regions (CDR-L1, L2, L3, H1, H2, and H3) of the antibody [30–32]. The CDR-H3 region of an antibody has the most sequence and structural diversity and is known to play the most important role in antigen-binding specificity among the six CDRs [33–36]. Moreover, the synthetic human scFv library used in this study also has the most diverse sequence and structural features of CDR-H3 [29].

The nucleotide and amino acid sequence of phagemid vector for scFvs (VH-linker-VL) were analyzed using 19 clones of different sequences isolated by conventional bio-panning and 6 clones (included in 19 sequences) selected by semi-automated cell panning (cell panning) established in this study. The CDR-H3 sequences of 19 different clones selected by bio-panning were confirmed to have various lengths and amino acid configurations. Six clones (A1D06, S2D05, S3A06, S3B09, S1E12, and A1A10) selected by cell panning were marked with gray highlights (Figure 3B). The CDR-H3 sequences of the six clones were compared for similarity through phylogenetic tree analysis using a Clustal W (multiple sequence alignment programs) tool. Clones A1D06, S2D05, S3A06, and S3B09 (Clade A) were significantly similar in sequence and length to CDR-H3, but S1E12 and A1A10 showed differences (Figure 3B). To compare the CDR-H3 structural similarity between the selected six clones, Fv modeling was first performed using the SAbPred (a structure-based antibody prediction server) tool (Figure S3) [37]. The structure alignment of the CDR-H3 loop was performed using the POSA (partial order structure alignment) tool for the whole variable heavy chain (VH), and the structural similarity of CDR-H3 was analyzed using a flexible multiple structure alignment approach for Clade A [38,39]. The average of rootmean-square deviation (RMSD) in the VH region (tertiary structure) between each clone in Clade A was less than 1 Å, which is the general criterion considered for significant similarity (Figure 3C) [40]. Therefore, clones of Clade A with sequence and structurally similarity to CDR-H3 were considered because they have similar binding properties against FGFR3. However, the overall CDR-H3 sequence and structural similarity pattern of Clade A may not be completely identical. It has a limitation in that there may be ambiguous contradictions due to the difference between the methodology of the two alignments.

#### *2.5. Generation of Anti-FGFR3 Antibodies (IgG) and Analysis of Physicochemical Properties*

The VH and VL sequences of the anti-FGFR3 scFv binders selected using cell panning were analyzed, and each sequence was cloned into heavy (Immunoglobulin G1, IgG1) and light chain (lambda) expression vectors for reformatting IgG (Figure S4). Anti-FGFR3 antibody clones were produced by co-transfection of heavy and light chain vectors using the Expi293 expression system and highly purified clones were obtained using an affinity chromatography column. The production yield of each clone was as follows: A1D06, 20 mg/L; S2D05, 120 mg/L; S3A06, 161 mg/L; S3B09, 100 mg/L; S1E12, 180 mg/L; A1A10, 38 mg/L after 6 days of incubation.

Using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis, the assembly of heavy and light chains was confirmed and the molecular weight of whole IgG was analyzed. Under non-reducing conditions, all clones were observed to have a molecular weight of approximately 150 kDa, whereas under reducing conditions, heavy and light chains were observed to have molecular weights of approximately 50 kDa and 25 kDa, respectively (Figure 4A). By using size exclusion-high-performance liquid

chromatography (SEC-HPLC), all clones were analyzed to confirm their purity, and their physical properties showed to be 95% or more (Figure 4B). It indicates that there was no loss of monomer such as through fragmentation (i.e., low-molecular-weight, LMW species) or aggregation (i.e., high-molecular-weight, HMW species). and 25 kDa, respectively (Figure 4A). By using size exclusion-high-performance liquid chromatography (SEC-HPLC), all clones were analyzed to confirm their purity, and their physical properties showed to be 95% or more (Figure 4B). It indicates that there was no loss of monomer such as through fragmentation (i.e., low-molecular-weight, LMW species) or aggregation (i.e., high-molecular-weight, HMW species).

*2.5. Generation of Anti-FGFR3 Antibodies (IgG) and Analysis of Physicochemical Properties* 

The VH and VL sequences of the anti-FGFR3 scFv binders selected using cell panning were analyzed, and each sequence was cloned into heavy (Immunoglobulin G1, IgG1) and light chain (lambda) expression vectors for reformatting IgG (Figure S4). Anti-FGFR3 antibody clones were produced by co-transfection of heavy and light chain vectors using the Expi293 expression system and highly purified clones were obtained using an affinity chromatography column. The production yield of each clone was as follows: A1D06, 20 mg/L; S2D05, 120 mg/L; S3A06, 161 mg/L; S3B09, 100 mg/L; S1E12, 180 mg/L; A1A10, 38

Using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis, the assembly of heavy and light chains was confirmed and the molecular weight of whole IgG was analyzed. Under non-reducing conditions, all clones were observed to have a molecular weight of approximately 150 kDa, whereas under reducing conditions, heavy and light chains were observed to have molecular weights of approximately 50 kDa

*Int. J. Mol. Sci.* **2021**, *22*, 6240 9 of 20

mg/L after 6 days of incubation.

**Figure 4.** Physicochemical property analysis of anti-FGFR3 antibodies derived from cell panning. (**A**) Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of purified antibodies in non-reducing and reducing conditions. In the non-reducing condition, 150 kDa band indicates the whole IgG, and in the reducing condition, 25 kDa and 50 kDa indicate the light chain and heavy chain of the antibody, respectively. (**B**) The purity of the anti-FGFR3 antibodies were analyzed using size exclusion-high performance liquid chromatography (SEC-HPLC). All anti-FGFR3 antibodies showed more than 95% purity of monomer. **Figure 4.** Physicochemical property analysis of anti-FGFR3 antibodies derived from cell panning. (**A**) Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of purified antibodies in non-reducing and reducing conditions. In the non-reducing condition, 150 kDa band indicates the whole IgG, and in the reducing condition, 25 kDa and 50 kDa indicate the light chain and heavy chain of the antibody, respectively. (**B**) The purity of the anti-FGFR3 antibodies were analyzed using size exclusion-high performance liquid chromatography (SEC-HPLC). All anti-FGFR3 antibodies showed more than 95% purity of monomer.

To verify the target-specific binding ability of the anti-FGFR3 antibodies, purified human FGFR3 isoform (FGFR3-IIIb, -IIIc) and several species of FGFR3 (mouse FGFR3- IIIc, cynomolgus monkey FGFR3-IIIc) were used in an ELISA assay. All clones showed specific binding for the purified human FGFR3-IIIc antigen, which were used for biopanning, dependent on their concentration, but not to the negative protein (Fc-tagged). The six clones were considered to have apparent specificity for purified human FGFR3- IIIc (Figure 5A). Clade A (A1D06, S2D05, S3A06, and S3B09 clones), clustered based on the similarity of CDR-H3 sequences, was confirmed to have cross-reactivity between human FGFR3-IIIb and cynomolgus FGFR3-IIIc, but did not show specificity for mouse FGFR3-IIIc. The clone S1E12 was not specific to human FGFR3-IIIb, but it had apparent binding specificity for mouse FGFR3-IIIc and cynomolgus FGFR3-IIIc. It was hypothesized that this clone has a binding epitope for the IIIc region of the FGFR3 extracellular domain. The first half of the Ig III domain of FGFR3 is encoded by the invariant exon (IIIa), and the other half is the region where the variant occurs by splicing to IIIb or IIIc [24]. The clone A1A10 was estimated to have a homologous region of human (IIIb and IIIc), mouse, and cynomolgus FGFR3 as an epitope, as it binds to the other FGFR3 proteins as well as human FGFR-IIIc. Clones of Clade A were considered to have similar binding patterns to the human FGFR3 isoform or interspecies FGFR3 due to sequence and

To accurately measure the binding kinetics (KD value) of anti-FGRF3 clones, we determined the affinity of the antibody to each purified protein based on surface plasmon

structural similarities of CDR-H3.

#### *2.6. Binding Properties of Anti-FGFR3 Antibodies*

To verify the target-specific binding ability of the anti-FGFR3 antibodies, purified human FGFR3 isoform (FGFR3-IIIb, -IIIc) and several species of FGFR3 (mouse FGFR3-IIIc, cynomolgus monkey FGFR3-IIIc) were used in an ELISA assay. All clones showed specific binding for the purified human FGFR3-IIIc antigen, which were used for bio-panning, dependent on their concentration, but not to the negative protein (Fc-tagged). The six clones were considered to have apparent specificity for purified human FGFR3-IIIc (Figure 5A). Clade A (A1D06, S2D05, S3A06, and S3B09 clones), clustered based on the similarity of CDR-H3 sequences, was confirmed to have cross-reactivity between human FGFR3-IIIb and cynomolgus FGFR3-IIIc, but did not show specificity for mouse FGFR3-IIIc. The clone S1E12 was not specific to human FGFR3-IIIb, but it had apparent binding specificity for mouse FGFR3-IIIc and cynomolgus FGFR3-IIIc. It was hypothesized that this clone has a binding epitope for the IIIc region of the FGFR3 extracellular domain. The first half of the Ig III domain of FGFR3 is encoded by the invariant exon (IIIa), and the other half is the region where the variant occurs by splicing to IIIb or IIIc [24]. The clone A1A10 was estimated to have a homologous region of human (IIIb and IIIc), mouse, and cynomolgus FGFR3 as an epitope, as it binds to the other FGFR3 proteins as well as human FGFR-IIIc. Clones of Clade A were considered to have similar binding patterns to the human FGFR3 isoform or interspecies FGFR3 due to sequence and structural similarities of CDR-H3.

To accurately measure the binding kinetics (KD value) of anti-FGRF3 clones, we determined the affinity of the antibody to each purified protein based on surface plasmon resonance (SPR) analysis. Each clone showed the same binding pattern as the ELISA assay for isotype and interspecies of FGFR3 proteins (Figure S5). Clade A bound to hFGFR3-IIIc, hFGFR3-IIIb, and cynoFGFR3-IIIc, but did not bind to mFGFR3-IIIc. Clone S1E12 showed specific affinity for FGFR3-IIIc regardless of species and did not bind to FGFR3-IIIb. Clone A1A10 showed specificity for all types of FGFR3 proteins. The affinity of each clone for the FGFR3-purified protein is distributed from about ten nanomolar to sub-nanomolar, and it is considered that could be applied as a therapeutics (Table 3).

We verified that the selected clones bind to the FGFR3 expressed on cell surface using FACS analysis. All clones showed specific binding to FGFR3-overexpressing cell (PDC #1) and did not bind to FGFR3-negative cell (PDC #2) (Figure 5B). Thus, the clones selected through the introduction of cell panning were proved to have specific binding affinity for the purified FGFR3 protein as well as the natural FGFR3 tertiary structure on the cell surface membrane.

resonance (SPR) analysis. Each clone showed the same binding pattern as the ELISA assay for isotype and interspecies of FGFR3 proteins (Figure S5). Clade A bound to hFGFR3-IIIc, hFGFR3-IIIb, and cynoFGFR3-IIIc, but did not bind to mFGFR3-IIIc. Clone S1E12 showed specific affinity for FGFR3-IIIc regardless of species and did not bind to FGFR3-IIIb. Clone A1A10 showed specificity for all types of FGFR3 proteins. The affinity of each clone for the FGFR3-purified protein is distributed from about ten nanomolar to sub-nanomolar,

We verified that the selected clones bind to the FGFR3 expressed on cell surface using FACS analysis. All clones showed specific binding to FGFR3-overexpressing cell (PDC #1) and did not bind to FGFR3-negative cell (PDC #2) (Figure 5B). Thus, the clones selected through the introduction of cell panning were proved to have specific binding affinity for the purified FGFR3 protein as well as the natural FGFR3 tertiary structure on the cell

**Table 3.** Binding affinities (KD) of selected clones to human FGFR3 isotypes and interspecies FGFR3.

FGFR3-IIIc Human 2.78 4.54 0.63 21.2 1.39 6.37 FGFR3-IIIb Human 3.52 5.43 1.66 23.5 n.b. 15.8 FGFR3(IIIc) Cynomolgus monkey 2.1 5.62 1.02 32.8 2.36 7.77 FGFR3 (IIIc) Mouse n.b. n.b. n.b. n.b. 1.93 7.49

**KD, nmol/L** 

**A1D06 S2D05 S3A06 S3B09 S1E12 A1A10** 

**Clade A** 

and it is considered that could be applied as a therapeutics (Table 3).

surface membrane.

n.b.: non-binding.

**Antigen Species** 

**Figure 5.** Binding characterization analysis of anti-FGFR3 antibody derived following the introduction of semi-automated bio-panning. (**A**) The binding specificity of selected FGFR3 antibodies to FGFR3 isoforms and interspecies was evaluated using ELISA. (**B**) The binding ability of FGFR3 antibodies to FGFR3-overexpressing cells (PDC#1) and FGFR3 negative cells (PDC#2) was determined using flow cytometry. **Figure 5.** Binding characterization analysis of anti-FGFR3 antibody derived following the introduction of semi-automated bio-panning. (**A**) The binding specificity of selected FGFR3 antibodies to FGFR3 isoforms and interspecies was evaluated using ELISA. (**B**) The binding ability of FGFR3 antibodies to FGFR3-overexpressing cells (PDC#1) and FGFR3 negative cells (PDC#2) was determined using flow cytometry.

To confirm the mechanism of inhibition of S3B09 on FGFR3-overexpressing cells, the induction of FGFR3 degradation by the antibody was verified. Changes in FGFR3

or S3C04 with the FGFR3-overexpressing cell line for 1 h, changes in the FGFR3 expression level on the surface of living cells were confirmed. S3B09 decreased the FGFR3 expression level by 28% compared to control IgG; however, S3C04 did not affect the FGFR3 expression level (Figure 6B). Additionally, after treatment with S3B09 or S3C04, the total FGFR3 protein level was evaluated using the lysate of the FGFR3-overexpressing cells, and the total FGFR3 protein level decreased by approximately 20% when treated with S3B09 (Figure 6C). These results suggest that S3B09 antibody selected by the introduction of cell panning effectively

inhibits the growth of FGFR3-overexpressing cells through FGFR3 degradation.

*2.7. Biological Function Analysis of Anti-FGFR3 Antibodies to FGFR3-Overexpressing Cells*  To compare biological functions, in vitro functional assays for S3B09 of Clade A, **Table 3.** Binding affinities (KD) of selected clones to human FGFR3 isotypes and interspecies FGFR3.


significantly inhibits the cell growth compared to control IgG and S3C04 against FGFR3 n.b.: non-binding.

overexpressing PDCs.

#### *2.7. Biological Function Analysis of Anti-FGFR3 Antibodies to FGFR3-Overexpressing Cells*

To compare biological functions, in vitro functional assays for S3B09 of Clade A, which is the representative clone selected by cell panning, and S3C04 selected by conventional bio-panning were performed on FGFR3-overexpressing PDCs. The cell growth inhibition of anti-FGFR3 antibodies (S3B09, S3C04) was evaluated for FGFR3-overexpressing cells incubated with FGF1 ligand. To accurately analyze the anti-proliferation effect of the antibody, cell viability was assessed. After 96 h of antibody treatment, it was confirmed that the S3B09 selected by cell panning inhibited the tumor growth by approximately 30% (Figure 6A). The S3C04 selected by conventional bio-panning inhibited the tumor growth by less than 10%. It means that S3B09 antibody significantly inhibits the cell growth compared to control IgG and S3C04 against FGFR3-overexpressing PDCs. *Int. J. Mol. Sci.* **2021**, *22*, 6240 13 of 20

**Figure 6.** Biological functional assay of anti-FGFR3 antibodies. Inhibitory effect of clone S3B09 (Clade A) and S3C04 selected from conventional bio-panning on proliferation of FGFR3 overexpressing cells (PDC #1) was evaluated. Cells were cultured in the presence of 10 ng/mL FGF1 plus 10 μg/mL heparin sulfate or with S3B09 or S3C04 antibodies. Relative cell growth was evaluated through (**A**) cell viability after 96 h incubation with antibodies. Data represent mean ± SD; \*\*\*, *p* < 0.001 using one-way ANOVA. (**B**) Cell surface FGFR3 degradation assay. Individual antibodies were treated on FGFR3-overexpressing cells (PDC #1) and incubated for 1 h at 37 °C, and FGFR3 expression on the cell surface was detected through FACS analysis. (**C**) Total FGFR3 degradation assay. After the FGFR3-overexpressing cells (PDC #1) were treated with the antibody, total FGFR3 contained in the cell lysate was detected through sandwich ELISA. Data represent mean ± SD; \*, *p* < 0.03 using one-way ANOVA. **Figure 6.** Biological functional assay of anti-FGFR3 antibodies. Inhibitory effect of clone S3B09 (Clade A) and S3C04 selected from conventional bio-panning on proliferation of FGFR3 overexpressing cells (PDC #1) was evaluated. Cells were cultured in the presence of 10 ng/mL FGF1 plus 10 µg/mL heparin sulfate or with S3B09 or S3C04 antibodies. Relative cell growth was evaluated through (**A**) cell viability after 96 h incubation with antibodies. Data represent mean ± SD; \*\*\*, *p* < 0.001 using one-way ANOVA. (**B**) Cell surface FGFR3 degradation assay. Individual antibodies were treated on FGFR3-overexpressing cells (PDC #1) and incubated for 1 h at 37 ◦C, and FGFR3 expression on the cell surface was detected through FACS analysis. (**C**) Total FGFR3 degradation assay. After the FGFR3-overexpressing cells (PDC #1) were treated with the antibody, total FGFR3 contained in the cell lysate was detected through sandwich ELISA. Data represent mean ± SD; \*, *p* < 0.03 using one-way ANOVA.

**3. Discussion**  In phage display with WCP, the binders are separated in a state where the membrane protein is expressed in a natural tertiary structure to recognize the epitope of the naturally exposed region [3,10,13]. Therefore, WCP could represent a suitable strategy to obtain antibodies that are conformational-specific for membrane protein such as G proteincoupled receptors, ligand-gate ion channels, receptor tyrosine kinases, and immunoglobulin-like receptors, etc. [3,41–44]. Various cell-based panning strategies have been studied and optimized over a long period of time such as shadow-stick selection technique, FACS sorting technique, and bio-panning and rapid analysis of selective interactive ligands (BRASIL) [3,45–48]. However, technical optimization is required to reduce the non-specific binders to the common cell surface proteins or the irrelevant proteins and to improve the time-consuming process using the intact cells [12,14]. To efficiently separate the phage binder that binds to the cell surface membrane To confirm the mechanism of inhibition of S3B09 on FGFR3-overexpressing cells, the induction of FGFR3 degradation by the antibody was verified. Changes in FGFR3 expression level on the cell surface was analyzed using flow cytometry and the change in total FGFR3 protein level was analyzed using sandwich ELISA. After incubating the S3B09 or S3C04 with the FGFR3-overexpressing cell line for 1 h, changes in the FGFR3 expression level on the surface of living cells were confirmed. S3B09 decreased the FGFR3 expression level by 28% compared to control IgG; however, S3C04 did not affect the FGFR3 expression level (Figure 6B). Additionally, after treatment with S3B09 or S3C04, the total FGFR3 protein level was evaluated using the lysate of the FGFR3-overexpressing cells, and the total FGFR3 protein level decreased by approximately 20% when treated with S3B09 (Figure 6C). These results suggest that S3B09 antibody selected by the introduction of cell panning effectively inhibits the growth of FGFR3-overexpressing cells through FGFR3 degradation.

#### panning along with a semi-automated cell panning process. We optimized a method of **3. Discussion**

PDCs and cancer cell lines.

labeling cells with biotin-X-DHPE (biotinylated phospholipid), the substance with a biotin tag on a cell membrane-like structure, to apply living cells to magnetic beads-based panning. Since this substance has a hydrophilic head and a hydrophobic tail similar to the structure of cell membrane phospholipids, it can be easily immobilized to the cell membrane while minimizing cell damage. Since the biotin tag has a strong affinity for Streptavidin, Streptavidin-coated magnetic beads can be efficiently used with cells during the bio-panning procedure. The PDCs were conjugated with biotin-X-DHPE and attached In phage display with WCP, the binders are separated in a state where the membrane protein is expressed in a natural tertiary structure to recognize the epitope of the naturally exposed region [3,10,13]. Therefore, WCP could represent a suitable strategy to obtain antibodies that are conformational-specific for membrane protein such as G protein-coupled receptors, ligand-gate ion channels, receptor tyrosine kinases, and immunoglobulin-like receptors, etc. [3,41–44]. Various cell-based panning strategies have been studied and optimized over a long period of time such as shadow-stick selection technique, FACS

implemented this method using an automatic instrument as the magnetic particle processor owing to its advantages in terms of time and labor required to efficiently isolate binders [17–19]. Furthermore, it may be applicable for high-throughput screening using multiple libraries, and binders can be selected simultaneously using various cells such as

Abnormal FGFR3 signaling due to overexpression and/or mutation induces tumor proliferation and metastasis in multiple tumors [20–25]. Several antibody therapeutics targeting FGFR3 have been developed such as mAb, Vofatamab (Rainier Therapeutics) currently undergoing phase 1/2(b) clinical trial and Antibody-drug conjugate, LY3076226 (Eli Lilly) currently undergoing phase 1 clinical trials [49–51]. In our laboratory, we tried

protein, we designed a screening process that enriches the binder using conventional bio-

sorting technique, and bio-panning and rapid analysis of selective interactive ligands (BRASIL) [3,45–48]. However, technical optimization is required to reduce the non-specific binders to the common cell surface proteins or the irrelevant proteins and to improve the time-consuming process using the intact cells [12,14].

To efficiently separate the phage binder that binds to the cell surface membrane protein, we designed a screening process that enriches the binder using conventional bio-panning along with a semi-automated cell panning process. We optimized a method of labeling cells with biotin-X-DHPE (biotinylated phospholipid), the substance with a biotin tag on a cell membrane-like structure, to apply living cells to magnetic beads-based panning. Since this substance has a hydrophilic head and a hydrophobic tail similar to the structure of cell membrane phospholipids, it can be easily immobilized to the cell membrane while minimizing cell damage. Since the biotin tag has a strong affinity for Streptavidin, Streptavidin-coated magnetic beads can be efficiently used with cells during the bio-panning procedure. The PDCs were conjugated with biotin-X-DHPE and attached to the Streptavidin-coated magnetic beads under optimized conditions to maintain viable cells with the natural conformation of the membrane protein [16]. Additionally, we implemented this method using an automatic instrument as the magnetic particle processor owing to its advantages in terms of time and labor required to efficiently isolate binders [17–19]. Furthermore, it may be applicable for high-throughput screening using multiple libraries, and binders can be selected simultaneously using various cells such as PDCs and cancer cell lines.

Abnormal FGFR3 signaling due to overexpression and/or mutation induces tumor proliferation and metastasis in multiple tumors [20–25]. Several antibody therapeutics targeting FGFR3 have been developed such as mAb, Vofatamab (Rainier Therapeutics) currently undergoing phase 1/2(b) clinical trial and Antibody-drug conjugate, LY3076226 (Eli Lilly) currently undergoing phase 1 clinical trials [49–51]. In our laboratory, we tried to screen the specific and functional antibody targeting FGFR3; however, it was difficult to isolate the binders with degradation and/or internalization properties (data not shown).

In this study, we enriched the functional antibodies through the additional semiautomated cell panning process using PDCs (glioblastoma, GBM) with a high level of FGFR3 expression. The 19 individual clones were isolated using the 1034 clones obtained from four or five rounds of the conventional bio-panning, and the 6 individual clones using 560 clones obtained from the additional semi-automated cell panning. The IgGs reformatting and production were performed on isolated clones, and target specificity was confirmed through affinity ELISA analysis and SPR analysis for the purified FGFR3 protein. Through the evaluation of the functional analysis, the final 4 candidates were selected. The semi-automated cell panning showed the enrichment of the desired 4 candidates from the 6 clones in comparison to the 4 of 19 clones obtained from conventional biopanning. All of the 4 clones showed binding specificity for FGFR3-overexpressing cells, and non-specific binding or unwanted binding patterns were not observed. We performed cell proliferation and target degradation assays to analyze the biological function against FGFR3-overexpressing cells. The candidate clone obtained through the introduction of semi-automated cell panning was shown to inhibit tumor growth and degrade FGFR3 on the cell surface; however, the remaining clones (15 clones of 19 clones) obtained through conventional bio-panning alone showed poor biological-related function against the FGFR3 overexpressing cells.

In summary, we optimized the semi-automated cell panning method and then introduced an additional process to enrich binders that have specific binding properties with natural conformation of cell surface protein. It was applied to efficiently select FGFR3 specific antibodies that have binding specificity and biologically relevant function. Finally, we selected a clone that showed anti-tumor effects and FGFR3 degradation against FGFR3 overexpressing cells. This method may be an efficient screening tool for isolating clones that recognize membrane protein structure and have biological functions in antibody discovery

through phage display. Furthermore, it has the potential to extend cell panning procedure to HTS systems and fully automated systems.

#### **4. Materials and Methods**

#### *4.1. Immobilization of Cells Using Biotin-X-DHPE and Coated Magnetic Beads*

PDCs (Glioblastoma, GBM) were dispensed into the tubes at a density of 5.0 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/mL. The cells were centrifuged at 1500 rpm for 3 min with 1% FBS (in PBS, pH 7.4, Thermo Fisher Scientific, Waltham, MA, USA) and washed twice to remove the supernatant. Fluorescein DHPE (N- (Fluorescein-5-Thiocarbamoyl)-1,2-Dihexadecanoyl-sn-Glycero-3-Phosphoethanolamine, Triethylammonium Salt, Invitrogen, F362, Carlsbad, CA, USA) or biotin-X-DHPE (1,2-dihexadecanoylsn-glycero-3-phosphoethanolamine, Invitrogen, B1550, Carlsbad, CA, USA), a cell membrane-like substance, was incubated with the cells to attach magnetic beads to the surface of living cells in a rotator at room temperature or 37 ◦C for 30 or 60 min. Cells labeled with biotin-X-DHPE were further incubated with Streptavidin-FITC (Invitrogen, SA1001, Carlsbad, CA, USA), which has biotin-specific binding force, at 4 ◦C for 1 h and washed twice with 1% FBS wash buffer. Flow cytometry (BD FACSAria™ III Cell Sorter) was used to analyze whether biotin-X-DHPE was fused to the cell surface. To attach magnetic beads to cells labeled with biotin-X-DHPE, 2 mg of Dynabeads™ M-280 Streptavidin (Invitrogen, 11206D, Carlsbad, CA, USA) was washed twice in PBS using a magnetic separation rack and incubated with the cells at room temperature for 1 h. After the reaction, the cells were washed through a magnetic separation rack and observed using a microscope to confirm whether the cells were attached to the biotin-X-DHPE and the magnetic beads complex.

#### *4.2. Optimization of Cell Immobilization Conditions*

The direct beads mounting method is as follows: 3 µg of biotin-X-DHPE and 200 µg of Dynabeads™ M-280 Streptavidin were first incubated at room temperature for 30 min and then isolated using a magnetic separation rack to form biotin-X-DHPE-magnetic beads complex. Then, the complex was incubated with 1.0 <sup>×</sup> <sup>10</sup><sup>6</sup> to 5.0 <sup>×</sup> <sup>10</sup><sup>6</sup> living cells in a rotor at room temperature for 1 h and then the washing was performed to isolate the immobilized cells.

The indirect beads mounting method is as follows: 3 µg of biotin-X-DHPE was incubated with living cells using a rotor at room temperature for 30 min and then washed. The pre-washed Dynabeads™ M-280 Streptavidin was then incubated with cells labeled with biotin-X-DHPE to attach the beads to the cell surface.

To increase the efficiency of attachment between cells and magnetic beads, reaction conditions were optimized using PBS (pH 7.4), 2 mM EDTA/0.1% BSA, and 0.1% Pluronic F-68 (Gibco, 24040032, Carlsbad, CA, USA).

#### *4.3. Bio-Panning Using Phage Display*

Bio-panning for FGFR3-specific antibody screening was performed using the synthetic human scFv library [29]. Conventional bio-panning using the recombinant FGFR3-Fc tagged protein was performed using two methods: magnetic beads-based semi-automated bio-panning (solution phase) and immobilized antigen coating bio-panning (solid phase). In the case of the magnetic beads-based semi-automated bio-panning method, the recombinant FGFR3-Fc protein (R&D systems, 766-FR-050, Minneapolis, MN, USA) was biotinylated using Biotinylation kit (Abcam, ab201796, Cambridge, UK) and then a magnetic particle processor (Thermo Fisher Scientific, KingFisher™ Flex, Waltham, MA, USA) was used to perform semi-automated bio-panning. This automatic process was performed according to the protocol pre-designed through BindIt Software 3.3 and repeated five times to amplify the FGFR3-specific scFv binder pool. For the immobilized antigen coating method, the immuno-tube was coated with 3 µg each of recombinant FGFR3-Fc tagged protein and recombinant negative-Fc tagged protein. To remove the scFv binder that binds to the Fc tag through negative selection, the phage library pool was first incubated on an immuno-tube coated with a negative-Fc tagged protein. The supernatant was further

incubated on the immuno-tube coated with the FGFR3-Fc tagged protein to obtain scFv binders specifically binding to FGFR3. The scFv binder pool that specifically binds to FGFR3 was amplified by performing four rounds of bio-panning.

#### *4.4. Semi-Automated Cell Panning Using Immobilized Cells*

Semi-automated cell panning was performed using beads-attached cells to rapidly isolate clones that bind to the surface of FGFR3-overexpressing cells from output of the previous two conventional bio-panning methods. First, the rescued phage pool was incubated with the FGFR3-negative cells and Streptavidin-coated magnetic beads for negative selection and only the supernatant was collected using a centrifuge. The recovered phage pool was placed in a 24-deep-well plate of KingFisher™ Flex, and approximately 1.0 <sup>×</sup> <sup>10</sup><sup>6</sup> to 2.0 <sup>×</sup> <sup>10</sup><sup>6</sup> of FGFR3-overexpressing cells labeled by biotin-X-DHPE and Streptavidincoated magnetic beads were added into a dedicated plate (24-deep well) of an automated magnetic particle processor. Wash buffer (0.1% PBST), followed by elution buffer, was also added to the plates, and semi-automated cell panning was then performed using the predesigned BindIt Software 3.3 protocol. From the phage pool recovered by semi-automated cell panning, clones which specifically bind to the FGFR3 protein but not the negative protein were isolated via affinity ELISA analysis for scFv screening.

#### *4.5. 3D-Structure Modeling and Alignment Analysis of Anti-FGFR3 Antibodies*

Structure-based prediction and design for antibody engineering were performed using a web-server called SAbPred [37]. This tool annotates antibody sequences, which are required for both heavy and light chains for modeling paired antibody, and automatically generates a homology model of the antibody Fv region. Six antibody sequences were selected for this study. The putative Fv model was annotated and refined into the region of CDR-H3 or not by using PyMOL.

Structural alignment was performed using a multi-protein structure alignment server called POSA, which uses multiple flexible structural alignment [38,39]. Fv models (PBD IDs) of six clones were submitted, and RMSD values and visualization modeling were verified based on the alignment results. Finally, annotation and refinement were performed using PyMoL.

#### *4.6. IgG Reformatting and Production of Anti-FGFR3 Antibodies*

The antibody variable region of isolated FGFR3-specific scFv was analyzed using phagemid vectors. The variable region sequences of heavy chain (IgG1) or light chain isolated from phagemid vectors were inserted into each mammalian expression vectors, respectively. Anti-FGFR3 antibodies were produced using the Expi293 transient mammalian expression system (Gibco, A14635, Carlsbad, CA, USA) through co-transfection of the above-mentioned vectors. Following transfection, the culture supernatant was purified using the ÄKTA protein purification system (GE Healthcare Life Sciences, Uppsala, Sweden) with HiTrap Mabselect SuRe (GE Healthcare Life Sciences, 11-0034-93, Uppsala, Sweden). After purification, enrichment was performed with Amicon® Ultra Centrifugal Filter (Merck Millipore, MA, USA). The characteristics of the highly purified antibodies were analyzed using SDS-PAGE and SEC-HPLC.

#### *4.7. ELISA Binding Assay*

One microgram per milliliter of each human (FGFR3-IIIb and -IIIc) (R&D systems, 1264- FR-050, Minneapolis, MN, USA), mouse (R&D systems, 710-MF-050, Minneapolis, MN, USA), and cynomolgus FGFR3 (Sino Biological, 90313-C02H, Beijing, China) protein was coated on 96-well EIA/RIA plates (Costar, #3590, Corning, NY, USA) at 4 ◦C for over-night, respectively. The plates were blocked with 3% skim milk containing anti-FGFR3-antibodies and incubated for 1 h at room temperature. After washing with PBST (0.1%), the antihuman Fab antibody-conjugated horseradish peroxidase (HRP) (Thermo Scientific, 31482, Waltham, MA, USA) was added at a ratio of 1:3000 in 3% skim milk. Following the wash,

the plate was treated with TMB solution (Thermo Scientific, N301, Waltham, MA, USA) as an HRP substrate, and the reaction was stopped with STOP solution (Cell Signaling Technology, #7002, Danvers, MA, USA). The absorbance for each well was detected at 450 nm wavelength with an Infinite® M200 pro (Tecan, Männedorf, Switzerland).

#### *4.8. Surface Plasmon Resonance Analysis*

The binding affinity (KD values) of anti-FGFR3 antibodies was measured using Biacore 3000 (GE Healthcare Life Sciences, Uppsala, Sweden). The human (FGFR3-IIIb and - IIIc), mice, and cynomolgus FGFR3 proteins were immobilized on a Sensor Chip CM5 (GE Healthcare Life Sciences, 29149604, Uppsala, Sweden) with Amine coupling kit (GE Healthcare Life Sciences, BR100050, Uppsala, Sweden). The KD (Ka and Kd) value was assessed according to the concentration gradient of antibodies.

#### *4.9. Cell Binding Analysis Using Flow Cytometry*

Binding efficiency of phage pools to FGFR3-overexpressing cells was determined using flow cytometry (BD Biosciences, FACSAria III, Mountainview, CA, USA). The phage pools (approximately 5.0 <sup>×</sup> <sup>10</sup><sup>12</sup> phage particles) precipitated with PEG were titrated in advance and blocked using 1% BSA at room temperature for 1 h. FGFR3-overexpressing cells at a density of approximately 3.0 <sup>×</sup> <sup>10</sup><sup>5</sup> to 5.0 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/mL were washed in eBioscience™ Flow Cytometry Staining Buffer (Invitrogen, 00-4222-57, Carlsbad, CA, USA) and then incubated with pre-blocked phage pool at 4 ◦C for 1 h. After washing twice with a staining buffer, anti-HA tag antibody (Cell signaling technology, # 3724S, Danvers, MA, USA) was diluted 1:800 in a staining buffer and incubated at 4 ◦C for 1 h. After washing, goat anti-Rabbit IgG (H+L) cross-adsorbed secondary antibody conjugated with Alexa Fluor 488 (Invitrogen, A-11008, Carlsbad, CA, USA) was diluted 1:200 in a staining buffer and incubated for 30 min at 4 ◦C. The binding pattern of phage pools was analyzed using flow cytometry.

The cell-surface-binding efficiency of anti-FGFR3 antibodies was analyzed using flow cytometry. About 3.0 <sup>×</sup> <sup>10</sup><sup>5</sup> to 5.0 <sup>×</sup> <sup>10</sup><sup>5</sup> FGFR3-overexpressing or FGFR3 negative cells were incubated with anti-FGFR3 antibodies with 200 nM at 4 ◦C for 1 h. After washing twice with eBioscience™ Flow Cytometry Staining Buffer, the cells were stained with the goat anti-human IgG (H+L) cross-adsorbed secondary antibody conjugated with Alexa Fluor 488 (Invitrogen, A-11013, Carlsbad, CA, USA) diluted 1:200 in staining buffer at 4 ◦C for 30 min. Mean fluorescence intensity was analyzed by flow cytometry.

#### *4.10. Assessment of Cell Growth*

FGFR3-overexpressing cell were seeded with FGF1 ligand (10 ng/mL) plus heparin sulfate (10 µg/mL) in a 96-well plate at a density of 20,000 cells/well and incubated with 500 nM anti-FGFR3 antibodies for 96 h. The cell growth was assessed using an Ez-Cytox cell viability assay kit (DAEIL Lab, EZ-1000, Seoul, Korea) at an optical density of 450 nm using a microplate reader.

#### *4.11. Target Degradation Assay*

FGFR3 degradation of the selected antibody was confirmed by analyzing the FGFR3 on the cell surface following antibody treatment. FGFR3-overexpressing cells (3.0 <sup>×</sup> <sup>10</sup><sup>5</sup> ) were incubated with 100 nM of anti-FGFR3-antibodies for 2 h at 37 ◦C. After washing with 1% FBS (pH 7.4, PBS), the cells were stained with human FGFR3 PE-conjugated antibody (R&D systems, FAB766P, Minneapolis, MN, USA) and the FGFR3 expression level on the cell surface was analyzed by FACS analysis.

FGFR3 degradation through total FGFR3 comparison of cells was determined using solid phase sandwich ELISA. Approximately 5000 to 10,000 FGFR3-overexpressing cells were cultivated on a 96-well plate, treated with 200 nM of anti-FGFR3 antibodies, and then incubated at 37 ◦C for 1 h. The cells were centrifuged to remove the supernatant and lysed with cOmplete™ Lysis-M buffer (Roche Diagnostics, 04719956001, Mannheim, Germany). The lysate was quantified and analyzed using sandwich ELISA at an optical density of 450 nm using human total FGFR3 DuoSet IC ELISA (R&D systems, DYC766-2, Minneapolis, MN, USA).

#### *4.12. Statistical Data Analysis*

All statistical analysis were performed using Graph Pad Prism software (GraphPad Software, Inc., La Jolla, CA, USA). One-way ANOVA using paired t-test was used to compare two or more data sets and the statistical significance was set at *p*-value < 0.03.

#### *4.13. Ethical Statement*

GBM specimens were obtained from patients undergoing surgery based on consent in accordance with the appropriate Institutional Review Boards. The study was approved by the Institutional Review Board of Samsung Medical Center (IRB No. 2010-04-004) and performed in accordance with the principles of the Declaration of Helsinki. Written informed consents were obtained.

#### **5. Patent**

A patent application has been filed in South Korea (application number: 10-2019- 0125222).

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/ijms22126240/s1.

**Author Contributions:** Conceptualization, B.M., D.-H.N. and Y.Y.; methodology, B.M. and M.Y.; software, B.M.; validation, B.M., M.Y. and M.C.; formal analysis, B.M. and H.K.; investigation, B.M., M.Y. and M.C.; resources, D.-H.N. and Y.Y.; data curation, D.-H.N. and Y.Y.; writing—original draft preparation, B.M., Y.Y.; writing—review and editing, B.M., H.K. and Y.Y.; visualization, B.M.; supervision, D.-H.N. and Y.Y.; project administration, D.-H.N. and Y.Y.; funding acquisition, D.-H.N. and Y.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by a grant from the Korea Health Technology R&D through the Korea Health Industry Development Institute, funded by the Ministry of Health & Welfare, Republic of Korea (HI14C3418). This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. 2017M3A9C8064720).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of the Institutional Review Board of Samsung Medical Center (IRB No. 2010-04-004).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** This research was partly supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2020R1I1A1A01075559).

**Conflicts of Interest:** D.-H.N. and B.M. are the inventors of IP for "Method of Biopanning Using Magnetic Bead attached cell". D.-H.N. is the CTO of AimedBio Inc. and owns shares of AimedBio Inc. which owns IP. B.M., M.Y. and M.C. are under paid employment by AimedBio Inc. and owns shares of AimedBio Inc. which owns IP. The other authors declare that they have no competing interests.

#### **References**


### *Article Candida* **Cell-Surface-Specific Monoclonal Antibodies Protect Mice against** *Candida auris* **Invasive Infection**

**Jonothan Rosario-Colon, Karen Eberle, Abby Adams, Evan Courville and Hong Xin \***

Department of Microbiology, Immunology, and Parasitology, Louisiana State University—Health Sciences Center, New Orleans, LA 70112, USA; jcolo1@lsuhsc.edu (J.R.-C.); keberl@lsuhsc.edu (K.E.); abbya@lsu.edu (A.A.); ecour2@lsuhsc.edu (E.C.)

**\*** Correspondence: hxin@lsuhsc.edu; Tel.: +1-504-568-8121

**Abstract:** *Candida auris* is a multidrug-resistant fungal pathogen that can cause disseminated bloodstream infections with up to 60% mortality in susceptible populations. Of the three major classes of antifungal drugs, most *C. auris* isolates show high resistance to azoles and polyenes, with some clinical isolates showing resistance to all three drug classes. We reported in this study a novel approach to treating *C. auris* disseminated infections through passive transfer of monoclonal antibodies (mAbs) targeting cell surface antigens with high homology in medically important *Candida* species. Using an established A/J mouse model of disseminated infection that mimics human candidiasis, we showed that C3.1, a mAb that targets β-1,2-mannotriose (β-Man<sup>3</sup> ), significantly extended survival and reduced fungal burdens in target organs, compared to control mice. We also demonstrated that two peptide-specific mAbs, 6H1 and 9F2, which target hyphal wall protein 1 (Hwp1) and phosphoglycerate kinase 1 (Pgk1), respectively, also provided significantly enhanced survival and reduction of fungal burdens. Finally, we showed that passive transfer of a 6H1+9F2 cocktail induced significantly enhanced protection, compared to treatment with either mAb individually. Our data demonstrate the utility of β-Man<sup>3</sup> - and peptide-specific mAbs as an effective alternative to antifungals against medically important *Candida* species including multidrug-resistant *C. auris*.

**Keywords:** *C. auris*; candidiasis; multidrug resistance; monoclonal antibodies; universal antibodies; cell wall; passive immunization

#### **1. Introduction**

*Candida auris* is an emerging fungal pathogen first identified in Tokyo, Japan in 2009 [1]. It has since emerged throughout much of the world, with many countries reporting multiple clinical cases [2]. Unlike other pathogenic *Candida* species, *C. auris* has a propensity to colonize abiotic surfaces as well as the human skin [3]. This makes the nosocomial spread of the pathogen especially prevalent and contributes to a higher potential to disseminate into bloodstream infections compared to other *Candida* species. Consequently, ICU patients and nursing home residents are highly vulnerable to nosocomial infections with *C. auris*, and migration from the skin to a disseminated bloodstream infection is especially common in patients with underlying comorbidities, those under immunosuppressed conditions, or those who have undergone invasive surgical interventions [4,5]. This ease of spread has contributed to a slew of healthcare-associated outbreaks, with contamination of ICUs persisting for several weeks [2,6]. Furthermore, with the ongoing COVID-19 pandemic caused by the novel coronavirus, SARS-CoV-2, the rate of hospitalizations is currently extremely high. With many ICU units being filled to capacity, this creates the perfect environment for further *C. auris* ICU outbreaks [7–9]. Once systemic, *C. auris* infection is often fatal, having a case mortality rate of 33–60% [4,10–12], which is much higher than that of other pathogenic *Candida* species. Mortality is most often attributed to multiorgan failure, with the kidney and heart being most susceptible [13–15].

**Citation:** Rosario-Colon, J.; Eberle, K.; Adams, A.; Courville, E.; Xin, H. *Candida* Cell-Surface-Specific Monoclonal Antibodies Protect Mice against *Candida auris* Invasive Infection. *Int. J. Mol. Sci.* **2021**, *22*, 6162. https://doi.org/10.3390/ ijms22116162

Academic Editors: Annamaria Sandomenico and Menotti Ruvo

Received: 11 May 2021 Accepted: 3 June 2021 Published: 7 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

A defining feature of *C. auris*, among other *Candida* species, is its multidrug resistance. Although antifungal resistance has been reported in other *Candida* species, most notably with *Candida glabrata*, the degree of antifungal resistance observed in *C. auris* is unprecedented [2]. In a study that looked at antifungal resistance in 99 clinical isolates of *C. auris* from the United States, 89% of isolates were resistant to fluconazole, 30% were resistant to amphotericin B, and 6% were resistant to echinocandin drugs [16]. In another study of 1385 United States clinical isolates of *C. glabrata*, 9.6% of isolates were resistant to fluconazole and 6% were resistant to echinocandin drugs [17]. Similarly, *C. auris* clinical isolates from across the globe have consistently shown high resistance to antifungals within the azole and polyene drug classes [4,18]. Being so, the typical course of treatment for *C. auris* bloodstream infections is the daily administration of echinocandin drugs, such as micafungin, caspofungin, or anidulafungin. A major limitation of antifungals, however, is their associated drug toxicities. Immunocompromised patients, who are most susceptible to disseminated infection, are in a fragile state and often unable to tolerate additional organ toxicity caused by commonly prescribed antifungal drugs, therefore rending these drugs ineffective [19]. Furthermore, there have been several reports of *C. auris* isolates that are pan-resistant to all three major antifungal drug classes, which greatly limits treatment options [4]. Due to its high degree of antifungal resistance, the potential to spread throughout the hospital environment, and its associated high mortality rate, *C. auris* is the first fungal pathogen to be labeled a serious global public health threat, and new treatments are urgently needed [20].

To overcome the problems of *C. auris* antifungal resistance and drug toxicity, we sought to investigate if prophylactic treatment using *Candida*-specific monoclonal antibodies (mAbs) could induce protection against *C. auris* bloodstream infections in A/J mice, as an alternative to conventional antifungal drug treatment. Protective mAb therapy is an emerging, yet highly promising strategy for the treatment of microbial diseases [21,22]. Antibodies are known to confer protection to various pathogens via several mechanisms, including neutralization, opsonization, and complement activation [23]. As of today, the United States Food and Drug Administration (FDA) has approved five different synthetic mAb for the treatment of various viral and bacterial diseases, including human immunodeficiency virus (HIV) and *Clostridioides difficile* infections [24]. Additionally, two new synthetic mAb-based treatments, Eli Lilly's Bamlanivimab and Regeneron's REGN-COV2 cocktail, are currently in clinical trials and have shown promising efficacy against COVID-19 [25], and the FDA has approved both drugs for emergency use authorization (EUA) [26,27]. Presently, there are no mAb-derived drugs for the treatment of fungal diseases, even within clinical trials. Particularly with pathogens such as *C. auris*, which have developed high levels of drug resistance and cause high mortality in immunocompromised patients, mAb therapy is an attractive treatment option.

Since the cell wall is the first point of contact between *Candida* and the host's immune system, we developed "universal mAbs" that target various *Candida* cell wall epitopes that share high homology among various *Candida* species. A major benefit of universal mAbs is that they could potentially be applied for the treatment of candidemia caused by multiple pathogenic species of *Candida*, such as *Candida albicans*, *Candida glabrata*, *Candida tropicalis*, *Candida krusei*, and *C. auris*. This is especially important because infected individuals often do not receive a timely diagnosis due to unspecific symptoms of invasive candidiasis.

Overall, we hypothesized that prophylactic treatment with universal mAbs would induce extended survival and enhanced fungal clearance within an A/J mouse model of *C. auris* disseminated infection. A/J mice are deficient in complement protein C5 and its cleaved product C5a, a pro-inflammatory chemoattractant important for anti-*Candida* protection [28–30]. This renders A/J mice highly susceptible to *C. auris* disseminated infection without the need for immunosuppressive drugs [31]. Using this model, we identified three mAbs that provided significant protection, as evidenced by extended survival and lower fungal burdens in the kidney, brain, and heart, compared to control mice. In addition, our results showed that two of our mAbs could be administered as a cocktail to further

enhance their effectiveness. Overall, our results demonstrate the efficacy of passive transfer with universal mAbs as a novel treatment against multidrug-resistant *C. auris*.

#### **2. Results**

#### *2.1. In Vitro and In Vivo Efficacy of Antifungals against Multidrug-Resistant C. auris*

*C. auris* isolates can be grouped into five clades (I-V) originating from different geographic regions [4]. Within each clade, isolates may have differences in morphology, levels of virulence, growth rates, and antifungal-resistance profiles [4,32]. Being so, in preparation for our animal studies, we first investigated the antifungal susceptibility of two clinical isolates of *C. auris* belonging to distinct clades: AR-0386 (CAU-06) of Clade IV and AR-0389 (CAU-09) of Clade I. AR-0386 is a highly aggregative South American isolate that has been shown to be less virulent than *C. albicans* in mouse models [33], while AR-0389 is a nonaggregative South Asian isolate that is highly virulent in mouse models [31,34]. It has been reported that nonaggregating *C. auris* isolates such as AR-0389 are among the most virulent clinical isolates, with virulence comparable to that of *C. albicans* in the invertebrate *Galleria mellonella* model [35].

We first performed an in vitro minimum inhibitory concentration (MIC) assay using two commonly administered antifungal drugs, micafungin, and itraconazole. These two antifungals belong to the echinocandin and azole drug classes, respectively, and isolates AR-0386 and AR-0389 have been reported to be susceptible to both drugs in vitro [34,36]. As a comparison, we also tested the MICs for *C. albicans* reference strain SC5314. After 48-h of drug exposure, the micafungin MIC50 was determined to be 0.063 µg/mL for AR-0386, 0.125 µg/mL for AR-0389, and 0.031 µg/mL for *C. albicans* (Table 1). For itraconazole, the 48 h MIC50 was 2.0 µg/mL for AR-0386, 0.25 µg/mL for AR-0389, and 0.031 µg/mL for *C. albicans*. These results showed that AR-0386 and AR-0389 were susceptible to micafungin and itraconazole in vitro, although both isolates were much more resistant to itraconazole than was *C. albicans*.

**Table 1.** Micafungin and itraconazole MIC50 for *C. auris* isolates AR-0386 and AR-0389 and *C. albicans* isolate SC5314 at 24 and 48 h.


A limitation of in vitro assays is that they reflect the limited environment within the test tube, which is considerably different from the environmental conditions encountered in vivo. Being that MIC data cannot always reliably predict in vivo drug susceptibility [37], we next tested the micafungin and itraconazole susceptibility of *C. auris* using a complement C5-deficient A/J mouse model of disseminated infection [33]. To begin, we established an appropriate sublethal challenge dose that would result in 80–100% survival within 10 days post challenge using the highly virulent AR-0389 strain (Figure 1A). Our results showed that doses below 8 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs resulted in 100% survival by day 10. Accordingly, we decided to use a dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs for our in vivo antifungal susceptibility study. After the challenge, mice were treated daily with a minimum protective dose of micafungin (0.25 mg/kg/day) or itraconazole (1.67 mg/kg/day), which were determined via an antifungal pilot study (data not shown). Upon termination on Day 6, micafungin-treated mice showed no reduction in fungal burdens in the kidney, brain, or heart compared to control mice (Figure 1B). Similarly, itraconazole provided no significant reduction in organ burdens using the minimum dose.

**Figure 1.** In vivo efficacy of antifungals against multidrug-resistant *C. auris*: (**A**) the 10-day survival curve of 7-week-old female A/J mice challenged with *C. auris* AR-0389 doses ranging from 1 <sup>×</sup> <sup>10</sup><sup>7</sup> to 1 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs; (**B**) quantification of kidney, brain, and heart fungal burdens from 7-week-old female A/J treated daily for 5 days with a minimum protective dose of micafungin or itraconazole. Mice were challenged with a sub-lethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs of *C. auris* isolate AR-0389. Starting 24 h later, mice received daily administration of 200 µL of DPBS, micafungin (0.25 mg/kg body weight), or itraconazole (1.67 mg/kg body weight). Mice were sacrificed on day 6 post challenge; (**C**) quantification of kidney and brain fungal burdens from 16- to 17-week-old male and female neutropenic C57BL/6 mice treated daily for 14 days with a minimum protective dose of micafungin or itraconazole. To induce neutropenia, mice were administered cyclophosphamide (200 mg/kg body weight) on day 3 and every 7 days after (150-mg/kg body weight). On day 0, mice were challenged with a sublethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs of *C. auris* isolate AR-0389. Starting 24 h later, mice received daily administration of 200 µL of DPBS, micafungin (0.25 mg/kg body weight), or itraconazole (5.0 mg/kg body weight). Mice were sacrificed on day 15 post challenge. MFG = micafungin, ITC = itraconazole. Data are mean + SD (**B**,**C**). *n* = 4 (**B**) *n* = 5 (**A**,**C**). Log-rank (Mantel–Cox) test (**A**) or two-tailed *t*-test (**B**,**C**) were used to identify significant differences. \* *p* < 0.05; *ns* = not significant.

Different inbred mouse strains have differences in MHC haplotypes, immunophenotype features, and isoforms of metabolic enzymes which can contribute to varying rates of drug metabolism [38,39]. Being so, we repeated our in vivo micafungin and itraconazole susceptibility assay using an immunosuppressed C57BL/6 mouse model of disseminated infection to compare *C. auris* susceptibility to that of A/J mice. C57BL/6 mice were immunosuppressed with cyclophosphamide prior to challenge with a sub-lethal dose of *C. auris* AR-0389. Beginning 18 h post challenge, mice were administered micafungin or itraconazole daily for 14 days and then sacrificed on day 15. As with our A/J mouse model, we saw no protective effect using itraconazole, even with a higher dose (5.0 mg/kg/day) (Figure 1C). Interestingly, with the minimum dose of micafungin, we observed a significant reduction in fungal burdens in the kidney and brain (2.9 <sup>×</sup> <sup>10</sup><sup>4</sup> and 2.2 <sup>×</sup> <sup>10</sup><sup>3</sup> CFUs/g, respectively) by day 15 as compared to DPBS mice kidney and brain burdens (5.5 <sup>×</sup> <sup>10</sup><sup>6</sup> and 3.4 <sup>×</sup> <sup>10</sup><sup>4</sup> CFUs/g, respectively). Comparing these results to our MIC data showed that although *C. auris* AR-0389 is susceptible to both micafungin and itraconazole in vitro, only low-dose micafungin was effective at significantly reducing fungal burdens in immunosuppressed C57BL/6 mice. The data showed that *C. auris* is susceptible to minimum doses of micafungin in vivo, but this could vary with mouse strain and cannot be predicted solely using in vitro MIC data.

#### *2.2. Candida Cell Surface Binding of Universal Candida-Specific Monoclonal Antibodies*

Considering that antifungals have limited efficacy against multidrug-resistant *C. auris*, we next investigated the protective efficacy of a panel of universal monoclonal antibodies (mAbs) that target different *Candida* cell surface epitopes, which share high homology among various *Candida* species (Table 2). First, we validated antibody binding to cell surface epitopes by flow cytometric analysis using *C. auris* AR-0386 and AR-0389 and *C. albicans* SC5314. We focused on three mAbs that were shown to be protective in our preliminary studies: C3.1, which targets β-1,2-mannotriose (β-Man3, IgG3), a mannose sugar that is abundantly expressed and distributed on the outer cell wall of most *Candida* species [40,41]; 6H1, which targets hyphal wall protein 1 (Hwp1, IgG2a), a cell wall mannoprotein that is involved in adhesion, biofilm formation, and hyphal development in several *Candida* species [42]; 9F2, which targets phosphoglycerate kinase 1 (Pgk1, IgG1), a metabolic enzyme that is primarily involved in glycolysis and gluconeogenesis within the cytoplasm [43,44].

**Table 2.** Universal *Candida* monoclonal antibodies and their cell surface targets.


<sup>1</sup> The 9F2 and 10E7 antibodies target two different epitopes on Pgk1.

Isolates were incubated with each primary mAb, washed, and then incubated with goat anti-mouse secondary antibody conjugated to Alexa Fluor 488. Antibody binding was then detected using flow cytometry (Figure 2A–C). C3.1 (anti-β-Man3, IgG3) showed 56.6% binding to AR-0386, 98.4% binding to AR-0389, and 86.5% to *C. albicans*, while 6H1 (anti-Hwp1, IgG2b) showed binding of 38.1% to AR-0386, 2.6% to AR-0389, and 21.2% to *C. albicans*. Finally, binding of 9F2 (anti-Pgk 1, IgG2a) was at 33.0% for AR-0386, 3.4% for AR-0389, and 38.6% for *C. albicans*. Additionally, fluorescent microscopy imaging of cells stained with each antibody (Figure 2D–F) depicted levels of fluorescence that corresponded with our flow cytometry data. Since mannose sugars are abundantly expressed on the outer cell wall [40,41], a high level of C3.1 binding in all isolates was expected. Pgk1 and Hwp1, on the other hand, are not major components of the *Candida* cell wall and not as abundantly expressed as β-Man3, which was reflected in our mAb-binding data. It was, however, surprising to see modest Hwp1 binding, since *C. auris* has not been demonstrated to express Hwp1, and its genome has not been shown to contain an ortholog of the *Hwp1* gene. This pointed to the possibility that our anti-Hwp1 mAb could be cross-reactive with another *C. auris* cell wall protein, in which further investigation is required to identify the target as well as confirm its sequence. It was also surprising to see such disparate levels of 6H1 or 9F2 binding between the two *C. auris* isolates. The lower levels of 9F2 and 6H1 binding to AR-0389 could be an indication of epitope masking, which could explain the higher level of virulence of this isolate, compared to AR-0386. Overall, the data showed that the universal mAbs bind to *C. auris* in an isolate-specific manner.

**Figure 2.** Cell surface binding of universal *Candida*-specific monoclonal antibodies: (**A**–**C**) flow cytometry scatter plots depicting levels of cell surface binding of monoclonal antibodies as a measure of Alexa Fluor 488 expression in (**A**) *C. auris* isolate AR-0386, (**B**) *C. auris* isolate AR-0389, and (**C**) *C. albicans* isolate SC5314; (**D**–**F**) confocal microscopy analysis (1000X) showing antibody cell surface staining of (**D**) *C. auris* isolate AR-0386, (**E**) *C. auris* isolate AR-0389, and (**F**) *C. albicans* isolate SC5314 using mAbs C3.1, 6H1, and 9F2. G11.2 = β-1,2-mannotriose (IgG1), C3.1 = β-1,2-mannotriose (IgG3), Hwp1 = hyphal wall protein 1, Pgk1 = phosphoglycerate kinase 1. Bar = 25 µm.

#### *2.3. In Vivo Protective Efficacy of Universal Candida β-1,2-Mannotriose- and Peptide-Specific Monoclonal Antibodies*

We next evaluated if the prophylactic passive transfer of our universal mAbs would protect mice against *C. auris* disseminated infection. We began with C3.1 (anti-β-Man3, IgG3), a mAb that has previously been shown to be highly protective against *C. albicans* disseminated bloodstream infections in mice [45]. Since the *C. auris* cell wall has a high composition of β-1-2-mannans—reportedly even higher than in *C. albicans* [46], we hypothesized that mAb C3.1 would also protect against disseminated infection caused by *C. auris*. To compare the efficacy of C3.1 to that of micafungin, A/J mice were treated one time with C3.1 or DPBS or treated daily for 7 days with micafungin. 24 h after C3.1 treatment, mice were challenged with a sublethal dose of *C. auris* AR-0386. By day 35 post infection, there was a significant increase in survival of C3.1-treated mice (100% survival), compared to DPBS control mice (40% survival) (Figure 3A). Micafungin-treated mice survival (60% survival) was not significantly extended, compared to the DPBS group. When quantifying fungal burdens, C3.1-treated mice had a significant reduction in kidney and brain burdens (1.2 <sup>×</sup> <sup>10</sup><sup>4</sup> and 5.0 <sup>×</sup> <sup>10</sup><sup>1</sup> CFUs/g, respectively), compared to control mice (6.6 <sup>×</sup> <sup>10</sup><sup>8</sup> and 6.2 <sup>×</sup> <sup>10</sup><sup>6</sup> CFUs/g, respectively) (Figure 3B). Remarkably, all C3.1 treated mice had undetectable brain burdens by day 35. While there was also a reduction in heart burdens in C3.1-treated mice (8.2 <sup>×</sup> <sup>10</sup><sup>2</sup> CFUs/g), compared to DPBS control (4.4 <sup>×</sup> <sup>10</sup><sup>6</sup> CFUs/g), this change was not statistically significant. Consistent with survival data, there was no significant reduction in the kidney, brain, or heart burdens in micafungintreated mice (2.8 <sup>×</sup> <sup>10</sup><sup>8</sup> , 4.9 <sup>×</sup> <sup>10</sup><sup>4</sup> , and 4.4 <sup>×</sup> <sup>10</sup><sup>6</sup> CFUs/g, respectively), as compared to DPBS mice. The data showed that mAb C3.1 treatment outperformed low-dose micafungin, a gold-standard drug for the treatment of invasive *C. auris* infection in an A/J mouse model.

**Figure 3.** In vivo protective efficacy of universal *Candida* β-1,2-mannotriose-specific monoclonal antibody: (**A**) the 35-day survival curve and (**B**) quantification of kidney, brain, and heart fungal burdens from 7-week-old female A/J mice treated with mAb C3.1 or micafungin. Mice were treated with 200 µL DPBS, mAb C3.1 (0.24 mg/200 µL DPBS), or micafungin (0.5 mg/kg body weight daily for 7 days) then challenged 18 h later with a sub-lethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs of *C. auris* AR-0386. β-Man<sup>3</sup> = β-1,2-mannotriose. Data are mean + SD (**B**). *n* = 5. Log-rank (Mantel–Cox) test (**A**) or two-tailed *t*-test (**B**) were used to identify significant differences. \* *p* < 0.05; *ns* = not significant. ND = not detectable.

We then evaluated the prophylactic efficacy of our peptide-specific mAbs. In addition to the mAbs 6H1 and 9F2, we also screened one additional universal mAb, 10E7, which targets a different epitope on Pgk1 (GPV-P3, IgG1). A/J mice were treated with each mAb, followed by a lethal dose challenge with *C. auris* AR-0386 18 h later. Survival was observed for 40 days, and fungal burdens were quantified in the kidney, brain, and heart. With this lethal challenge dose, all control mice died on day 5 post challenge (Figure 4A). On the other hand, 9F2 and 10E7 mice had prolonged survival (50% and 25% survival, respectively) by day 40, with 9F2 inducing significantly higher survival, compared to control mice. Although not statistically significant, both mAb-treated groups had slightly reduced fungal burdens in the kidneys, with 9F2 also inducing significantly lower heart burdens (2.5 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs/g), compared to DPBS mice (2.1 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs/g) (Figure 4B). This reduction in heart burdens is host significant, because A/J mice ultimately succumb to cardiac failure, therefore making the heart the best indicator for evaluating disease progress and protection [33,47]. There was no difference in brain burdens among mAb-treated and control groups.

**Figure 4.** In vivo protective efficacy of universal *Candida* peptide-specific monoclonal antibodies: (**A**) the 40-day survival curve and (**B**) quantification of kidney, brain, and heart fungal burdens from 7-week-old male A/J mice treated with two Pgk1-specific mAbs; (**A**,**B**) mice were treated with 200 µL DPBS or antibody (0.285 mg/200 µL DPBS), then challenged 18 h later with a lethal dose of 1 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs of *C. auris* AR-0386; (**C**) quantification of kidney, brain, and heart fungal burdens from 7-week-old female A/J mice treated with mAb 6H1 or micafungin. Mice were treated with 100 µL DPBS or mAb (0.135 mg/100 <sup>µ</sup>L DPBS), then challenged 18 h later with a sublethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs of *C. auris* AR-0389. GPV-P3 = phosphoglycerate kinase 1 (IgG1), Pgk1 = phosphoglycerate kinase 1 (IgG2a). Data are mean + SD (**B**,**C**). *n* = 4. Log-rank (Mantel–Cox) test (**A**) or two-tailed *t*-test (**B**,**C**) were used to identify significant differences. \* *p* < 0.05; *ns* = not significant.

In a separate experiment, we also tested the prophylactic protective efficacy of 6H1 (anti-Hwp1, IgG2b) in A/J mice using a sublethal challenge dose of *C. auris* AR-0389. Mice were sacrificed on day 6, and fungal burdens were quantified as previously (Figure 4C). Based on the data, 6H1-treated mice had lower fungal burdens in the kidney, brain, and heart (3.8 <sup>×</sup> <sup>10</sup><sup>7</sup> , 6.7 <sup>×</sup> <sup>10</sup><sup>5</sup> , and 1.5 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs/g, respectively), compared to DPBS mice kidney, brain, and heart burdens (2.9 <sup>×</sup> <sup>10</sup><sup>8</sup> , 1.8 <sup>×</sup> <sup>10</sup><sup>6</sup> , and 1.8 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs/g, respectively). Of these, only the kidneys had a statistically significant reduction in fungal burdens, while the reduction in heart burdens was nearly significant (*p* = 0.0798). Collectively, the data showed that passive transfer of universal *Candida* peptide-specific mAbs, 9F2 and 6H1 provided significantly extended survival (9F2) and significantly reduced fungal burdens in the kidney (6H1) and heart (9F2), compared to control mice in an A/J mouse model of *C. auris* disseminated infection.

#### *2.4. In Vivo Protective Efficacy of Monoclonal Antibody Cocktails*

Being that several of our antibodies were able to induce protection in mice when administered individually, we further tested if combining two different mAbs would induce enhanced or even synergistic protection in mice. Since our mAbs are specific to different cell surface antigens, we hypothesized that using a cocktail of two mAbs could result in a more effective "double hit" protection against *C. auris*. The selected twomAb cocktail consisted of 6H1, which performed best in the kidney (Figure 4C), and 9F2, which performed best in the heart (Figure 4B). A/J mice were treated with 6H1, 9F2, or both, followed by a sublethal dose challenge of AR-0386 24 h later. Survival was observed for 35 days. Mice that received the 6H1+9F2 cocktail had significantly higher survival (100% survival) by day 35, compared to mice that received only 6H1 (20% survival) or 9F2 (0% survival) (Figure 5A). Regarding fungal burdens (Figure 5B), mice that received the cocktail had significantly lower burdens in the kidney (1.5 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs/g), compared to mice that received only 9F2 (1.0 <sup>×</sup> <sup>10</sup><sup>9</sup> CFUs/g). Kidney burdens in the cocktail group were also lower than in the 6H1 group (6.7 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs/g), although this was not statistically significant. Similarly, fungal burdens in the brain were significantly lower in mice that received the cocktail (5.8 <sup>×</sup> <sup>10</sup><sup>3</sup> CFUs/g), compared to mice that received only 9F2 (9.3 <sup>×</sup> <sup>10</sup><sup>6</sup> CFUs/g). The cocktail group brain burdens were also lower than the 6H1 group (3.4 <sup>×</sup> <sup>10</sup><sup>4</sup> CFUs/g), although this was not statistically significant. We also observed a consistent trend of lower heart burdens for the mAb cocktail treated group (1.1 <sup>×</sup> <sup>10</sup><sup>4</sup> CFUs/g), compared to 6H1 mice (3.3 <sup>×</sup> <sup>10</sup><sup>6</sup> CFUs/g) or 9F2 (8.3 <sup>×</sup> <sup>10</sup><sup>6</sup> CFUs/g), although not statistically significant. Overall, the data showed that protective mAbs, such as 6H1 and 9F2, can be combined into cocktails to provide enhanced protection against *C. auris* disseminated infection compared to treatment with either mAb individually.

**Figure 5.** *Cont*.

**Figure 5.** In vivo protective efficacy of monoclonal antibody cocktails: (**A**) the 35-day survival curve and (**B**) quantification of kidney, brain, and heart fungal burdens from 7-week-old female A/J mice treated with mAbs 6H1, 9F2, or a cocktail of 6H1 + 9F2. Mice were treated with two 400-µL-doses of mAb 6H1 given 18 h apart, one 200-µL-dose of mAb 9F2, or a combination of both mAbs consisting of one dose of 9F2 and two doses of 6H1 given 18 h apart. Then, 18 h after first dose of mAb, mice were challenged with a sublethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs of *C. auris* AR-0386. Hwp1 = hyphal wall protein 1, Pgk1 = phosphoglycerate kinase 1. Data are mean + SD (B). *n* = 5. Log-rank (Mantel–Cox) test (**A**) or two-tailed *t*-test (**B**) were used to identify significant differences. \* *p* < 0.05; \*\* *p* < 0.01; *ns* = not significant.

#### **3. Discussion**

*C. auris* is the first fungal pathogen to cause a serious global public health threat [20]. In vitro and in vivo studies of drug efficacy against this MDR pathogen have shown that most isolates are highly resistant to azoles and polyenes, which severely limits effective drug treatments. As this pathogen easily spreads throughout hospital ICUs and assisted living facilities, *C. auris* seriously threatens the lives of patients living with comorbidities. In those who are most at risk, such as the immunocompromised, antifungal drugs are often not effective and cause additional organ toxicity, which can further exacerbate the conditions of these highly susceptible patients [19].

In this study, we demonstrated for the first time that passive transfer of *Candida* peptide- and glycan-specific universal mAbs is an effective means of immunotherapy for protecting against *C. auris* invasive infections in an established A/J mouse model. During systemic infection, *Candida spp.* disseminates through the bloodstream and enters target organs within hours of infection [48]. During early infection, *Candida* cells are rapidly eliminated from circulation, and the pathogen is often undetectable in the blood [49]. Consequently, our study evaluated antibody efficacy by quantifying fungal burdens in the kidney, brain, and heart, as well as overall survival. A key characteristic of our mAbs is that they target epitopes that share high homology among various clinically significant *Candida* species. This "universal" targeting allows for their potential application in preventing candidemia caused by different *Candida* species and isolates regardless of isolate-specific antifungal-resistance profiles.

The two *C. auris*isolates analyzed, AR-0386 (CAU-06) and AR-0389 (CAU-09), come from different clades and have unique genetic variations resulting in isolate-specific morphologies, rates of proliferation, virulence, and antifungal resistance [32]. These genetic differences may also affect cell wall composition. The lower level of antibody binding observed with isolate AR-0389 could be due to epitope masking by cell wall polysaccharides, which could account for evasion of immune responses and higher levels of virulence observed with isolate AR-0389, compared to AR-0386. This hypothesis is supported by evidence showing that the unmasking of cell wall components, such as β-(1,3)-glucan, can lead to attenuated *C. albicans* virulence in mouse models of systemic infection [50,51]. More work is required to determine why the same mAb has different binding patterns to the cell surface epitopes of *C. auris* isolates, and how this is related to isolate virulence, host–pathogen interaction, and immune escape.

Of our mAb panel, C3.1 (anti-β-Man3) had the highest level of cell surface binding, provided the best overall protection, and significantly outperformed micafungin as a

treatment for invasive *C. auris* infection. β-mannose is a major glycan component of the outer cell wall of many *Candida spp*., and it is abundantly expressed on the surface of *C. auris* [46]. Our previous data with *C. albicans* has shown that C3.1-mediated protection depends on its ability to rapidly and efficiently fix complement to the fungal surface, which is associated with enhanced phagocytosis and killing of the fungus [40,45]. Being that A/J mice are C3-competent, C3.1 likely protects against *C. auris* via the same mechanism. Furthermore, it has been shown that immune complexes of IgG3 can bind to FcγRI receptors on phagocytic cells [52]. Thus, C3.1 opsonization of *C. auris* may lead to additional FcγRI binding, increased adherence and internalization, and enhanced phagocytosis; however, quantitation of these values will be the subject of a later study.

In contrast to C3.1, 6H1 (anti-Hwp1) and 9F2 (anti-Pgk1) mAbs target epitopes that are not abundantly expressed in the *Candida* cell wall, which was reflected in our flow cytometry data. Although Pgk1 is primarily a glycolytic enzyme localized in the cytoplasm, it is also exposed on the *Candida* surface and has been identified as a cell-wall-associated moonlight protein that is immunoreactive during invasive fungal infections in humans [43,44,53]. Hwp1, on the other hand, is expressed on the cell surface during hyphal morphology [54]; however, some evidence does suggest that it may also be expressed during pseudohyphal morphology [55], which has been induced in *C. auris* in vitro [56]. Interestingly, the *C. auris* genome does not appear to contain an ortholog of the Hwp1 gene [54,56], which points to the possibility of cross reaction of the anti-Hwp1 mAb with another cell wall protein. Nonetheless, both mAbs were able to induce significant protection in our mouse model.

Our animal model also showed the enhanced efficacy of two-mAb cocktails as a prophylactic treatment against *C. auris* disseminated infection. A benefit of mAb cocktails is that each antibody can individually induce a different effector mechanism, which can work in concert to inhibit the growth and dissemination of the pathogen [57]. This is a strategy that has been highly effective in treating other infectious diseases, such as HIV infection using highly active antiretroviral therapy (HAART), which uses a cocktail of three or more drugs that inhibit different steps of the virus's replication cycle [58]. In the case of the 6H1+9F2 cocktail, the enhanced protection could be due to improved access of phagocytic cells to *C. auris* yeast, leading to increased oxidative damage. Research conducted by other groups has shown that Pgk1 confers protection to *C. albicans* against reactive oxygen species (ROS) [59] and that immunization with recombinant Pgk1 protein leads to a significant reduction in kidney burdens in mice infected with *C. albicans* or with *C. glabrata* [60]. Additionally, in an in vivo rat venous catheter model of infection, *C. albicans* Hwp1 mutants were shown to display severe biofilm defects with few hyphae [61]. This may indicate that 6H1 and 9F2 mAbs could function together to disrupt biofilm formation and increase susceptibility to respiratory burst by phagocytic cells, although this mechanism would need to be further investigated. Alternatively, the binding of one mAb could induce a conformational change that results in the unmasking of the second mAb's target epitope, leading to functional cooperativity between mAbs targeting different epitopes. Further investigation may also show that these different mechanisms are not mutually exclusive and may both contribute to the observed enhanced efficacy.

In recent years, several novel antifungal compounds have been developed that have proven effective against MDR *Candida* species. One such compound, carvacrol, was shown to have antifungal activity against clinical isolates of *C. auris* while also inducing synergistic antifungal activity when combined with fluconazole, amphotericin B, caspofungin, and micafungin [62]. As with these compounds, mAbs have the potential to be combined with antifungals to induce synergistic protection while also significantly reducing drug MICs and associated toxicity. Future experiments will evaluate the enhanced protection of combining our mAbs with conventional antifungals as well as the therapeutic efficacy of these mAbs.

Finally, it is important to note that similar to most pathogenic *Candida* species, *C. auris* has the propensity to form biofilms within its host. It is well established that within biofilms, *Candida spp.* exhibit higher resistance to antifungal drugs. This is largely due

to the upregulation of efflux pumps [63] and the production of extracellular polymeric substances (EPS), which can interfere with drug diffusion [64]. Although we observed that micafungin was effective at reducing fungal burdens within our immunosuppressed C57BL/6 model, this efficacy would likely be reduced against biofilm-derived *C. auris* cells. In one mouse study of disseminated candidiasis using biofilm-derived *C. glabrata* cells, micafungin treatment was ineffective at reducing liver and kidney burdens [65]. MAbs could be effective here as well. In one study, after incubating *C. albicans* and *Candida dubliniensis* with antibodies targeting the surface antigen, complement receptor 3-related protein (CR3- RP), both species had a reduction in surface adherence and in biofilm thickness in vitro [66]. Using a combination therapy, the binding of mAbs to the surface of the pathogen could interfere with biofilm formation, leading to increased diffusion of antifungals.

In summary, the potential for mAb therapy against microbial pathogens is vast since mAbs inherently have high specificity for their targets without selecting for resistance. The application of protective mAbs against *C. auris* disseminated infection represents a highly promising alternative to the often-ineffective use of antifungal drugs against this MDR pathogen. Not only can effective mAbs protect against severe infection more rapidly than antifungal drugs [67–69], but specific antibodies may also be synergistic with conventional antimicrobials. The data presented here have significant implications for both immunotherapy and vaccine development in the future, and the demonstration of preclinical efficacy of the immunoprotective mAbs in this study will provide compelling data that can be advanced into the clinical setting.

#### **4. Materials and Methods**

#### *4.1. Candida Isolates and Culture Conditions*

Two antifungal-resistant isolates of *C. auris*, AR-0386 (CAU-06) and AR-0389 (CAU-09), were supplied by the United States Centers for Disease Control and Prevention (CDC, Atlanta, GA, USA). *C. albicans* reference strain SC5314 was supplied by the American Type Culture Collection (MYA-2876, ATCC, Manassas, VA, USA). For passive transfer of mAb experiments, the inoculum was serially passaged daily for three days in 25 mL glucose yeast peptide broth at 37 ◦C and then washed three times in Dulbecco's phosphate-buffered saline (DPBS). Cell density was measured using a hemocytometer and adjusted to the desired density in DPBS. For MIC assays, culture was plated onto Sabouraud Dextrose Broth (SDB) agar plates and incubated at 35 ◦C for 24 h. Five colonies were selected and suspended in 1 mL of sterile deionized H2O, and cell density was measured using a hemocytometer. The density was then adjusted to desired concentration in RPMI 1640 + 0.165 M MOPS medium (with L-glutamine and phenol red, without bicarbonate).

#### *4.2. Mice*

Male and female A/J mice were purchased from the Jackson Laboratory (JAX, Bar Harbor, ME, USA) and Envigo (Indianapolis, IN, USA). Male and female C57BL/6 mice were purchased from the Jackson Laboratory (JAX, Bar Harbor, ME, USA). At the time of studies, A/J mice were 7 weeks old, and C57BL/6 mice were 16–17 weeks old. Mice were maintained in the Louisiana State University Health Sciences Center's AAALAC-accredited animal facility (#000037, LSUHSC-NO, New Orleans, LA, USA), and all animal experiments were performed using a protocol approved by the Louisiana State University Health Sciences Center's Institutional Animal Care and Use Committee (#3559, 1/18/2019, LSUHSC-NO IACUC, New Orleans, LA, USA).

#### *4.3. Immunosuppression*

16–17-week-old male and female C57BL/6 mice were immunosuppressed using cyclophosphamide monohydrate (#C0768, Sigma-Aldrich, St. Louis, MO, USA) three days prior to challenge by intraperitoneal (i.p.) injection using a dose of 200 mg/kg of body weight. Immunosuppression was maintained with additional i.p. injections of a 150 mg/kg dose of cyclophosphamide every 7 days.

#### *4.4. Antifungals*

Micafungin (≥97% HPLC) was purchased from Sigma-Aldrich (#SML2268, Sigma-Aldrich, St. Louis, MO, USA), and itraconazole (≥98% TLC) was purchased from Sigma-Aldrich (#I6657, Sigma-Aldrich, St. Louis, MO, USA). For MIC assay, micafungin and itraconazole were dissolved in RPMI 1640 + 0.165 M MOPS medium (with L-glutamine and phenol red, without bicarbonate) + 1% DMSO to the desired concentrations. For animal experiments, micafungin was dissolved in DPBS to the desired concentration, and itraconazole was dissolved in sterile deionized H2O + 10% DMSO to the desired concentration.

#### *4.5. Antifungal Susceptibility*

*C. auris* and *C. albicans* micafungin and itraconazole minimum inhibitory concentrations (MICs) were determined using the broth microdilution method (BMD) according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI) Reference Method for Broth Dilution Antifungal Susceptibility Testing of Yeasts; Approved Standard (CLSI M27-A3, 3rd Ed. 2008. Vol 28, No 14).

#### *4.6. Antibodies*

All monoclonal antibodies were isolated from hybridoma cells, purified, and sterile filtered in phosphate-buffered saline (PBS) by GenScript Biotech Corporation (Gen-Script, Piscataway, NJ, USA) and by Autoimmune Technologies (AiT, New Orleans, LA, USA). Antibody concentrations were determined using a Pierce BCA Protein Assay Kit (#23227, Thermo Scientific, Waltham, MA, USA), according to the manufacturer's directions. A standard curve was constructed using bovine gamma globulin standard (BGG) (#23212, Thermo Scientific, Waltham, MA, USA). Absorbance was read on a plate reader at 562 nm, and sample absorbances were compared to the BGG standard curve to determine antibody concentration.

#### *4.7. Antibody Titers*

Titers of mAbs were determined via enzyme-linked immunosorbent assay (ELISA). A 96-well polystyrene plate was coated with whole synthetic protein (Pgk1, Hwp1) (Gen-Script, Piscataway, NJ, USA) or mannan extract at 4 µg/mL in bicarbonate coating buffer and incubated overnight at 4 ◦C. The following day, the wells were blocked using 1% BSA blocking buffer for 1 h at room temperature. Monoclonal antibodies were then added in duplicate to respective wells using twofold serial dilutions from 1:500 to 1:256,000 and incubated for 2 h at room temperature. Horseradish peroxidase (HRP)-conjugated goat anti-mouse polyvalent IgG, IgA, IgM secondary antibody (#A0412, Sigma-Aldrich, St. Louis, MO, USA) was then added (1:3000) and incubated in the dark for 1 h at room temperature. Tetramethyl benzidine (TMB) (#34022, Thermo Scientific, Waltham, MA, USA) was then added to each well and incubated in the dark for 30 min at room temperature. HCl was added to stop the reaction, and the optical density was measured at 450 nm using a spectrophotometer. As a negative control, wells were incubated with secondary antibody alone. Titers were given as the dilution whose OD reading was greater than two times that of the negative control.

#### *4.8. Antibody Cell Surface Staining*

Overnight cultures of *C. auris* AR-0386, AR-0389, and *C. albicans* SC5314 were washed three times with DPBS. Pellets were resuspended in 100 µL of C3.1 (anti-β-Man3), 9F2 (anti-Pgk1), or 6H1 (anti-Hwp1) antibodies in 1X PBS + 1% BSA and incubated for 1 h at room temperature. Cells were washed three additional times, and pellets were resuspended in Alexa Fluor 488-conjugated goat anti-mouse IgG, IgM secondary antibody (#A10680, Invitrogen, Carlsbad, CA, USA) (1:100) in 1X PBS + 1% BSA and incubated for 1 h at room temperature. Cells were washed three additional times and resuspended in 500 µL DPBS and acquired by flow cytometry at 488 nm (FACSDiva 8.0.3, FACSCanto II, BD Biosciences, San Jose, CA, USA), As a positive control, an additional high-binding

anti-β-Man<sup>3</sup> mAb, G11.2 (IgG1), was used. As a negative control, cells were stained with secondary antibody alone. Gating was set on the secondary antibody (Alexa Fluor 488 only) control. A portion of the stained cells was spread on slides for fluorescent imaging.

#### *4.9. In Vivo Model of Disseminated Infection*

For this, 7-week-old A/J mice or 16–17-week-old immunosuppressed C57BL/6 mice were treated via intraperitoneal (i.p.) injection with 200 µL of monoclonal antibody or DPBS. Then, 18 h later, mice were challenged via intravenous (i.v.) injection in the tail vein with *C. auris* AR-0386 at a sublethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> CFUs in 100 <sup>µ</sup>L DPBS or a lethal dose of 1 <sup>×</sup> <sup>10</sup><sup>8</sup> CFUs in 100 <sup>µ</sup>L DPBS or with *C. auris* AR-0389 at a sublethal dose of 4 <sup>×</sup> <sup>10</sup><sup>7</sup> in 100 µL DPBS, depending on the experiment. For experiments that measured antifungal efficacy, mice received daily i.p. administration of micafungin or itraconazole starting 24 h post challenge. All mice were monitored daily for death or the development of a moribund state, at which point they were sacrificed via CO<sup>2</sup> inhalation. All surviving mice were sacrificed at the conclusion of each study.

#### *4.10. Quantification of Fungal Burdens*

Upon death, the kidney, brain, and heart were extracted from mice, and each organ was homogenized in DPBS. The homogenate was then serial diluted and plated onto GYEP agar plates containing chloramphenicol. The plates were incubated for 48 h at 37 ◦C at which time CFUs were quantified. The limit of detection was 50 CFUs/g for each organ.

#### *4.11. Statistical Analysis*

Plots and statistical comparisons were performed using Prism Software (Version 9, GraphPad Software, San Diego, CA, USA). Survival data was evaluated by Kaplan–Meier analysis, and statistical significance was calculated using a log-rank (Mantel–Cox) test. For fungal burden data, results were expressed as mean ± SD, and statistical significance was calculated using a two-tailed *t*-test to compare mAb-treated groups to the control group. Each study contained five mice per group unless otherwise stated. Significant *p* values were defined as follows: \* *p* < 0.05; \*\* *p* < 0.01.

**Author Contributions:** Conceptualization and methodology, H.X.; investigation, J.R.-C., K.E., A.A., and E.C.; data curation, J.R.-C. and K.E.; writing—original draft preparation, J.R.-C.; writing—review and editing, H.X.; supervision, H.X.; funding acquisition, H.X. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Department of Defense CDMRP, Award PR171482.

**Institutional Review Board Statement:** The study was approved by and conducted according to the guidelines of the Louisiana State University Health Sciences Center's Institutional Animal Care and Use Committee (#3559, 1/18/2019, LSUHSC-NO IACUC, New Orleans, LA, USA). Mice were maintained in the Louisiana State University Health Sciences Center's AAALAC-accredited animal facility (#000037, LSUHSC-NO, New Orleans, LA, USA).

**Data Availability Statement:** Data supporting reported results are available upon request (jcolo1@lsuhsc.edu).

**Acknowledgments:** We thank Constance Porretta for technical assistance with flow cytometry acquisition and analysis, Louisiana State University Health Sciences Center (LSUHSC) for the support of our research, and the National Institutes of Health (NIH) for funding support.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**

