Development and evaluation of single domain antibodies for vaccinia and the L1 antigen

Abstract

There is ongoing interest to develop high affinity, thermal stable recognition elements to replace conventional antibodies in biothreat detection assays. As part of this effort, single domain antibodies that target vaccinia virus were developed. Two llamas were immunized with killed viral particles followed by boosts with the recombinant membrane protein, L1, to stimulate the immune response for envelope and membrane proteins of the virus. The variable domains of the induced heavy chain antibodies were selected from M13 phage display libraries developed from isolated RNA. Selection via biopanning on the L1 antigen produced single domain antibodies that were specific and had affinities ranging from 4×10(-9) M to 7.0×10(-10) M, as determined by surface plasmon resonance. Several showed good ability to refold after heat denaturation. These L1-binding single domain antibodies, however, failed to recognize the killed vaccinia antigen. Useful vaccinia binding single domain antibodies were isolated by a second selection using the killed virus as the target. The virus binding single domain antibodies were incorporated in sandwich assays as both capture and tracer using the MAGPIX system yielding limits of detection down to 4×10(5) pfu/ml, a four-fold improvement over the limit obtained using conventional antibodies. This work demonstrates the development of anti-vaccinia single domain antibodies and their incorporation into sandwich assays for viral detection. It also highlights the properties of high affinity and thermal stability that are hallmarks of single domain antibodies.

Figure 1. Sequences of L1 binding sdAb.

The alignment shows the sequences of L1 binders isolated in sandwich-style selection. Three families, based on homology within the CDR regions were isolated. CDR regions are underlined. Red denotes positions where the amino acid is 90% conserved in the compared sequences, blue indicates low consensus (50%) and black are not conserved. The most dominant family (L1-H7 family) showed some variation in amino acid sequences of the framework regions that showed an impact of binding affinities and thermal stability. Arrows mark the positions of the cysteine substitutions in the L1-G2+ construct.

Figure 2. Sandwich immunoassay for L1 detection.

These assays were performed on at least two different days; representative binding assays are shown. Panels A-D are MAGPIX assays and are plotted as signal over background versus the L1 concentration; a signal over background of 3 was set as the limit of detection and is indicated by a blue line. Each MAGPIX assays measures binding from at least 50 beads and the error of the mean fluorescence intensity is within 5%. Panel A, mAb419 capture with the biotinylated (bt)-sdAb tracers. Panel B, mAb417 capture with the bt-sdAb tracers. Panel C shows combinations of the mAb capture and tracer pairs. Panel D shows the sdAb L1-E6 and L1-H7 captures paired with the mAb reporters. Panels E and F are ELISA data, each point represents the average of three measurements and the error bars represent the standard deviation. Control lines were collected with the bt-L1-H2 tracer in combination with a non-L1 binding capture sdAb. Panel E shows sdAb capture and mAb reporters. Panel F shows sdAb capture and tracer pairs.

Figure 3. Refolding of sdAb.

Circular dichroism was used to monitor the secondary structure of purified sdAbs over a temperature range of 25–95°C. Representative plots are shown of clones H7 and H2 typifying the distinctly different refolding properties. Temperature is expressed in °C and the change in ellipticity is in milidegrees. CD experiments were performed in duplicate for most of the sdAbs; the observed Tm values agreed within a degree, and the refolding trends were the same among the replicate measurements.

Figure 4. Thermal stability assay.

Following a one hour incubation at the defined temperature optical density at 280 nm was measured to assess protein aggregation (upper panel). Samples G2 (blue triangles) and G2+ (gray circles) were measured in duplicate, and showed the same solubility behavior in each replicate. Since all of the mAbs behaved similarly they are all represented by the same symbol (red upside-down triangle). Binding activity following incubation was measured using a direct binding assay with the L1 antigen directly immobilized to four rows of a sensor chip for SPR analysis (lower panel). The mAbs have the same color and symbol but are differentiated by line type: mAb419 with short dashes, mAb418 with long dashes, and mAb417 with a solid line.

Figure 5. Increased melting temperature of L1-G2+.

Fluorescence based melting assay showing melting of L1-G2 in blue and the mutant with the added disulfide, L1-G2+, in black. Melting temperature is determined from a plot of the first derivative of the melt curve. The measurements were performed in triplicate on two different days and showed identical results; for clarity only one replicate is shown.

Figure 6. Specificity of L1 binding sdAbs.

Direct binding assays were used to demonstrate target specificity for L1 binding antigens. Antigen was immobilized to magnetic beads and biotinylated (bt)- sdAbs (noted at the top of each graph) served as the tracers. Independent direct binding assays were performed different days; a representative data set is shown. The MAGPIX assays typically measures binding from at least 50 beads and the error of the mean fluorescence intensity (MFI) is ≤5%.

Figure 7. Amino acid alignments of vaccinia virus binding sdAbs.

The representatives from each family that were chosen for further characterization are presented. Red denotes positions where the amino acid is 90% conserved in the compared sequences, blue indicates low consensus (50%) and black are not conserved. The three CDRs are underlined.

Figure 8. Direct binding ELISA to assess binding to inactivated vaccinia virus.

The anti-L1 mAbs 417 and 419 as well as sdAb selected on L1 (L1-E6, L1-E2, L1-G2, L1-H2 and L1-H7) and sdAb selected on vaccinia (Vacc-E7 and Vacc-D9) were tested for binding to inactivated vaccinia virus immobilized on the wells of a 96-well plate. Measurements were done in triplicate; data is plotted as signal minus background, and the error bars represent the standard deviation.

Figure 9. MAGPIX limit of detection assays for vaccinia virus.

Top two panels show purified sdAbs used as both captures and tracers in bead-based assays to detect the vaccinia virus. The bottom panel shows the assay using conventional antibodies. Capture antibodies are plotted with corresponding biotinylated (Bt) tracer sdAbs indicated above each graph. The Y-axis is a measure of the signal over background while the X-axis represents the concentration of the vaccinia virus. Insets show a close-up of the lower part of the graph, the blue line indicates the LOD, defined as a signal to background ratio of 3. Limits of detection for best performing sdAb capture-tracer pairs approach 4×10⁵ pfu/ml. Sandwich assays for vaccinia detection were performed on multiple days and reproduced; a representative data set is shown. The MAGPIX assays measures binding from at least 50 beads and the error of the mean fluorescence intensity is within 5%.