Broadly reactive anti-VHH antibodies for characterizing, blocking, or activating nanobody-based CAR-T cells

Abstract

Background: Production of chimeric antigen receptor T cell (CAR-T) therapies depends on antibody reagents to label, isolate, and expand T cell products. We sought to create antibody tools specific for the variable domain of heavy-chain only antibodies (VHHs), also known as nanobodies, used in some CARs.

Methods: We immunized a mouse with VHH and selected two murine monoclonal antibodies (mAbs) that bind to distinct epitopes in conserved framework regions of llama-derived VHHs, and not to human VH domains. Anti-VHH mAbs were characterized by enzyme-linked immunosorbent assay, surface plasmon resonance, and hydrogen-deuterium exchange mass spectrometry; were then tested for cell/tissue labeling and for modulating cellular activity in VHH-CAR-T cells.

Results: We produced a high-quality dual-clonal anti-VHH antibody product and confirmed reactivity to over 98% of VHH proteins regardless of their antigenic specificity, with no reactivity to human or mouse IgG and reduced reactivity to conventional llama or alpaca IgG. Anti-VHH binding did not disrupt VHH/antigen interaction, and thus was appropriate for secondary labeling to assess cellular or tissue reactivity of VHH molecules. Despite not interfering with antigen binding, anti-VHH antibodies (Abs) potently blocked VHH-CAR-T activation and cytolytic killing of target cells. When immobilized, anti-VHH Abs induced strong activation and expansion of VHH CAR-T cells; with 730-fold mean expansion, >94% CAR purity, and retained CD8/CD4 heterogeneity. Functionally, anti-VHH antibody-expanded CAR-T cells maintained strong antigen-specific activity without functional exhaustion.

Conclusions: Overall, these data identify useful anti-VHH mAbs that can be applied to better understand and manipulate VHH-based CAR-T cells or other VHH-based immunotherapies.

Keywords: CAR labeling; CAR-T; Chimeric Antigen Receptors; Nanobodies; VHH.

Figure 1

Selection of two broadly cross-reactive mouse anti-VHH mAbs by mouse immunization with a purified VHH. (A) Clonal hybridoma culture supernatants were screened via ELISA for reactivity to human total IgG protein, selecting only those with negative reactivity to human IgG for further testing. Non-human reactive supernatants were then screened via ELISA for cross-reactive binding to three different VHHs, with a total of 10 clones showing reactivity to 3 of 3 VHHs. (B) Flow cytometry was then conducted with Jurkat cells bearing stable expression of surface-linked VHH-chimeric antigen receptor (Jurkat-VHH-CAR) to confirm the reactivity of seven selected anti-VHH hybridoma supernatants. (C) Anti-VHH hybridoma supernatants were similarly screened for reactivity against primary CAR-T cells from two different healthy donor blood samples expressing five different VHH-CARs, two control scFv-CARs, or untransduced T cells. Bar graphs show the median fluorescence intensity of anti-VHH staining on gated EGFP+ CAR-expressing cells from one experiment performed in duplicate +/− SEM. Supernatants with the broadest cross-reactivity (2A3 and 3H12) were selected for downstream testing.

Figure 2

ELISA and SPR-based reactivity of 3H12 and 2A3 mAbs. (A) ELISA demonstrating 3H12 and 2A3 reactivity toward 76 passively absorbed llama-derived VHHs (“Cocktail” refers to an equimolar mix of 3H12 and 2A3); binding of mAbs to VHHs was detected with donkey-anti-mouse IgG-HRP as shown with absorbance readings at 450 nm. (B) Similarly, reactivity of 3H12 and 2A3 to a llama-VH domain, or a control VHH domain, was tested via ELISA, (C) summary of VHH detection by 3H12 and 2A3. (D) SPR sensorgrams showing monovalent binding affinities of select VHHs for mAbs. 3H12 and 2A3 mAbs were captured, and VHHs flowed at the concentration ranges indicated. NB: no binding. (E) VHH consensus sequences in FR1 (3H12) and FR2 (2A3) that may be predictors of mAb reactivity. (F) SPR sensorgram showing a single S11L mutation in FR1 of VHH-53 imparts 3H12 binding.

Figure 3

Anti-VHH mAbs are most reactive to VHH fraction of llama or alpaca serum. (A) Whole serum from human, mouse, and the camelid species llama, llama glama, and alpaca, Vicugna pacos, were assayed by protein ELISA to determine the species reactivity with a cocktail of antibody clones 2A3 and 3H12. (B) Similarly, 2A3 and 3H12 cocktail was examined via ELISA for reactivity to fractionated serum antibodies from llama, alpaca, human, and mouse. (C) Serum fractions were analyzed by SDS-PAGE to demonstrate that conventional IgG1 antibodies were isolated in the G2 fraction of serum from all species, while camelid heavy chain antibodies were isolated in the G1, A1, and A2 fractions of llama and alpaca serum. *Asterisks denote fractions where conventional VH/VL antibodies are expected.

Figure 4

Anti-VHH mAbs 3H12 and 2A3 bind to unique but adjacent epitopes. (A) YSD was used to map binding; inset table shows the % binding of anti-VHH mAbs via yeast whole-cell ELISA, normalized to the full-length VHH. (C) Hydrogen-deuterium exchange mass spectrometry was performed on purified VHH alone or bound to excess anti-VHH mAbs; VHH domain-wire diagram shows regions of significantly reduced (blue) or increased (red) hydrogen exchange when a VHH test material (1ug36) was saturated with 2A3 or 3H12 anti-VHH mAb. (C) Competitive SPR was performed using an amine coupled VHH, and mAbs were sequentially injected binding of the first anti-VHH mAb prevented most but not all binding of the second mAb, regardless of injection order, an irrelevant non-VHH binding mAb was included as a control (“ctrl”; cetuximab).

Figure 5

Anti-VHH mAbs can be used to assess cell and tissue binding of novel VHHs. (A) A human lymphoma cell line with high expression of CD22 (Ramos) or a similar line wherein CRISPR gene targeting was used to knockout CD22 expression (Ramos-CD22ko) was stained with purified CD22-specific VHHs (1ug36 or 1ug13 clones) or a control VHH of irrelevant specificity (B131), cells were then washed and stained with a mix of AlexaFluor-647 conjugated anti-VHH secondary mAb cocktail (3H12 and 2A3), and cells were examined via flow cytometry. (B) To demonstrate the general utility of this approach for VHHs of unknown sequence identity and cellular reactivity, a panel of BCMA-specific VHHs was similarly tested for reactivity to a BCMA+ multiple myeloma cell line (RPMI8226) or a BCMA-negative control cell line (Jurkat), followed by secondary staining with an anti-VHH cocktail. (C) To examine whether a similar approach can be used for assessment of tissue-level reactivity of VHHs, human tonsillar tissue known to have a high density of CD22+ B cells was stained with CD22-specific VHH, tissue was then stained with 3H12 anti-VHH mAb, and a tertiary anti-mouse HRP polymer detection reagent; images show expected follicular tissue distribution consistent with CD22-specific reaction.

Figure 6

Soluble anti-VHH mAbs can block CAR-T cell responses. (A-B) To test the effect of soluble anti-VHH mAbs, EGFR-specific VHH-CAR expressing Jurkat cells were combined with EGFR-expressing target cells and varying doses of anti-VHH mAbs separately or as a cocktail, incubating co-cultures for 48 hours before assessing CAR-Jurkat cell activation via CD69-staining and flow cytometry. (C) A similar experiment was performed with lentivirally transduced primary human CAR-T cells, wherein upregulation of both CD69 and CD25 activation markers was diminished in a dose-dependent manner with a soluble anti-VHH mAb. (D) Primary CAR-T cells in co-culture with red-fluorescent protein-expressing EGFR-high SKOV3 target cells were examined for target cell survival via live fluorescence microscopy using an Incucyte device; images show a clear loss of target cells when co-cultured with EGFR-specific VHH CAR-T cells (top left) but not unmodified human T cells (bottom), or in the presence of soluble anti-VHH mAb treatment (right image). (E) Quantitation of red fluorescent signal over time, as measured via Incucyte, showing that anti-VHH mAb treatment completely blocks CAR-T specific lysis of target cells.

Figure 7

Immobilized anti-VHH mAbs can be used to activate and expand VHH-CAR-T cells. (A) Schematic of experimental setup to test whether plate-bound anti-VHH mAbs were able to activate VHH-CAR expressing cells. (B) Varying concentrations of anti-VHH mAbs were absorbed to a 96-well plate overnight, before washing and addition of CAR-Jurkat cells for an additional 24 hours; cells were then stained with CD69 antibody and assessed via flow cytometry. (C) To confirm this effect across varying cells expressing various VHH-CAR constructs, CAR-Jurkat cells were stimulated with plate-bound 3H12, 2A3, a mix of both anti-VHH mAbs, or plate-bound anti-CD3, and similarly examined for activation status via CD69 staining and flow cytometry. (D) A similar experimental setup was used to confirm that plate-bound anti-VHH mAbs can stimulate activation and proliferation of primary human CAR-T cells that co-express a green fluorescent marker (mNeonGreen). (E) Similarly, primary VHH CAR-T cells show upregulation of the CD69 activation marker and (F) expansion of green fluorescent marker co-expressed with the CAR transgene.

Figure 8

Functional testing of anti-VHH activated CAR-T cells. (A) Schematic of experimental setup to activate VHH CAR-T cells with plate-bound anti-VHH mAbs, followed by functional assay using co-culture with various target cells. (B) The short-term response to CD22+ Ramos target cells was examined across a range of effector to target ratios, wherein the number of red-fluorescent target cells and green-fluorescent CAR-T cells was enumerated using flow cytometry after 7 days of co-culture; inset P values show two-way ANOVA comparison of overlaid curves in graphs. (C) Co-cultures of 20 000 CAR-T cells with an equal number of CD22+ Ramos, CD22-knockout Ramos, or CD22-negative H292 lung cancer target cells; to drive CAR-T exhaustion, additional target cells were added to cultures at days 3, 7, 10, 14, and 17; 1 to 5 media exchanges and cell culture splits were also performed on days 7 and 14, = monitoring growth of red-fluorescent target cells via live regular fluorescent microscopy. All graphs show the response of non-activated CAR-T cells (solid lines) versus anti-VHH activated cells (broken lines) from a single experiment.