**2. Results**

#### *2.1. Selection of Human Anti-SARS-2 RBD Fabs*

To isolate a specific SARS-CoV-2 antibody, the KFab-I library [42,43] was panned against a recombinant SARS-2 RBD immobilized on magnetic beads (Figure 1a). After five rounds of panning, the phage ELISA was performed on immobilized SARS-2 RBD surfaces, using each panning library to monitor the enrichment (Figure 1b). Ninety-five monoclonal phages were randomly picked from the third and fourth rounds, and the binding on the SARS-2 RBD was evaluated by ELISA (Figure S1). Of the 190 individual clones from the third and fourth rounds, 70 clones showed higher absorbances at 450 nm (A450 nm) than those of the negative control (no immobilized SARS-2 RBD control) in the ELISA read-out. The 70 clones were sequenced: 55 were confirmed to be complete and the remaining clones had mutations, such as frame-shifts and stop codons. By analyzing the CDR sequences of the 55 clones, ten unique Fab clones were identified in total and, of these, D12 (Fab) was dominantly selected (60% of sequenced clones (33 out of the 55)), while the other clones showed selection frequencies of about 11% (G3 (Fab)), 9% (E10 (Fab)), 7% (E4 (Fab)), 4% (F7 (Fab)), and 2% (C2 (Fab), C12 (Fab), G9 (Fab), H1 (Fab), and H3 (Fab)) (Figure 1c). In addition, it was observed that nine out of the ten Fabs had CDR3 lengths greater than 12 amino acid residues (70% and 20% for 12 and 14 amino acid residues, respectively), while 10% of the Fabs had shorter CDR3 lengths, such as eight amino acid residues (Figure 1c). In order to confirm that the ten selected clones bound to the SARS-2 RBD and SARS-2 S1 proteins as well, a binding assay was performed, revealing that all of the antibody clones bound to the RBD and the S1 proteins (Figure 1d). In parallel, a binding analysis for RBD variants was also performed using the ten selected clones (Figure S2). This showed that all the phage clones bound to each RBD variant. Two of them (G9 and E4) bound weakly to one RBD variant (N354D; D364Y).

**Figure 1.** Panning of the phage-displayed synthetic Fab library on an immobilized SARS-2 receptor-binding domain (RBD). (**a**) Monitoring of the phage titers over four rounds (R2–R5) of panning. Black and gray bars indicate the ratio of the phage output to the input titers, presented as a percentage (%), from panning on immobilized SARS-2 RBD (black, SARS-2 RBD (+)) and non-immobilized SARS-2 RBD (gray, SARS-2 RBD (−)) surfaces. The ratio of the output to the input (%) = (phage output titer ÷ phage input titer) × 100. (**b**) Phage ELISA performed on the immobilized SARS-2 RBD surfaces using each panning library phage. (**c**) Frequency of ten Fab phage clones selected in the third and fourth rounds (left) and the distribution of HCDR3 lengths (right). The selection frequency of a unique clone (%) = (number of unique clones ÷ total number of phage ELISA positives) × 100. (**d**) Monoclonal ELISA of ten Fab phage clones against the SARS-2 RBD (red) and SARS-2 S1 protein (green). AA: amino acid residue; NC: negative control.

### *2.2. Production and Characterization of Human Anti-SARS-2 RBD Fabs*

To order to produce and characterize the selected clones as Fab proteins, clones were cloned into an in-house *E. coli* expression vector (pKFAB). The Fab proteins were expressed and purified as described in the Section 4. The protein yields of these Fab clones were 11 mg/L, 6.5 mg/L, 106 mg/L, 15.5 mg/L, 12.5 mg/L, 125.5 mg/L, 8.5 mg/L, 17.5 mg/L, 9.5 mg/L, and 40 mg/L for C2 (Fab), C12 (Fab), D12 (Fab), F7 (Fab), H1 (Fab), E4 (Fab), E10 (Fab), G3 (Fab), G9 (Fab), and H3 (Fab), respectively (Figure S3 and Table 1).


**Table 1.** Physicochemical properties of human anti-SARS-2 RBD antibodies.

Fab: antigen-binding fragment; IgG: immunoglobulin G; n.d.: not determined; *T*m: melting temperature. *T*m1 and *T*m2 are the first and second apparent melting temperatures determined by differential scanning fluorimetry (DSF), respectively; *EC*50: half maximal effective concentration; *K*D: equilibrium dissociation constant; *IC50*: half maximal inhibitory concentration; Mon.: monomer; Agg.: aggregate; *IC*50<sup>P</sup> and *IC*50A: *IC*<sup>50</sup> determined by a pseudo-typed virus (D614 spike) and authentic SARS-CoV-2 virus, respectively.

> The apparent affinities of the ten Fabs for the SARS-2 RBD were assessed using an ELISA (*EC*50, nM) (Figure 2a and Table 1). While six Fabs (G9 (Fab), C2 (Fab), F7 (Fab), H1 (Fab), H3 (Fab), and E4 (Fab)) had low to intermediate apparent affinities for the SARS-2 RBD (*EC*<sup>50</sup> = 112–663 nM), the remaining four Fabs—D12 (Fab), E10 (Fab), G3 (Fab), and C12 (Fab)—showed relatively higher apparent affinities (19 nM, 62 nM, 67 nM, and 83 nM, respectively) (Figure 2a).

> To examine the potential neutralizing ability of the selected Fabs, we conducted a competitive binding assay between the SARS-2 RBD and ACE2 protein or ACE2-overexpressed cells (Figure 2b). It was found that five Fabs (C2 (Fab), C12 (Fab), D12 (Fab), F7 (Fab), and H1 (Fab)) significantly antagonized the interaction between the SARS-2 RBD and biotinylated ACE2 protein (Figure 2c). The same five Fabs seemed to block the interaction between SARS-2 RBD-mFc protein and ACE2-overexpressed cells in a flow cytometry analysis as well (Figure 2d and Figure S4).

> Next, in order to determine whether the ten Fabs could cross-react with the S1 proteins from other coronaviruses, such as SARS-CoV and MERS-CoV, an ELISA was conducted. This showed that three of the Fabs (E4 (Fab), E10 (Fab), and G3 (Fab)) indeed cross-reacted with the SARS-CoV S1, whereas no Fabs bound with the MERS-CoV S1 (Figure S5).

**Figure 2.** Characterization of human anti-SARS-2 RBD Fabs. (**a**) Soluble ELISA of ten serially diluted human anti-SARS-2 RBD Fabs on immobilized SARS-2 RBD surfaces to measure their apparent affinities (*EC*50, nM). (**b**) Schematic drawings of a competitive ELISA of human anti-SARS-2 RBD Fabs between the SARS-2 RBD and ACE2 protein (left) or ACE2 overexpressed cells (right). (**c**) Competitive ELISA of human anti-SARS-2 RBD Fabs antagonizing the interaction between ACE2 and the SARS-CoV-2 RBD. (**d**) Competitive flow cytometry analysis of human anti-SARS-2 RBD Fabs antagonizing the interaction between ACE2 on cells and the SARS-CoV-2 RBD (tagged with mouse Fc (mFc)). Arrows indicate potentially neutralizing clones. mFc-PE: anti-mouse PE (phycoerythrin) conjugate; MFI: mean fluorescence intensity; n.s: not significant (*p* > 0.05); NC: negative control. \* and \*\*: *p* < 0.05 and *p* < 0.01, respectively.

### *2.3. Production and Characterization of Human Anti-SARS-2 RBD IgGs*

To produce and characterize the five anti-SARS-2 RBD antibodies that seemed to have neutralizing activities in IgG forms, the five Fabs were individually reformatted to IgG forms. That is, the individual VH and VL sequences from each of the Fabs were cloned into heavy- (IgG1 Fc) and light-chain (Ck1) expression vectors, respectively. The five IgGs were transiently expressed in HEK293 cells and subsequently purified as described in the Section 4. The resulting IgGs were highly pure and their protein yields were 9.6 mg/L, 12.9 mg/L, 13.5 mg/L, 13.2 mg/L, and 12.5 mg/L for C2 (IgG), C12 (IgG), D12 (IgG), F7 (IgG), and H1 (IgG), respectively (Figure S6 and Table 1).

In order to confirm whether the purified IgGs could bind to the SARS-2 RBD and its variants—and also whether they could cross-react with other coronavirus S1 proteins, as observed with the Fabs—an ELISA binding assay was conducted, revealing that all the IgGs bound to the SARS-2 RBD and SARS-2 S1, as well as the RBD variants, whereas all of them did not bind to the MERS-CoV S1 (Figure 3a). In particular, one clone, H1 (IgG), was found to cross-react with the SARS-CoV S1 and three IgGs (C2 (IgG), D12 (IgG), and F7 (IgG)) seemed to bind with the SARS-CoV S1 but the binding was too weak to confirm their cross-reactivity.

**Figure 3.** Characterization of anti-SARS-2 RBD immunoglobulin Gs (IgGs). (**a**) Binding analysis of five human anti-SARS-2 RBD IgGs—C12 (IgG), H1 (IgG), C2 (IgG), D12 (IgG), and F7 (IgG)—to the SARS-2 RBD and its variants (top) and the SARS-CoV-2 S1 (D614G) and other coronavirus S1 proteins (bottom), respectively. (**b**) Soluble ELISA of five serially diluted human anti-SARS-2 RBD IgGs on immobilized SARS-2 RBD surfaces to measure their apparent affinities (*EC*50, nM). (**c**) ELISA detection for five human anti-SARS-2 RBD IgGs blocking the binding of the ACE2 protein with the SARS-CoV-2 RBD (top) and analysis of the flow cytometry for the blocking effect between the SARS-CoV-2 RBD and an ACE2-overexpressed cell (bottom). (**d**) Size-exclusion chromatography analysis of five human anti-SARS-2 RBD IgGs. The positions of the molecular mass markers, shown as kDa, on the retention time *x*-axis are indicated above the peaks. The data are presented as the mean ± standard error (SEM). MFI: mean fluorescence intensity; NC: negative control; \*, \*\*, and \*\*\*: *p* < 0.05, *p* < 0.01, and *p* < 0.001, respectively.

To determine whether the apparent affinities of the anti-SARS-2 RBD IgGs were altered by reformatting the Fabs into the IgGs, the apparent affinities of the IgGs were examined using ELISA (*EC*50, nM). As shown in Figure 3b, the five clones (C2 (IgG), C12 (IgG), D12 (IgG), F7 (IgG), and H1 (IgG)) increased their apparent affinities approximately 100- to 1800-fold compared to their Fab formats (Figure 3b and Table 1), which might have been due to an avidity effect [42,43]. Next, a size-exclusion chromatography analysis was performed to assess their non-aggregation properties, revealing that the IgGs were monomeric without forming high molecular weight (HMW) aggregates (Figure 3d and Table 1). The five IgGs were further analyzed using a protein thermal shift (PTS) assay to determine their thermal stabilities; the assay showed that all the IgGs had *T*<sup>m</sup> over 70.0 ◦C, confirming that they were thermally stable (Figure S7 and Table 1). To determine whether the thermal stability of the IgGs was due to the intrinsically high stability of the Fabs, the five Fabs were analyzed with the same PTS assay and the results showed that all of the Fabs had *T*<sup>m</sup> values over

76.0 ◦C, indicating that the high thermal stability of the IgGs was derived from the intrinsic properties of the Fabs (Figure S8).

Next, to determine whether the neutralizing activities of the SARS-2 RBD IgGs remained after reformatting the Fabs into IgGs, we performed a competitive binding assay demonstrating the IgGs' antagonizing activities in the interaction between the SARS-2 RBD and ACE2 protein or ACE2-expressed cells (Figure 3c). All the IgGs significantly antagonized the interaction between the RBD and biotinylated ACE2 protein (Figure 3c top) and also inhibited the interaction between the SARS-2 RBD-mFc protein and ACE2 overexpressed cells in a flow cytometry analysis, although C12 (IgG) showed a slightly reduced inhibition compared to the other IgGs (Figure 3c bottom and Figure S9).

#### *2.4. Neutralization Assay against SARS-CoV-2 Pseudovirus and Authentic SARS-CoV-2*

To evaluate the neutralization potency of the five human SARS-CoV-2 RBD IgGs, we carried out a pseudo-typed virus neutralization assay using a lentiviral HIV-1 pseudotyping system [44]. The five IgGs were found to display strong neutralizing activity against the SARS-CoV-2 pseudo-typed virus, among which C2 (IgG) and D12 (IgG) showed the most potent activity. The *IC*<sup>50</sup> values of C2 (IgG) and D12 (IgG) in the pseudo-typed virus neutralization were 0.015 and 0.035 μg/mL, respectively (Figure 4a and Table 1).

**Figure 4.** In vitro neutralization assay of human anti-SARS-2 RBD IgGs. Pseudo-typed virus-based neutralization (**a**) and a neutralization assay using authentic SARS-CoV-2 (**b**). (**c**) Correlation in neutralization potencies between pseudo-typed virus- and authentic virus-based assays. (**d**) Correlation between affinities of anti-SARS-2 RBD IgGs and their neutralization potencies for the authentic virus. The data are showed as the mean ± standard error (SEM).

Based on our previous competitive binding assays, we examined the five IgGs in order to evaluate their neutralizing effects on authentic SARS-CoV-2. The observations of luminescent signals showed that two of the IgGs, C2 (IgG) and D12 (IgG), exhibited high protection upon SARS-CoV-2 exposure for three days. The *IC50* values of C2 (IgG) and D12 (IgG) in the authentic SARS-CoV-2 neutralization were 0.018 and 0.036 mg/mL, respectively (Figure 4b, Figure S10, and Table 1). The rest of the IgGs showed less neutralization compared to the two IgGs: 0.102 mg/mL, 0.151 mg/mL, and 0.232 mg/mL for H1 (IgG), F7 (IgG), and C12 (IgG), respectively (Table 1). In order to know whether there was any correlation present between the neutralization potencies from the pseudo-typed virus and the authentic virus, we compared and plotted the values from the assays and found that a strong correlation was indeed present between the neutralization potencies from the two different neutralization assays (Figure 4c). In addition, we also found that there was a strong correlation between the affinity and the neutralization potency of the anti-SARS-2 RBD IgGs as well (Figure 4d).

The two human anti-SARS-2 RBD IgGs, C2 and D12, were further characterized by BLI (Octet) in order to determine their affinities with the SARS-2 RBD, and it was found that C2 (IgG) and D12 (IgG) had binding affinities of 0.13 nM and 0.57 nM, respectively (Figure S11). This confirmed that some avidity effects were reflected in the apparent affinities from the previous ELISA (Figure 3b and Table 1), which was performed with an immobilized SARS-2 RBD, unlike the BLI (Octet), which was undertaken with the human IgGs immobilized on the sensor.
