IgG production.

Based on the sequences of αF1sc AM2 and αF1sc AM8, full-length antibodies (immunoglobulin G type 1, IgG1s) were produced by ATUM (www.atum.bio). Resulting antibodies were named αF1Ig AM2 and αF1Ig AM8 and their production yield was 0.26 mg/mL culture and 0.63 mg/mL culture, respectively (above the average, i.e. > 0.1 mg/mL culture).

Antibody stability measurements by DLS.

Antibody drugs not only need to bind a target molecule, but they also need to be developable [53] i.e., they need to be expressed in high yield, have good solubility, be stable during long-term storage, and not be toxic. Although antibodies have evolved to be stable at 37°C in serum, tracking antibody thermostability by dynamic light scattering (DLS) allows us to assess antibody-specific shelf life, propensity to aggregation, fragmentation, and potential generation of chemical by-products that could be toxic in vivo. Our DLS measurements were performed in the absence of preservatives (only saline solution) to assess the unaided resilience of our antibodies and shorten the duration of the experiment (stability at 37°C) or avoid precipitation of adjuvants (high temperature stress). DLS has been used previously to demonstrate the structural integrity of monoclonal antibodies in solution [54]. DLS measurement of the hydrodynamic radius of both αF1Ig AM2 and αF1Ig AM8 in PBS before heat stress, revealed single peaks (100% of total species detected) corresponding to 12.12 and 14.91 nm radius, respectively. At 37°C, hydrodynamic radius measured consistently in the 10–17 nm range for both antibodies, as well as the control antibody, over the course of a 150-day incubation period (Fig 3). Possible aggregates (higher hydrodynamic radius) appeared at the 60^th day of incubation as peaks representing only 0.1% to 0.3% of the molecules detected. Since larger particles (aggregates) scatter light better than small particles, [55] these data suggest a negligible level of aggregation. We conclude that our anti-pestis antibodies could be stable beyond 150 days in harsh environments (37°C is an extreme ambient temperature) when stored with preservatives.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Antibody stability by dynamic light scattering analysis.

The hydrodynamic radius of anti-F1 affinity matured antibody 2 and 8 (αF1Ig AM2, orange; αF1Ig AM8, green) was ~14 nm and did not change after 150 days storage at 37°C in PBS. These results indicate lack of protein aggregation in the conditions tested. A mouse anti-lysozyme IgG stored at 4°C for the duration of the experiment was used as a control (blue). In this figure, two per moving average trendline is indicated as dotted lines.

https://doi.org/10.1371/journal.pone.0305034.g003

IgGs were also subjected to DLS temperature scan from 37 to 55°C. For αF1Ig AM2, temperatures and hydrodynamic radius sizes were as follows: 37°C = 17.1 nm, 45°C = 16.9 nm, and 55°C = no signal. Since the signal from the antibody disappeared somewhere between 45°C and 55°C, we decided to take more measurements between 45°C and 55°C to more accurately ascertain at what temperature the aggregation occurred. Unfortunately, due to constrains in antibody availability we were only able to perform a more accurate assay with αF1Ig AM8. For this antibody the temperatures and sizes measured were: 37°C = 15.3 nm, 45°C = 15.7 nm, 47°C = 14.8 nm, 49°C = 14.6 nm, 51°C = 16.6 nm, 53°C = 6.7 nm, and 55°C = no signal. Lack of signal at 55°C suggests antibody precipitation out of solution at Tm = 55 for both antibodies. We concluded that at least αF1Ig AM8 is thermostable up to 51°C. We were able to assess that aggregation and fragmentation at 0.3 uM antibody concentrations is negligible up to 45°C (αF1Ig 2) and 51°C (αF1Ig AM8). These thermostability data together with the high production yields (section above) suggest good manufacturability.

Kinetics study.

A kinetic study of affinity matured IgGs was performed by ELISA and SPR. Affinity constants (K_Ds), together with affinity improvements, are reported in Table 1. Flow cytometry-based kinetics of yeast displayed scFvs are also reported here for ease of comparison (better described in Figs 2 and S1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Affinities of anti-F1 antibodies for F1 antigen.

https://doi.org/10.1371/journal.pone.0305034.t001

Capsular antigen F1 is expressed on Y. pestis surface at >34°C; therefore, for ELISA-based kinetic studies of αF1 AM2 and 8 we used Y. pestis A1122 grown at 37°C and immobilized on plastic to generate a lawn of cell-associated F1 antigen and perform whole cell ELISA (S2A Fig, top). Furthermore, in order to execute the experiments under Biosafety Level 1 conditions, Y. pestis cells were fixed (i.e. killed) knowing that fixation does not affect the F1 interaction with our antibodies [41]. Functional affinity for cell-expressed F1 was measured at various concentrations of antibody and data were fit in the one-site specific binding equation (Eq #1 in Materials and Methods, Fig 4A). αF1Ig AM2 and αF1Ig AM8 appear to have 2- and 3-fold higher affinity for capsule-embedded F1 than the parental clones (Table 1). The difference in affinity increments between yeast displayed scFvs obtained by flow cytometry (3 and 9-fold affinity increment for αF1sc 2 and αF1sc 8 respectively) and soluble IgGs measured by ELISA, is likely due to different antibody formats (more below) and antigen formats. Specifically, soluble recombinant F1V antigen was used in the flow cytometry assays, and native F1V was used in ELISA. The recombinant F1V used for flow cytometry sorting is dimer-enriched antigen purified from E. coli. In this construct the signal peptide on the native F1 gene is removed and a dipeptide linker (EF) is introduced at junction between F1 and the V proteins. The resulting chimera is produced as soluble protein. Although this recombinant protein was designed to mimic the native protein, the required manipulations could have potentially altered the structure and stability of the proteins in subtle ways. During affinity maturation, we used the recombinant F1V fusion protein. It can be hypothesized that the increase in affinity might be due to additional contact points and/or structural adjustments occurring in the yeast displayed scFv to the presence of the V antigen in the chimera, contributed to the more pronounced changes in affinity. In vitro antibody selection and maturation technologies are highly sensitive to antigen formats, hence differences between recombinant fusion protein and native cell surface protein binding should be anticipated. Nevertheless, we can state that αF1Ig AM2 and αF1Ig AM8 both recognize native and recombinant F1 and that, based on ELISA, the affinity gain achieved by the scFv display-based maturation method seems to be preserved when the antibody format transitions from single chain to full length IgG.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Kinetic study of purified anti-F1 IgGs by ELISA and SPR.

(A) ELISA data obtained for original anti-F1 IgGs (αF1Ig 2 and 8) or affinity matured (αF1Ig AM2 and 8) antibodies are shown. Experiments were performed in triplicate and data points averaged. The standard deviation of each data point was calculated and shown as the error bar. Data were fitted to one-site specific binding equation. AB = antibody-F1 binding; AB_max = antibody-F1 binding at saturation; K_D = dissociation constant = half saturating antibody concentration ([Ab]). (B) Representative surface plasmon resonance sensograms for each parental and affinity matured antibody are shown.

https://doi.org/10.1371/journal.pone.0305034.g004

Affinity characterization was also performed by surface plasmon resonance (SPR, Figs 4B and S2C). For this analysis, αF1Ig AM2, αF1Ig AM8 and the corresponding parental IgGs were immobilized on sensor chips at various concentrations and exposed to various concentrations of monomer-enriched recombinant F1 antigen (S2A Fig, bottom). αF1Ig AM8 was found to have 4-fold higher affinity than parental αF1Ig 8, however no affinity increment was observed for αF1Ig AM2 vs αF1Ig 2. A major cause of this discrepancy could again be the difference in the antigen format used in the two assays. During flow-based kinetic studies (and the affinity maturation process) we used dimer-enriched F1V complex, whereas for SPR kinetics we used monomer-enriched F1. It is possible that αF1 antibody 2 binds at the intersection between F1 and V antigens and that the antibody changes during affinity maturation were tailored to the presence of V antigen. The lack of V antigen in the antigen used for the SPR assay (monomeric F1, not F1V) might explain why this assay failed to reveal affinity maturation. Also notice, that although SPR is the gold standard in affinity measurements, the ELISA used here reflects a more real-world scenario (i.e. antibody interacting with Y. pestis-displayed F1 antigen) than the SPR assay. Therefore, ELISA data should weigh more in the decision of whether αF1Ig AM2 has indeed been matured.

Based on these observations we could conclude that both matured IgGs have higher affinity for F1 antigen than parental IgGs. We do recognize that the increment in affinity was not as high as expected, however please note that increased manufacturability is an additional goal of the maturation process. This is the reason why during maturation we sort yeast that are not only strong antigen binders but also strong displayers. High level of display on the yeast surface is generally a proxy for high stability (e.g. thermostability), foldability and low aggregation (all associated to manufacturability). Therefore, matured antibodies are likely to have higher stability than the parental antibodies. Although in this work the stability of matured antibodies was not measured side by side with the parental counterpart, we previously reported that αF1Ig 2 and 8 start to degrade after 2.5 weeks storage at 37°C, whereas matured αF1Ig AM2 and AM8 (this paper) show no sign of aggregation after 150 days.

In summary, we can safely conclude that our yeast display-based affinity maturation process was successful and resulted in improved variants of parental antibodies αF1Ig 2 and αF1Ig 8, i.e. αF1Ig AM2 and aF1Ig AM8.

αF1Ig AM2 and αF1Ig AM8 anti-F1 antibodies are opsonic in vitro.

Opsonization is antibody-based phagocytosis of bacteria by phagocytes. Prior to examining the mAbs for pre-exposure prophylaxis, we evaluated the effect of the αF1Ig AM2 and αF1Ig AM8 on the interaction of Y. pestis with RAW264.7 cells (murine macrophage-like cells). In these assays, both antibodies were significantly opsonic (Fig 5). The bacteria pretreated with either mAb were phagocytosed to a significantly greater extent than the negative control mAb (α-influenza αM2 IgG Z3). The positive control mouse-derived αF1 mAb F1-04-A-G1 appeared to be more potently opsonic than αF1Ig AM2 and αF1Ig AM8, however this might be an artifact due to a better response of the mouse-derived RAW264.7 cells to a murine antibody (Fig 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The anti-F1 monoclonal antibodies are opsonic in nature.

As demonstrated in an in vitro gentamicin protection assay the mouse derived α-F1 mAb (F1-04-A-G1) and the human derived αF1sc AM2 and αF1sc AM8 mAbs were opsonic. RAW264.7 cells were infected with Y. pestis CO92 pgm- pPst- at an MOI of approximately 10 CFU. (*) indicates P <0.0001 when compared to Y. pestis (no antibody) alone and P < 0.0001 when compared to anti-influenza M2 (αM2 IgG Z3) negative control, (**) indicates P = 0.0007 when compared to Y. pestis alone and P < 0.0001 when compared to anti-influenza M2 negative control. This is the average of 2 experiments and representative of 5 experiments total.

https://doi.org/10.1371/journal.pone.0305034.g005

αF1Ig AM2 and αF1Ig AM8 anti-F1 antibodies protect mice from pneumonic plague.

Initially we tested the antibodies at 500 μg doses. The mAbs were administered intraperitoneally approximately 18 h prior to exposure to aerosolized Y. pestis CO92. In this first experiment, all control mice (PBS-treated, normal mouse IgG-treated, or α-influenza M2-treated) had succumbed to disease or were euthanized by day 4 post-exposure to aerosolized bacteria. However, 100% of the mice receiving mouse-derived αF1 mAb F1-04-A-G1, αF1Ig AM2 or αF1Ig AM8 survived the infection with no clinical signs of disease (Table 2, experiment #1). We also performed several dose-down studies and demonstrated that partial protection was achieved with either a 400 μg dose or 187.5 μg dose of αF1Ig AM2 (Table 2, experiments #2 and #3). However, doses of 125 μg or lower offered no overall protection from pneumonic plague (Table 2, experiment #4), mirroring the results obtained when mice were pre-treated with the mouse-derived αF1 mAb F1-04-A-G1. These dose-down studies did not include αF1Ig AM8 since data from a previous set of preliminary experiments (unpublished data, available upon request) showed that this antibody performs slightly worse than αF1Ig AM2. While the effective doses determined in these experiments are clearly higher than desirable for human use, the BALB/c mouse model of pneumonic plague is quite stringent, and these protective data serve as proof-of-concept for further optimization and/or antibody engineering that may be required for optimal drug product. Additionally, these anti-F1 mAbs were evaluated as single medical countermeasures, but there has been documented synergy when anti-F1 and anti-LcrV mAbs are used in combination. Hill et al. [36] demonstrated that greater protection from bubonic plague was associated with an anti-F1 mAb when combined with antibodies directed against the LcrV protein, a multi-functional protein that is a component of the type three secretion injectisome and essential for virulence [56,57]. It is also conceivable that simultaneous administration of αF1Ig AM2 and αF1Ig AM8 might result in a synergistic effect due to orthogonal binding of these two antibodies to the F1 antigen [41], resulting in lowering of the protective dosage of the antibody cocktail. Finally, it is important to note that due to inherent variability associated with the aerosolization of bacteria, the calculated inhaled dose was approximately 3-fold higher in the experiment evaluating the lowest dose ranges of antibodies (experiment #3 & #4). These lower doses of administered antibodies did extend the time to death or euthanasia compared to PBS (p < 0.0037), but the biological relevance of this delay remains equivocal. Thus, under the right conditions, lower concentrations of anti-F1 mAbs may be even more effective than we observed. Our efforts to develop novel anti-F1 mAbs will be beneficial to both the biodefense and public health research communities. While the overwhelming number of human plague cases have been caused by encapsulated strains of Y. pestis, nonencapsulated strains have been isolated in nature [58,59]. However, very few documented cases of human disease initiated by these strains exist [60]. Nevertheless, nonencapsulated strains remain virulent in animal models of plague and accordingly, other antigenic targets (e.g. LcrV) should always be considered when developing a medical countermeasure against plague [61–66]. We envision an effective anti-F1 mAb to be beneficial in most cases of plague, but a cocktail approach would certainly provide the most complete protection against a number of different Y. pestis strains with differing antigen profiles (e.g. nonencapsulated).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. In vivo experiments using the BALB/c mouse model of pneumonic plague to demonstrate protection of antibody therapy when given as pre-exposure prophylaxis.

https://doi.org/10.1371/journal.pone.0305034.t002

Generation of mutant libraries.

Mutant libraries of αF1sc 2 and 8 were created and sorted for selection of higher affinity/expression antibodies variants following a previously published protocol [49]. Briefly, the antibody genes were amplified from previously described yeast display plasmids pDNL6-αF1sc 2 and pDNL6-αF1sc 8 [41] by error-prone polymerase chain reaction (epPCR) using PNL6 forward and reverse primers (GTACGAGCTAAAAGTACAGTG and TAGATACCCATACGACGTTC). The epPCR reaction mixture included a 5-fold excess dATP/dGTP versus dTTP/dCTP (2 mM and 0.4 mM, respectively), DMSO (2%), MnCl₂ (0.2 mM), MgSO₄ (0.2 mM) and Taq polymerase (NEB, cat# M0273L). The thermocycling PCR program was the following: initial denaturation (95°C, 2 min), 30 cycles of denaturation (95°C, 30 sec), annealing (58°C, 30 sec), polymerization (68°C, 60 sec), and final extension (68°C, 7 min). Four hundred to 500 μL epPCR reaction generated 10–20 μg of insert DNA for both antibodies. Plasmid pDNL6-CBH1 [49,67] was digested with restriction enzymes BssHII, NheI, and NcoI. The quality of the plasmid and digestion was verified by gel electrophoresis. Of note, separation of plasmid and CBH1 insert was not necessary since formation of a library of pDNL6-antibody mutant plasmids is not based on ligation but on recombination upon yeast transformation. Both the epPCR inserts and the cut vector were purified (Qiagen, #28104) prior to yeast transformation. Preparation of 10⁷ mutant library used 20 μg of vector DNA and 10 μg of insert DNA. The yeast transformation protocol used 100 mL of S. cerevisiae EBY₁₀₀ cells at > OD₆₀₀ 2.0 (4 x 10⁷ cells/mL). The lithium acetate-based chemical transformation protocol was performed as described before [68].

Yeast library selection.

Transformed yeast were grown in selective medium (SD/CAA) for two days at 30°C and yeast induction was performed in SGR/CAA medium at 20°C for 16–20 h. The SD/CAA medium was composed of 5 g/L casamino acids (-ade, -urs, -trp), 20 g/L dextrose, 1.7 g/L yeast nitrogen base, 5.3 g/L ammonium sulfate, 10.19 g/L Na₂HPO₄.7H₂O, and 8.58 g/L NaH₂PO₄. The SGR/CAA medium was the same as selective medium except for dextrose, which was substituted with 20 g/L galactose, 20 g/L raffinose, and 1 g/L dextrose. Five hundred microliters of SD/CAA overnight culture was added to 5 mL of SGR/CAA for induction of antibody display. After overnight induction, 200–1000 μL of the yeast was washed with yeast wash buffer (YWB- 1X PBS, 0.5% BSA) by centrifugation. The yeast cells were then incubated with F1V dimer (BEI Resources, NR-2563) which was biotinylated using NHS-LC-LC biotinylation kit (ThermoFisher #21343). Incubation with biotinylated antigen was allowed to proceed for 30–60 min with rotation on a bench top rotator. Yeast cells were washed twice and stained with phycoerythrin (PE) conjugated anti-SV5 IgG (ThermoFisher #37–7500, 1 μg/mL, to assess scFv display levels) and streptavidin-Alexa 635 (ThermoFisher #S21375, 5μg/mL, to detect binding to biotinylated antigen). Flow cytometry-based analysis of yeast staining and subsequent sorting of best F1-binding/scFv-expressing cells were performed using BD FACS Aria. The initial mutant libraries were incubated with 500 nM bio-F1V, analyzed, and sorted. After three rounds of sorting, yeast populations binding F1V at 50 nM were sorted. Single clones from the third sort were subjected to monoclonal analysis to identify best clones. Plasmid DNA from these clones was prepared and another set of epPCR was performed to introduce further mutations. Two sets of mutant libraries and six rounds of sorting was performed to identify the αF1sc 8 variants that binds the F1V antigen with 1 nM affinity. Similarly, three sets of epPCR and ten rounds of sorting were performed to identify the best αF1sc 2 variants that also recognized F1V antigen at 1 nM. Single clones from the two best variant populations were analyzed using 5 nM F1V antigen. Plasmid DNA from each of the best monoclonals was prepared and sequenced by Sanger sequencing.

Kinetic characterization of yeast displayed affinity matured antibodies.

Affinity measurements were conducted using yeast displayed affinity-matured or parent antibodies incubated with serial dilutions of biotinylated-FIV (same antigen format used in affinity maturation process). Yeast cells were analyzed for expression and antigen binding by flow cytometry. Antigen-antibody binding was reported by the Alexa 633 fluorescence (medium fluorescence intensity, MFI; 635 excitation, 660/20 emission). This yeast-associated fluorescence derives from streptavidin-Alexa 633 conjugate, interacting with the yeast-bound biotinylated antigen. Antigen-antibody binding (Alexa 633 fluorescence MFI) at each concentration of antigen was normalized to the level of antibody display on yeast reported by Phycoerythrin fluorescence (MFI, 561 excitation, 582/15 emission). The resulting data (antibody-antigen binding and corresponding antigen concentration) were fit to the one-site specific binding equation using GraphPad software.

Eq 1

In this equation AB is antigen-antibody binding (Alexa 635 MFI, variable), AB_max is the antigen-antibody binding (Alexa 635 MFI) at antibody-binding saturation (constant), [Ag] is the initial antigen concentration (variable), and K_D is the apparent antibody-antigen dissociation constant.

IgG preparation.

Wild-type and affinity-matured scFvs were converted to IgGs by inserting the amino acid sequences corresponding to the variable heavy (VH) and variable light (VL) antibody regions into a standard IgG1 scaffold. The resulting protein sequences were submitted to ATUM (Newark, CA, USA) for codon-optimized back-translation, gene synthesis, and expression as full-length IgG1 antibodies in HEK293 cells. IgG were received frozen from ATUM and stored in small aliquots at -80°C before use in various assays.

Stability study.

αF1Ig AM2 and αF1Ig AM8 were separately incubated in PBS pH 7.4, in a total volume of 1 mL, at concentrations of 0.87 mg/mL and 0.97 mg/mL, respectively. Antibody solutions were stored in Eppendorf tubes inside a humid chamber at 37°C and periodically analyzed for stability by dynamic light scattering (DLS). DLS measurements or the hydrodynamic radius of each antibody and a control antibody (mouse anti-lysozyme IgG stored at 4°C in PBS pH 7.4) were taken immediately after obtainment of antibody solutions (day 0) and then over the course of 150 days’ storage at 37°C. Measurements were performed using a Malvern Zetasizer Nano ZS with the following settings: material = protein, refractive index = 1.45, dispersant = water, measurement angle = 173 degrees backscatter, equilibration time = 30 sec, cell = ZEN0040 (plastic disposable cuvettes), measurement duration = 10 sec, number of runs = 12, number of runs per measurement = 6, and run temperature = 25°C. Size is reported as hydrodynamic radius (nm) by majority (99.6–99.9%) peak volume. Six individual measurements were performed for each data point (i.e., number of days in PBS at 37°C). In addition, temperature ramping of both antibodies was performed using heat-DLS. Antibodies αF1Ig AM2 (0.74 mg/mL) and αF1Ig AM8 (1.49 mg/mL) were heated from 37°C to 55°C with 5-min equilibration times between temperature readings, with the other DLS conditions the same as described above.

Y. pestis culture used for initial in vitro antibody characterization.

Y. pestis A1122 glycerol stocks, obtained from archived culture stocks in the Los Alamos National Laboratory, were streaked on tryptic soy broth agar with 5% sheep’s blood (Teknova #T0148) and allowed to form distinct colonies by incubation at 26°C overnight. Liquid cultures were prepared by inoculating single colonies in brain heart infusion (BHI) broth (BD Biosciences #221812) and overnight growth. F1-positive Y. pestis was cultured at 37°C in BHI medium supplemented with 2.5 mM CaCl₂, whereas F1-negative Y. pestis cultures were grown at 23°C without CaCl₂. Live cells were washed by multiple rounds of a) 10 min centrifugation at 8,500 g, b) removal of supernatant, and c) resuspension of cell pellet in PBS. Y. pestis cell fixation was achieved using cold 4% paraformaldehyde (Sigma-Aldrich #1004960700), followed by extensive PBS wash (at least 3 cycles). Absence of live cells (before using fixed cells in BSL1 conditions) was assessed by plating 100 μL fixed cell suspension on a TSB agar plate and checking for colony formation after overnight incubation at 26°C. Fixed cells were stored at 4°C for several months in the presence of 0.1% sodium azide.

Kinetic characterization by whole-cell ELISA.

This assay used F1 expressed on the surface of Y. pestis grown at 37°C and was performed as described before [41]. Briefly 1 x 10⁴ fixed bacterial cells in 100 μL PBS per well were attached to a 96-well MaxiSorp ELISA plate (ThermoFisher #44240421) by overnight incubation at 4°C, followed by PBS washes and blocking with Wonder Block (WB; 0.3% BSA, 0.3% skimmed milk, and 0.3% fish gelatin in PBS). The primary antibody stocks (anti-F1 IgGs, or anti-influenza M2 IgG Z3 negative control [68] were serially diluted in 1:10 WB:PBS (light WB, LWB), and 100 μL dilution was added to each cell-coated and blocked well. After 1 h incubation, the antibody solution was removed, and the plate washed. HRP-conjugated anti-human antibody in LWB was added, and the plate incubated at 25°C for 1 hour. After washing, 100 μL/well TMB substrate (Sigma- Aldrich #CL07) was added to each well, the reaction was stopped by adding 100 μL/well 0.18 M H₂SO₄, and the absorbance at λ 450 nm of each well was measured. Experiments were performed in triplicate, background absorbance (obtained with negative control antibody) was subtracted from each data point, and the averages of the resulting numbers (data points) plus standard deviations (error bars) were plotted against the corresponding primary-antibody concentrations using KaleidaGraph. Data were fitted to the one-site specific binding equation defined above (Eq #1). Comparative analysis between the original and affinity matured clones were conducted to assess improvement in affinity for the antigen.

Kinetic characterization by SPR.

SPR analysis was performed by Carterra Inc. (Salt Lake City, UT) using their high-throughput instrument (LSA). Chip CMD200M (Carterra, #4287) was coated with protein A/G by priming the instrument with HBSTE running buffer (Hepes Buffer Saline with Tween and EDTA, Carterra, #3630), activating the chip surface for 8 min with 33 mM S-NHS and 133 mM EDC in 100 mM MES buffer pH 5.5, and injecting protein A/G (Pierce # 21186) for 10 min at 100 μg/mL in 10 mM NaAcOH pH 4.0. The chip surface was quenched with 1 M ethanolamine and conditioned by running 2 x 30 sec cycles of 10 nM glycine pH 2. Antibodies αF1Ig AM2 and 8 were added at eight different concentrations starting at 15 μg/mL plus seven 3-fold serial dilutions. Antibodies solutions were captured on the protein A/G chip in an array for 40 min. The F1 antigen (expressed in E. coli, BEI NR-44223 monomer enriched from BEI) was used at five different concentrations starting at 1 μM plus four 5-fold serial dilutions. After antibody capture a series of blank buffer injections were performed, followed by the various solutions of antigen. RBD from SARS-CoV-2 was used as a negative control (Acrobiosystems, #SPD-C52H3).

Bacterial strains and growth conditions used for in vitro macrophage-based assays and in vivo protection studies.

The Y. pestis pgm- pPst-, used for in vitro studies, was generated at USAMRIID and was kindly provided by Dr. Susan Welkos (USAMRIID, Fort Detrick, Frederick, MD) [35]. Y. pestis pgm- pPst-is an attenuated strain derived from the fully virulent Y. pestis CO92 strain, which is cured of the pPst plasmid containing the plasminogen activator (Pla) virulence locus (pla) and is pigmentation-deficient (pgm) [69–71]. This strain allows to perform moderate to high-throughput assays under BSL-2 condition for initial down selection. Y. pestis pgm- pPst- was grown on Remel® Sheep Blood Agar plates (Remel, Thermo Fisher #01202) and incubated at 37°C for 24 h. Bacterial colonies were harvested and used to inoculate 10 mL of brain heart infusion (BHI) broth (BD Biosciences #211059) and incubated in the BHI medium for 2 hours at 37°C with shaking at 200 rpm prior to infecting the macrophages. To ensure the bacteria were harvested in the log phase of growth the OD₆₀₀ of the culture post-incubation was not allowed to exceed 1.0 prior to incubation with antibodies.

Mouse monoclonal antibodies.

The anti-F1 mouse monoclonal antibody F1-04-A-G1 was provided by provided by James Burans and Jennifer Adrich (Naval Medical Research Center, Silver Spring, MD) [35].

Cell culture.

RAW264.7 murine macrophage-like cells derived from an Abelson murine leukemia virus tumor (ATCC TIB-71) were grown at 37°C in 5% CO₂ in low glucose DMEM containing 10% fetal bovine serum, 1% L-glutamine, 1% non-essential amino acids, and 1% HEPES buffer. Cells were used before passage 15 and seeded in 96-well plates.

Quantification of viable intracellular Y. pestis (Gentamicin Protection Assay).

Bacterial cultures were suspended in DMEM from cultures grown in brain heart infusion broth, and multiplicity of infection (MOI) was estimated for an OD₆₀₀ of 1.0 (~ 5.34 x 10⁸ colony forming units (CFU) per milliliter). As previously described for macrophage infection assays, RAW264.7 cells (1.5 x 10⁴ cell/well) were seeded into 96-well plates (Greiner Bio-OneT CellStarT μClearT 96-Well, Cell Culture-Treated, Flat-Bottom Microplate, # 655090) one day prior to infection [72]. Y. pestis, at 8 x 10⁶ CFU/mL, was pre-incubated with 10 μg/mL antibodies in DMEM for 1 hous at 37°C prior to infection. Macrophages were then infected at an approximate MOI of 10 in triplicate wells and the plates were centrifuged at 200 x g for 5 min to initiate infection, and then placed into a 37°C incubator with 5% CO₂. After 1 h of infection, gentamicin (8 μg/mL) was added to the wells to kill extracellular bacteria and the plates were incubated for an additional hour at 37°C in the presence of 5% CO₂. After incubation, macrophages were washed two times in PBS and lysed using 0.1% Triton X-100 in PBS. Serial dilutions of lysates were plated in duplicate on sheep’s blood agar and incubated for 2 days at 28°C for CFU enumeration. The geometric means of CFU per mL were plotted for control samples not incubated with antibody compared to those incubated with 10 μg/mL of each antibody.

Treatment of mice with antibodies and exposure of mice to aerosolized Y. pestis.

Female BALB/c mice what were 7–9 weeks old (Charles River, Frederick, MD) received antibodies via intraperitoneal injections approximately 18 hous prior to exposure to aerosolized Y. pestis. Mice treated with commercially available normal mouse IgG (Rockland antibodies and assays, Limerick, PA) were used as controls. Aerosolized challenge doses of virulent Y. pestis CO92 (pneumonic plague model) were prepared as follows. The bacteria were streaked from freezer stocks onto tryptose blood agar base slants (BD Biosciences, #223220) and grown for approximately 48 hours at 30°C. Flasks of Heart infusion broth (HIB, BD Biosciences #23840) plus 0.2% xylose (Sigma Aldrich, St. Louis, MO, X1500) were inoculated with the bacterial growth from the freshly grown slants for an approximate initial OD₆₂₀ of 0.01 and were grown for approximately 24 hours at 30°C. The cultures were harvested by centrifugation and suspended in HIB medium (no xylose) to the estimated concentration yielding the desired number of LD₅₀. Exposure of mice to aerosolized bacteria was accomplished as previously described [73,74]. Briefly, mice were transferred to wire mesh cages and up to four wire mesh cages were placed in a whole-body aerosol chamber within a class three biological safety cabinet located inside a BSL-3 laboratory. Mice were exposed to aerosolized Y. pestis strain CO92 created by a three-jet collision nebulizer. Samples were collected from the all-glass impinger (AGI) vessel and analyzed by performing CFU calculations to determine the inhaled dose of Y. pestis. The median lethal dose for Y. pestis CO92 in female BALB/c mice is approximately 6.8 x 10⁴ inhaled CFUs [74].

Ethics statement.

Animal research at the United States Army Medical Research Institute of Infectious Diseases (USAMRIID) was conducted under an animal use protocol approved by the USAMRIID Institutional Animal Care and Use Committee (IACUC) in compliance with the Animal Welfare Act, PHS Policy, and other Federal statutes and regulations relating to animals and experiments involving animals. The facility where this research was conducted is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals (National Research Council, 2011). Challenged mice were observed at least daily for 21 days for clinical signs of illness. Humane endpoints were used during all studies, and mice were humanely euthanized when moribund, according to an endpoint criteria score sheet approved by the IACUC prior to beginning the study. Animals were scored on a scale of 0–16: 0–2 = no significant clinical signs (e.g., slightly ruffled fur); 3–7 = significant clinical signs such as subdued behavior, hunched appearance, absence of grooming, labored breathing of varying severity (increased monitoring was warranted); >8 = distress. Those animals receiving a score of > 8 were humanely euthanized by CO₂ exposure using compressed CO₂ gas or by injection of pentobarbital-based euthanasia solution; both methods of euthanasia were followed by cervical dislocation for confirmation of death. However, even with multiple observations per day, some animals died as a direct result of the infection.

Statistics.

Gentamicin protection assay results were compared by stratified Wilcoxon (Van Elteren) test. The log-rank test was used to compare mouse survival curves post-challenge and Fisher’s exact test was used to compare survival rates. Plaque counts were compared by stratified Wilcoxon (Van Elteren) test. Analysis was implemented in SAS version 9.4 (SAS Institute Inc., Cary, NC).