Antibody-epitope conjugates deliver immunogenic T-cell epitopes more efficiently when close to cell surfaces

ABSTRACT

Antibody-mediated delivery of immunogenic viral CD8⁺ T-cell epitopes to redirect virus-specific T cells toward cancer cells is a promising new therapeutic avenue to increase the immunogenicity of tumors. Multiple strategies for viral epitope delivery have been shown to be effective. So far, most of these have relied on a free C-terminus of the immunogenic epitope for extracellular delivery. Here, we demonstrate that antibody-epitope conjugates (AECs) with genetically fused epitopes to the N-terminus of the antibody can also sensitize tumors for attack by virus-specific CD8⁺ T cells. AECs carrying epitopes genetically fused at the N-terminus of the light chains of cetuximab and trastuzumab demonstrate an even more efficient delivery of the T-cell epitopes compared to AECs with the epitope fused to the C-terminus of the heavy chain. We demonstrate that this increased efficiency is not caused by the shift in location of the cleavage site from the N- to the C-terminus, but by its increased proximity to the cell surface. We hypothesize that this facilitates more efficient epitope delivery. These findings not only provide additional insights into the mechanism of action of AECs but also broaden the possibilities for genetically fused AECs as an avenue for the redirection of multiple virus-specific T cells toward tumors.

KEYWORDS:

• antibody-epitope conjugates (AECs)
• virus-specific T-cells
• immunotherapy
• redirecting T-cells
• bispecific-antibodies

Introduction

The use of antibody-epitope conjugates (AECs) has recently emerged as a new approach in which CD8⁺ virus-specific T cells are redirected toward cancer cells.^1–6 AECs rely on antibody-mediated delivery of immunogenic virus T-cell epitopes to cancer cells and have demonstrated their effectiveness with multiple antibody targets and epitopes originating from Epstein Barr virus (EBV) or cytomegalovirus (CMV).^7,8 Increasing immunogenicity of tumors through delivery of viral epitopes from EBV and CMV is attractive since these viruses are highly prevalent in the human population and are known to induce a potent CD8⁺ T-cell memory response.^8–10

For AECs, multiple release strategies have proven to be effective, ranging from release within the endo-lysosomal pathway,¹ the extracellular environment,^2,3,5,6 or the cytoplasm.⁴ AEC strategies that rely on extracellular delivery use viral epitopes with a free C-terminus. Protease expression levels and the amino acids/protease cleavage site in front of the epitope can affect the therapeutic efficiency of these AECs.^5,6 Extension of epitopes by one or a few amino acids at the C-terminus abolishes the capacity of AECs to deliver the epitope in MHC class I¹¹ unless the peptide epitope is imported into the cytoplasm.⁴ This suggests that a free C-terminus may be essential for extracellular delivery of epitopes.

Previously we demonstrated that epitopes can be genetically fused to either the C-terminus of the light chain (LC) or heavy chain (HC) of an antibody.¹² However, the efficiency of viral epitope delivery for AECs with epitopes fused to the C-terminus of the LC was reduced, possibly due to reduced accessibility. We therefore explored whether it would be possible to fuse the viral epitope to the N-terminus of the LC instead of the C-terminus. The data presented here demonstrate that this is feasible, and interestingly these AECs are even more efficient in delivering the viral epitopes to cancer cells than AECs in which the epitopes are fused with the C-terminus of either the LC or the HC. This approach increases and broadens the options for the development of AECs for use in therapeutic strategies.

Materials and methods

Antibodies and peptides

All AECs and trastuzumab were produced at Genmab via transient expression in ExpiHEK293 FreeStyle cells as previously described.¹³ The proteins were purified by Protein A affinity chromatography, and, if required, protein aggregates were removed via size-exclusion chromatography (SEC) to yield protein product with a > 95% monomeric content as analyzed on HPLC-SEC. Non-modified cetuximab was sourced from Merck (Germany). The amino acid sequence attached to the C-terminus of the heavy chain was: -GGSGLSGRSDNHYVLDHLIVV, and to the N-terminus of the LC was: YVLDHLIVVLSGRSDNHGGSG-. The BRLF1-YVL epitope is underlined. All antibodies used in the coculture experiments were stored in phosphate-buffered saline (PBS) at −80°C. The following antibodies were used for flow cytometry: cetuximab, trastuzumab, Goat Anti-human IgG-A488 (Jackson ImmunoResearch, #109-546-098) or -PE (Jackson ImmunoResearch, #109-116-098). The peptides used in the coculture experiments are indicated in Table 1 and were dissolved in dimethyl sulfoxide at a concentration of 20 mg/ml. All peptides were synthesized with Fmoc chemistry, and their identity was confirmed with mass-spectrometry.

Table 1. Overview of the peptides used in the coculture assay of figure 1A. The EBV epitope is underlined.

Peptide abbr.	Peptide	Protease
YVL	YVLDHLIVV
Cl-1-YVL	LSGRSDNH-YVLDHLIVV	uPa, matriptase, legumain
YVL-Cl-1	YVLDHLIVV-LSGRSDNH
YVL-Cl-2	YVLDHLIVV-PRSAKELR	MMP-14
YVL-Cl-3	YVLDHLIVV-VPLSLYSG	MMP-2, -7 and 9

Generation and analysis of bispecific antibodies

The following monoclonal antibodies and AECs were produced with either the K409R or the F405L mutation; CTX-F405L, CTX-NL-F405L, b12-K409R, and b12-NL-K409R. cFAE was performed as previously described¹⁴ for the following antibody combinations: CTX-F405L and b12-NL-K409R (bs-CTXxb12-NL) and CTX-NL-F405L and b12-K409R (bs-CTX-NLxb12). To determine whether cFAE was successful, bispecific IgG1 molecules were cleaved specifically above the hinge region using FabALACTICA (Genovis) into intact and homogeneous Fab and Fc fragments. The relative intensities of Fc domains from parental homodimer and bispecific IgG were determined by mass spectrometry.

Cell lines

All adherent cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco), 1% Pen/Strep (Gibco), 10% Fetal Calf Serum (FCS, Biowest). The generation of KO cell lines and the transduction of HLA-A2, tEGFR and tHer2 were performed as previously described.⁶

The T cells used in the experiments were EBV-BRLF1 (YVLDHLIVV presented in HLA-A *02:01) specific T-cell clones or CD8⁺ T cells derived from peripheral blood mononuclear cells (PBMCs) that were transduced with the virus-specific T-cell receptor as previously described.¹² T cells were cultured in a T-cell medium; Iscove’s Modified Dulbecco’s Medium (IMDM, Lonza), containing 5% FCS (Gibco), 5% human serum, 3 mM L-glutamine (Lonza), 1% Pen/Strep, and 200 IU/ml IL-2, and stimulated every 10–16 days with phytohemagglutinin (PHA) and irradiated allogeneic peripheral blood mononuclear cells in a 1:3 ratio. To remove expansion-related cytokines, T cells were washed 3 times with IMDM supplemented with 0.5% human serum albumin (HSA, Albuman, and Sanquin) before they were used in coculture experiments.

FACS experiments

In a 96-well, U-bottom plate, 100,000 cells/well were plated. The cells were washed 2× with PBS supplemented with 0.5% bovine serum albumin and 0.02% sodium azide (PBA) and incubated on ice for 30 min with the primary antibody. Hereafter, cells were washed 2× with PBA and incubated 20 min on ice with a secondary antibody. The cells were washed 2x with PBA and analyzed with fluorescence-activated cell sorting (FACS) on a LSRII (BD). The amount of cell surface receptors expressed by the cell lines was quantified with the QIFI kit (Agilent Dako) according to the manufacturer’s instructions. The data were analyzed with FlowJo software (v10.7.1).

Coculture assays

To determine T-cell activation, 5,000 target cells/well were cultured overnight in a 384-well flat-bottom tissue culture plate to allow them to adhere. The following day, the target cells were exposed to titrated peptide or antibody dilutions, which were prepared in IMDM supplemented with 0.5% HSA and incubated for 30 min or 1 hr at 37°C, respectively. To remove non-bound antibody, the cells were washed 2× with IMDM supplemented with 0.5% HSA. Next, 4,000 T cells/well were added in IMDM supplemented with 0.5% HSA and 100 IU/ml IL-2. After an overnight coculture, interferon (IFN)-γ production in the supernatant was measured by ELISA as a marker for T-cell activation (Diaclone).

T-cell cytotoxicity measurements

To measure T-cell cytotoxicity on the adherent cell lines, the 384-wells plates were, after harvesting the supernatant, supplied with a T-cell culture medium and incubated for additional 48 hours at 37°C. To remove the T cells, the plates were washed 3× with IMDM supplemented with 0.5% HSA. Next, a DMEM culture medium supplemented with 10% AlamarBlue HS cell viability reagent (ThermoFisher) was added. Viability was measured according to manufacturer’s protocol in relative fluorescence units (RFU), and killing was calculated as previously described.⁶

Statistical analysis

The used statistical tests are indicated in the figure legends. The calculations have been performed in GraphPad Prism software (V9.3.1.). The significance levels are indicated as*p < 0.05, **p < 0.01 and ***p < 0.001.

Patient materials

Healthy donor material from the Leiden University Medical Center Biobank for Hematological Diseases was used in this study. This research was approved by the Institutional Review Board of the Leiden University Medical Center (approval number B16.039). Materials were obtained after written informed consent in accordance with the Declaration of Helsinki.

Results

Peptide epitopes extended with a protease cleavage site at the C-terminus can activate CD8⁺ T cells

Before the T-cell epitope and the protease cleavage site were fused to the N-terminus of the LC of an antibody, we studied whether the BRLF1-YVL peptide (YVL) extended with a C-terminal cleavage site could be processed and presented by tumor cell lines. It was previously demonstrated that a cleavage site designed to be cleaved by the proteases urokinase-type plasminogen activator (uPA), membrane-type serine protease 1 (MT-SP1/matriptase), and legumain (Cl-1-YVL) could be efficiently processed and presented by cancer cells when placed at the N-terminus of the peptide epitope.^12,15 To expand this concept, we decided to analyze C-terminal extended peptide epitopes containing three different protease cleavage sites. The first uses the same cleavage site, designed to be cleaved by the uPA, MT-SP1/matriptase, and legumain (YVL-Cl-1), while the second cleavage site (YVL-Cl-2) was designed for membrane metalloprotease (MMP)-14, and the third (YVL-Cl-3) for MMP-2, −7, and −9 (Table 1).

HeLa cells transduced with HLA-A2 (HeLa-A2) were exposed to a titration of the various peptides and cocultured for 18 hrs with BRLF1-specific T cells. T-cell activation was measured in the supernatant by means of an IFN-γ ELISA (Figure 1(a)). A 10–100-fold difference was observed between YVL, which does not require proteolytic processing, and all peptide epitopes containing a protease cleavage site. Interestingly, no large differences could be observed in the efficiency of activation of the T cells between the peptides with different C-terminal extensions, in comparison to the previously used N-terminal fusion Cl-1-YVL. We conclude that the BRLF1-YVL epitope with either C- or N-terminal extensions can be processed and presented in HLA-A2 by the tumor cells and that N- and C-terminally extended EBV epitopes can both be processed, presented, and recognized in a T-cell activation assay.

Figure 1. AECs, with epitopes attached to the N-terminus of the LC can very efficiently deliver the BRLF1-YVL epitope. (a) HeLa-A2 cells were pulsed with titrated concentrations of the different peptides, followed by an overnight coculture with the EBV-specific T cells. The peptides contained either a cleavage site on the C- or N-terminus of the epitope (see Table 1 for a list of the peptides) and IFN-γ was measured in the supernatant with an ELISA to determine T-cell activation. (b) Schematic overview of the two AECs of CTX and TRS with either an EBV epitope at the C-terminus of the heavy chain (AEC-CH) or an epitope at the N-terminus of the light chain (AEC-NL) (c,d) To determine whether the epitope at the N-terminus of the light chain influences target binding, a flow cytometry experiment was performed for both the CTX- and TRS-AECs. CTX-AECs were titrated on HeLa-A2 cells (C) and TRS-AECs on HeLa-A2 cells transduced with truncated Her2 (HeLa-A2 tHer2) (D) and compared to the wildtype (WT) antibody. A representative graph of two individual experiments is shown. (e,f) HeLa-A2 (E) or -tHer2 (F) were exposed to different AECs, followed by an overnight coculture with the BRLF1-YVL-specific T cells. IFN-γ in the supernatant was measured with an ELISA to determine T-cell activation. Plotted values are the means of duplicates (SEM) of a representative experiment out of n=4. (g,h) From the coculture experiments in E and F the EC50 values were calculated. Plotted values are the means of duplicates (SEM) of four independent experiments. A paired t-test is performed to determine whether the difference between AEC-CH and –NL was significant.

AECs can be generated with EBV epitopes fused to the N-terminus without influencing target recognition. When the epitope is fused to the N-terminus of cetuximab and trastuzumab, an improved epitope delivery and therefore T-cell activation is observed compared to AECs with the EBV epitope fused to the C-terminus of the heavy chain of the antibody.

EBV-epitope fusion to the N-terminus of the light chain of CTX and TRS does not affect binding of the antibody to its target

To assess the impact of fusion of EBV epitopes to the N-terminus of the LC on the binding of cetuximab (CTX) and trastuzumab (TRS) to their respective targets, the N-terminus was visualized on the crystal structures of the antigen-binding fragment-domains (Fab-domains) bound to their respective targets: the epidermal growth factor receptor (EGFR, PDB 1YY9) and the human epidermal growth factor receptor 2 (Her2, PDB 1N8Z) (Figure S1). For both antibodies, the N-terminus of the LC is not in the immediate vicinity of the target-binding paratope, indicating that interference with binding will most likely be minimal.

Since no clear differences in stimulatory capacity between the Cl-1, −2 and −3 N-terminally extended peptides was observed (Figure 1(a)), we selected Cl-1 for further studies, as this is also the cleavage site used for the C-terminal fusion. AECs of CTX and TRS were produced with the BRLF1-YVL epitope and Cl-1 cleavage site genetically fused to the N-terminus of the LC (AEC-NL). The effectiveness of these AEC-NLs was compared with the AECs that were previously generated in which the BRLF1-YVL epitope was coupled to the C-terminus of the HC (AEC-CH) of both CTX and TRS (Figure 1(b)).⁶ AECs with the epitopes fused to the C-terminus of the LC were not taken along because we have previously demonstrated that these AECs have a reduced stimulatory capacity compared to AEC-CH.¹² To determine whether target recognition was influenced by the extension of the N-terminus, the AEC-NLs were titrated on the target-positive HeLa-A2 cell lines alongside the wildtype antibody (WT) and AEC-CH for both CTX and TRS, and antibody-target binding was measured by flow cytometry (Figure 1(c,d)). Our results clearly show that target recognition of the AEC-NL of CTX and TRS is comparable to their WT and AEC-CH counterparts, indicating that fusion of an epitope to the N-terminus of the LC of both CTX and TRS has no measurable impact on EGFR or Her2 target recognition, respectively.

AECs with epitopes at the N-terminus of the light chain have increased efficiency in T-cell epitope delivery

To investigate whether AEC-NL was able to deliver the viral T-cell epitope to antibody target-positive tumor cells and to compare their efficiency to AEC-CH, HeLa-A2 or HeLa-A2 cells transduced with a truncated Her2 lacking the intracellular domain (HeLa-A2 tHer2) were exposed to WT, AEC-CH or -NL of CTX and TRS, respectively. This was followed by an overnight coculture with BRLF1-specific CD8⁺ T cells. T-cell activation was determined by measuring IFN-γ in the supernatant. Both CTX- and TRS-NL showed a 10-fold higher efficiency in stimulatory capacity compared to their AEC-CH counterparts (Figure 1(e,f)). Strikingly, this difference was not observed with the previously tested extended peptides (Figure 1(a)). Moreover, when the experiment was repeated with non-specific T cells, no T-cell activation was observed for AEC-NL (Figure S2), which was in line with our previous findings for AEC-CH.¹² The half maximal effective concentration (EC₅₀) values calculated from 4 to 5 repeated experiments showed a consistent and significant difference between AEC-NL and -CH for both CTX and TRS (Figure 1(g,h)). Next, a viability assay was performed to investigate whether the increased T-cell activation also resulted in increased tumor cell kill. In accordance with the enhanced T-cell activation as measured by cytokine production, a 10-fold higher efficiency was also observed in tumor cell kill (Figure 2(a,b)). From these data, we conclude that AEC-NL can deliver the BRLF1-epitope more efficiently to antibody target positive tumor cells than AEC-CH, resulting in increased T-cell activation and tumor cell killing at low AEC concentrations on antibody target-positive HeLa-A2 cell lines.

Figure 2. AEC-NL triggers target cell killing with increased efficiency and is antibody target specific. (a,b) To determine whether the AEC-NLs resulted in a more efficient target cell killing, an AlamarBlue assay was performed, in which HeLa-A2 cells were incubated with AECs and cocultured for 72 hrs with the virus specific T cells. Plotted values are the means of duplicates (SEM) of a representative experiment out of n=4. (c,d) Both CTX (C) and TRS (D) AECs were tested for their target specificity on Hela-A2 EGFR KO (C) or Her2 KO (D) cell lines. The target cell lines were pulsed with 16 nM of the different AECs followed by an overnight coculture with the BRLF1-YVL-specific T cells. T-cell activation was measured with an IFN-γ ELISA. Plotted values are the means of duplicates (SEM) of four independent experiments. To determine the statistical significance, an RM one-way ANOVA with Tukey’s multiple comparisons was performed.

As demonstrated in these graphs, target cell killing is improved when the epitope is placed at the N-terminus of the light chain, and all AECs demonstrate to be antibody target-specific independent of where the epitope is placed.

To determine whether the observed T-cell activation by AECs-NL is indeed target specific, coculture experiments with WT, AEC-NL and -CH of CTX and TRS were repeated with the previously established EGFR and Her2 knockout (KO) HeLa-A2 cell lines (Figure 2(c,d)).⁶ No T-cell activation was measured when the respective KO HeLa-A2 cell lines were exposed to CTX- and TRS-NL, corroborating the antigen specificity of T-cell activation.

Enhanced epitope delivery was also observed on other antibody target positive tumor cell lines

To determine whether the increased stimulatory capacity of the AECs-NL could also be observed on other cancer cell lines, we exposed five tumor cell lines from different origins: MDA-MB231 (breast cancer), H292 and A549 (lung cancers), 518A2 (melanoma cancer), and SKOV3 (ovarian cancer) to AECs CTX-NL and -CH. Since A549, H292 and SKOV3 cell lines do not endogenously express HLA-A2, the cell lines were transduced with HLA-A2-expression vectors yielding A549-A2, H292-A2, and SKOV3-A2. All cell lines, except for 518A2, expressed antibody target EGFR to a high extend (>60,000 EGFR molecules/cell, Figure S3).

We exposed all five tumor cell lines to CTX-NL and -CH and cocultured the tumor cells overnight with BRLF1-YVL-specific T cells. Those tumor cell lines that already potently activated the BRLF1-specific T cells upon exposure to CTX-CH (MDA-MB231 and H292-A2), did not further improve T-cell activation upon exposure to CTX-NL (Figure 3(a,b)). In contrast, tumor cell lines with low (A549-A2 and 518A2) or moderate T-cell activation (SKOV3-A2) when exposed to CTX-CH showed pronounced T-cell activation after exposure to CTX-NL (Figure 3(c–e)). Strikingly, in the case of the 518A2 cell line, this is done in the context of relatively low EGFR cell surface expression levels (Figure S3). From these data, we conclude that placing the T-cell epitope at the N-terminus of the LC can improve the efficacy and potency of AECs compared to AECs with T-cell epitopes attached at the C-terminus of the HC.

Figure 3. AECs-NL demonstrate equivalent or higher potency compared to AECs-CH. To determine whether AEC-NL would also demonstrate increased epitope delivery efficiency on other antibody target-positive cell lines, the coculture experiments for CTX-AECs were repeated with the (a) MDA-MB231, (b) H292-A2, (c) A549-A2, (d) 518A2 and (E) SKOV3-A2 cell lines. (a–e) The upper graphs show representative figures of 3–4 independently performed experiments of the coculture experiment. Plotted values are the means of duplicates (SEM). From the coculture experiments, the EC50 values were calculated and plotted in the lower graphs. A paired t-test is performed to determine whether the difference between AEC-CH and –NL was significant. As no clear T-cell activation was observed on the 518A2 and A549-A2 cell lines when exposed to CTX-CH, the EC50 could not be properly calculated, and was either estimated or when no T-cell activation was observed set to >10^5.

The increased efficiency of AECs with an epitope fused to the N-terminus of the light chain was also observed on three other cell lines.

Closer proximity to the cell surface results in enhanced T-cell epitope delivery

The efficiency of T-cell epitope delivery was enhanced when tumor cells were exposed to AEC-NL compared to AEC-CH, whereas no differences were observed when tumor cells were exposed to the N- and C-extended peptides (Figure 1(a)). We hypothesize that this difference in T-cell epitope delivery may be due to the difference in location of the T-cell epitope on the antibody and that proximity of the protease cleavage site-linked T-cell epitope to the cell membrane may influence the efficiency of delivery. The distance between the cell surface and the protease cleavage site-linked T-cell epitope is smaller for AEC-NL compared to AEC-CH, and consequently also the distance between T-cell epitopes, proteases and HLA molecules expressed on the cell surface. To test this hypothesis, bispecific-AECs (bs-AECs) were generated with CTX and the monoclonal antibody b12, which is a neutralizing antibody recognizing the human immunodeficiency virus type 1 (HIV-1) gp120 protein. As this protein is not expressed by cancer cells, the Fab-domain of b12 is not able to recognize its cognate antibody target on the used cell lines (Figure S4). The generated bispecific antibodies contained either the BRLF1-YVL epitope on the N-terminus of the LC of either the CTX arm (bs-CTX-NLxb12) or b12 arm (bs-CTXxb12-NL) of the bispecific antibody (Figure 4(a)). As both bsAECs contain the EBV epitopes on the LC of the N-terminus, potential differences in efficiency can only be caused by the difference in distance or orientation toward the cell surface. The bispecific antibodies were generated with Duobody technology,¹⁴ and the efficiency of controlled Fab Arm Exchange (cFAE) was determined by MS analysis. CTX contains additional N-glycosylation sites in the Fab-domain that cannot be removed by PNGase treatment. To allow MS detection of the Fc fragments, the Fab-domains were therefore enzymatically cleaved off above the hinge. Relative quantification was determined by the mass difference between Fc fragments of the homodimers and bispecific antibodies (due to the K409R and/or F409L mutations). This analysis showed that cFAE was successful for both bs-CTX-NLxb12 and bs-CTXxb12-NL (Figure S5).

Figure 4. The distance between the cell surface and the protease cleavage site and epitope determines the efficiency. (a) Overview of the two bispecific antibodies generated with either an -NL epitope on the Fab-arm of CTX (bs-CTX-NLxb12) or b12 (bs-CTXxb12-NL). (b) HeLa-A2 or (c) SKOV3-A2 cells were exposed to titrated concentrations of CTX-NL, bs-CTX-NLxb12, or bs-CTXxb12-NL and cocultured with the BRLF1-specific T cells. T-cell activation was measured by determining the levels of IFN-γ secretion by the T cells within the supernatant after overnight coculture. Plotted values are the means of duplicates (SEM) and each graph shows a representative figure of more than three independently performed experiments. (d,e) From repeated coculture experiments the EC50 values were calculated. Plotted values are the means of duplicates (SEM) of four independent experiments. A paired one-way ANNOVA was performed to determine whether the differences between the AECs were significant. As no clear T-cell activation was observed on the SKOV3-A2 cell line when exposed to bs-CTXxb12-NL, the EC50 could not be calculated, and was estimated to be >10^5.

By generating two bispecific AECs with one epitope, we demonstrate that AECs have an increased efficiency when the epitopes are delivered in a closer proximity toward the cell surface.

Next, HeLa-A2 and SKOV3-A2 cell lines were exposed to CTX-NL, bs-CTX-NLxb12, and bs-CTXxb12-NL, followed by a coculture with the BRLF1-specific T cells. For the HeLa-A2 cells, bs-CTX-NLxb12 showed a higher efficiency compared to bs-CTXxb12-NL (Figure 4(b)). The difference between bs-CTX-NLxb12 and bs-CTXxb12-NL was even more pronounced on the SKOV3-A2 cell line, where no T-cell activation was observed for bs-CTXxb12-NL (Figure 4(c)). Moreover, a clear reduction in T-cell activation was observed on the HeLa-A2 and SKOV3-A2 cell lines when the epitope-to-antibody ratio was reduced from 2 (CTX-NL) to 1 (bs-CTX-NLxb12). These differences were also observed when the EC₅₀ values were calculated from the repeated experiments and showed for both cell lines a significant difference between bs-CTX-NLxb12 and bs-CTXxb12-NL (Figure 4(d,e)). This indicates that the distance toward the cell surface indeed influences the efficiency of epitope delivery.

Discussion

We have previously demonstrated that tumor specific-antibodies can be used as a delivery vehicle for viral T-cell epitopes, thereby sensitizing tumor cells for attack by virus-specific CD8⁺ T cells.^6,12 So far, most extracellular delivery strategies have used epitopes with a free C-terminus that were either chemically conjugated or genetically fused to the antibody. Here, we demonstrate that it is also feasible to couple viral T-cell epitopes to the N-terminus of an antibody’s LC instead of the C-terminus of the HC and that this strategy results in improved sensitization of tumor cells, leading to an increased efficacy and potency of the virus-specific CD8⁺ T cells attacking tumors. This approach broadens the concept of the AEC strategy and increases the number of possibilities in AEC design and engineering.

In this study, we demonstrate that CTX- and TRS-AECs with viral T-cell epitope attached to the N-terminus of the LC can be generated and that these AECs display similar binding characteristics as the WT antibodies (Figure 1(c,d)). This might not always be the case, as the T-cell epitope is placed close to the antigen-binding domain and therefore could interfere with target recognition. Moreover, being able to attach a T-cell epitope to the LC of an antibody without loss of antigen-binding characteristics could increase the options for linking more immunogenic T-cell epitopes to antibodies. The epitope-to-antibody ratio can be increased, but also a well-defined, genetically fused AEC can be generated that contains two different epitopes and/or protease cleavage sites. Increasing the viral T-cell epitope payload of AECs could increase the number of epitopes being delivered and the chance to encounter their respective virus-specific T cells within the tumor microenvironment (TME). In addition, increasing the chance of redirecting virus-specific T cells toward the tumor would increase the chance of triggering a tumor-specific T-cell response in the long term by the induced spread of tumor antigens.^16–18

For the design of these AECs, a protease cleavage site was used as a release mechanism for the T-cell epitopes when the antibody binds to the tumor cells. The Cl-1-YVL epitope combination used in this study has previously undergone testing across multiple cancer cell lines, demonstrating broad applicability and high efficiency when attached to the C-terminus of the HC.^6,12 We previously demonstrated that proteolytic activity and the expression of HLA-A2 by the targeted cells were detrimental for successful delivery and recognition of the target cells by the T cells. Replacing the amino acids (AA) of the protease cleavage site with D-AA, rendering it unrecognizable by proteases, led to a complete reduction in T-cell activation and target cell killing.^6,12 Therefore, the observed differences in loading efficiency on the different cell lines could be explained by the variability in protease, antibody target, and HLA class I expression levels.^2,5,6,12

Furthermore, protease-cleavage sites have also been used as a release mechanism for multiple prodrugs,¹⁹ antibody-drug conjugates,²⁰ and probodies.^15,21,22 Collectively, these therapeutic strategies have been shown to be effective, can improve safety, and that proteolytic release can take place at the cell surface and/or within the endo-lysosomal pathway of cancer cells. However, not much is known about extracellular processing of HLA class-I epitopes within the TME, in contrast to the conventional and well-studied intracellular antigen processing and presentation pathway of HLA class-I epitopes.^23,24

As the mechanism of epitope release of AECs is not yet fully understood, we decided to explore alternative designs, such as placing an epitope at the N-terminus instead of the C-terminus of tumor-targeting antibodies.⁴ We demonstrate here not only the feasibility of this approach, but also that the location of the epitope, and therefore the distance and/or the orientation toward the cell surface, affects the efficiency of the AECs in general. With the increased efficacy for proximal versus a more distal delivery of the EBV epitope (while keeping the payload delivery to the cell constant) delivered by two different antibodies, we hypothesize that when the epitope is released closer to the cell surface, the chance of encounter with and binding to the HLA-molecules increases. Although we have demonstrated that the efficacy of epitope delivery is not contingent on the target antigen, it can be influenced by factors such as the viral epitope used, the HLA class I molecule, and the protease cleavage site. To broaden and generalize these findings, further screening of additional protease cleavage site-viral epitope combinations will be necessary.

The data presented here also align with previously published data supporting a mechanism that relies on extracellular release of the EBV epitope mediated by proteolytic cleavage close to the cell membrane.^2–5 Moreover, it explains why the T-cell epitopes do not spread to nearby tumor cells lacking the antibody target as previously demonstrated.³ Taken together, our findings shed new light on the mechanism of action of AECs and provide further insights that can be expanded to other antibody-based therapeutic modalities that aim to deliver payloads in the TME via a mechanism that involves proteolytic cleavage. Additionally, it would be intriguing to explore whether the concept of AECs could be further expanded by also fusing an epitope to the N-terminus of the HC.

In conclusion, we have demonstrated that the AEC concept can be broadened by fusing T-cell epitopes to the N-terminus of the LC of tumor-specific antibodies. These AECs-NL demonstrate improved efficacy compared to AECs-CH in sensitizing tumor cells for attack by virus-specific T cells through peptide delivery in closer proximity to the cell surface. Taken together, the enhanced efficiency of AECs-NL and the ability to broaden the spectrum of T-cell epitope coupling options provides new opportunities for developing more effective therapeutic AEC modalities.

Ethical approval

Healthy donor material from the Leiden University Medical Center Biobank for Hematological Diseases was used in this study. This research was approved by the Institutional Review Board of the Leiden University Medical Center (approval number B16.039). Materials were obtained after written informed consent in accordance with the Declaration of Helsinki.

Acknowledgments

The authors would like to thank Ewald van den Bremer for his expertise, input, and help with the MS analysis of the generated bispecific antibodies, and Renoud Marijnissen and Rick Hibbert for their guidance, input, and support.

Disclosure statement

BB, JS, LG, and WL are (former) employees at Genmab BV and have ownership interests (including stocks, warrants, patents, etc.).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2024.2329321

Additional information

Funding

This work was performed under a grant provided by Genmab.