TRIM21 and Fc-engineered antibodies: decoding its complex antibody binding mode with implications for viral neutralization

Abstract

TRIM21 is a pivotal effector in the immune system, orchestrating antibody-mediated responses and modulating immune signaling. In this comprehensive study, we focus on the interaction of TRIM21 with Fc engineered antibodies and subsequent implications for viral neutralization. Through a series of analytical techniques, including biosensor assays, mass photometry, and electron microscopy, along with structure predictions, we unravel the intricate mechanisms governing the interplay between TRIM21 and antibodies. Our investigations reveal that the TRIM21 capacity to recognize, bind, and facilitate the proteasomal degradation of antibody-coated viruses is critically dependent on the affinity and avidity interplay of its interactions with antibody Fc regions. We suggest a novel binding mechanism, where TRIM21 binding to one Fc site results in the detachment of PRYSPRY from the coiled-coil domain, enhancing mobility due to its flexible linker, thereby facilitating the engagement of the second site, resulting in avidity due to bivalent engagement. These findings shed light on the dual role of TRIM21 in antiviral immunity, both in recognizing and directing viruses for intracellular degradation, and demonstrate its potential for therapeutic exploitation. The study advances our understanding of intracellular immune responses and opens new avenues for the development of antiviral strategies and innovation in tailored effector functions designed to leverage TRIM21s unique binding mode.

Keywords: TRIM21; affinity; antibody Fc mutants; antibody mediated viral neutralization; avidity; binding kinetics; structure-function; therapeutic IgG.

Figure 1

SPR assay orientations to characterize the interaction between TRIM21 and an antibody. Symmetrical and asymmetrical antibody Fc variants are investigated. In case of an asymmetrical Fc part, one Fc heavy chain contains a AAA mutation (schematically shown by red star), that completely abolishes TRIM21 binding. The used Fc variants and assay setups allow determining how Fc mutations influence the avidity-binding mode and dissecting avidity from affinity. (A) Antibody Fc variants are captured on the biosensor surface via an anti-Fab nanobody (vhh), Fc-only variants are coupled using standard amine coupling chemistry and cytokine Fc-Fusions are captured via anti-PGLALA F(ab’)2 fragment (55), while TRIM21 PRYSPRY domain is the analyte in solution (see Materials and Methods). Configuration (B) schematically shows the inverse to (A) while the PRYSPRY domain is captured via monovalent streptavidin. (C) To analyze the dimeric TRIM21 engagement of both IgG heavy chains, the antibody is captured via its Fab fragment, cytokine Fc-Fusions are captured via anti-PGLALA F(ab’)2 fragment (identical capture setup as in (A) and TRIM21-coiled-coil-PYRSPRY (TRIM21-CC-PS) is injected. Illustrations are created with BioRender.com.

Figure 2

Interaction analysis of human IgG1 (mAb1) Fc variants and TRIM21 PRYSPRY domain. (A–C) showing sensorgrams (SPR data) where PRYSPRY was injected in five different concentrations as two-fold dilution series to immobilized mAb1 Fc variants (capture level approx. 60 RU). Each plot shows the measured raw data (colored gradient) and the global fit analysis as solid black lines. For immobilized mAb1 WT (A) and mAb1 WT-AAA (B) PRYSPRY was injected at 500 nM highest concentration and for mAb1 AAA (C) at 2000 nM. The sensorgrams show the affinity binding mode applying a mono-exponential fit model (Langmuir 1:1). The determined kinetic parameters are described in (D). The k_ON, k_OFF and K_D values are results from a global fit analysis ± fitting error. (E, F) show the complementary mass photometry (MP) data displaying a 2:1 binding stoichiometry confirming the SPR data. For the PRYSPRY - mAb1 WT complex, the data reveals a double bound state and for mAb1 WT-AAA a single bound state, while the control mAb1 AAA shows no binding at all. A Gaussian distribution model was used to analyze the MP data. For individual masses of the molecules, see SI Info Supplementary Figure S1 .

Figure 3

Kinetic characterization of TRIM21 PRYSPRY binding to immobilized antibody Fc variants and cytokine-Fc Fusion constructs, and Fc only variant (Raw data SI Info Supplementary Figure S2 ). Detailed SPR assay setup is described in materials and methods. (A) The Affinity Rate Scale Plot enables the kinetic comparison of several binding experiments at one glance. The association rate (k_ON) and corresponding dissociation rate (k_OFF) are juxtaposed in opposition, connected via a vertical line, representing the binding strength (affinity). The further apart both parameters (k_ON and k_OFF) the stronger the interaction is. Compared to mAb1 Fc WT, the Fc variants YTE (M252Y, S254T, T256E), HH (T307H, N434H) and Y436A show decreased PRYSPRY affinity. The YTE affinity is 1.7-fold, HH 2.4-fold and Y436A 180-fold decreased. As shown in (B) the altered binding strength is mostly off rate driven, which becomes apparent in the overlay of normalized dissociation. The start of dissociation is normalized to 100%.

Figure 4

Affinity rate scale plot for captured TRIM21 PRYSPRY domain (ligand) and antibody (human IgG1) variable domain variants or antigen fusion constructs in solution (analyte). The injected constructs have different Fab regions but share the same Fc region. This allows the investigation of a potential Fab contribution to the PRYSPRY binding. All constructs were analyzed by applying a simple 1:1 Langmuir fit. The analyzed antibody variants do not show any Fab contribution. Notably, there is a faster on-rate (2x) for all constructs when compared to the reverse assay setup up (PRYSPRY as analyte). Raw data is shown in SI Info Supplementary Figure S3 .

Figure 5

Characterization of the TRIM21 dimeric nature (TRIM21-CC-PS) applying different technologies. (A) SEC-MALS data reveals 94% TRIM21-CC-PS dimer (90 kDa). (B) Mass photometry technology shows 99% TRIM21-CC-PS with 82 kDa. (C) Selection of EM 2D classes confirming TRIM21-CC-PS dimers. The coiled coil domains facilitate dimerization whereas the C-terminal PRYSPRY domains are placed at the opposite end of each coiled-coil domain.

Figure 6

Characterizing the interaction of TRIM21-CC-PS with three different Antibody Fc variants. (A–D) Applying MP, dashed lines indicate the main peak of the respective species over all measurements. The applied 3-dimensional Gaussian Fit Distribution is shown in black lines. (A) MP of TRIM21-CC-PS with mAb 1 WT, (B) MP of TRIM21-CC-PS with mAb 1 WT-AAA (C) MP of TRIM21-CC-PS with mAb 1 AAA. Only mAb 1 WT shows binding to TRIM21-CC-PS at low nM concentration in accordance with its low nM binding strength. (D) The amount (%) of TRIM21-CC-PS - mAb1 WT complex increases with excess of TRIM21-CC-PS, while TRIM21-CC-PS - mAb1 WT-AAA shows single events of complexed species for the applied concentrations, that could not be fitted robustly. (E–H) selected 2D averages of EM data, resolving TRIM21-CC-PS with Fc WT (E), TRIM21-CC-PS with Fc WT-AAA (F), TRIM21-CC-PS with mAb1 WT (G) and mAb1 WT alone (H).

Figure 7

Characterization of TRIM21-CC-PS with Antibody Fc variants. (A) shows the sensorgram (SPR) of mAb1 WT (ligand, approx. 8–10 RU) and TRIM21-CC-PS (analyte) where TRIM21-CC-PS was injected in seven different concentration, each for 180 sec as two-fold dilution series with 100 nM as highest concentration. Applied fit model is a simple 1:1 interaction reflecting 100% avid bound, 1:1 antibody - TRIM21-CC-PS species. (B) Variation in association time (10 -300 sec) injecting a constant concentration of 25 nM TRIM21-CC-PS to captured mAb1 WT reveals a biphasic binding kinetic, which can be described by applying a two state model providing fast and slow kinetic rates. (C) Rate-scale-plot comparing affinity and avidity measurements of Fc variants towards PRYSPRY or TRIM21-CC-PS. (D) The altered binding strength from affinity to avidity is mostly off rate driven, which becomes apparent in the overlay of normalized dissociation phases (k_OFF,AVIDITY) but can also occur as combination of both kinetic rate parameters, namely on and off rate.

Figure 8

TRIM21-CC-PS-Antibody-AAV2 Characterization. SPR Assay data is shown in (A, B). (A1, A2) Schematic SPR assay configuration to analyze the affinity to avidity interplay of TRIM21-CC-PS, anti-capsid antibody variants A20 and rAAVv-2. Biotinylated TRIM21-CC-PS is captured via monovalent streptavidin achieving a captured level of 190 RU (A1, high density) and 35 RU (A2, low density). Subsequent, anti-AAV2 capsid antibody variants (bivalent A20 Fc WT, one-armed A20 Fc WT and Fc WT-AAA) are injected to saturate the TRIM21-CC-PS surface, followed by the injection of rAAV-2. (B) Overlay of the normalized dissociation phases (Start of Dissociation: 100%) after the injection of 3.32 nM rAAVv-2 over low and high TRIM21-CC-PS-A20 densities. At higher antibody densities, more avid complexation occurs and a higher degree of rAAVv-2 surface decoration is possible. This allows less complex to dissociate over time to due to simultaneous engagement of both, TRIM21-CC-PS and AAV2, mediated via the A20 antibody variants. (C) Electron microscopy images of rAAVv-2 interactions with antibodies alone (left column) and TRIM21 additionally (right column). The scale bars represent 50 nm.

Figure 9

Schematic model suggesting how one TRIM21 dimer engages both sites of the Fc region in a two-step process. Upon Fc binding, the PRYSPRY detaches from the coiled-coil domain. The linker domain allows enough freedom of movement to allow engagement of the second PRYSPRY domain. Only after initial binding bivalent engagement of both Fc heavy chains is possible.