HEK293A, A549 and DLD-1 (ATCC) were maintained in Dulbecco’s modified Eagle medium (DMEM, Sigma-Aldrich, UK) supplemented with 10% (v/v) heat-inactivated foetal bovine serum (FBS, Gibco, UK). Transfections were performed in Opti-MEM (Gibco, UK), whilst virus infections were carried out using DMEM with 2% (v/v) FBS. Primary cells were maintained in X-VIVO 10 (Lonza, UK) with 1% (v/v) heat-inactivated human serum (HS, Sigma-Aldrich, UK). Cells were grown at 37 °C, 5% CO₂ and 95% humidity.

Human peripheral blood from anonymised healthy donors was obtained from the NHS Blood and Transfusion Service (Oxford, UK). PBS-diluted blood was overlaid onto Ficoll-Paque Plus (GE Healthcare, UK), then centrifuged (950 g, 30 min, RT). PBMCs were collected and washed with PBS, then resuspended in RPMI-1640 medium supplemented with 10% FBS. Cells were overlaid onto Percoll PLUS (46% in RPMI-1640, 10% FBS, 285 mOsm; GE Healthcare, UK) and centrifuged as before. The monocyte and lymphocyte fractions (the interphase and pellet, respectively) were collected and washed with PBS.

Monocytes were differentiated into macrophages through 6 days’ culture in medium containing 1% HS. Where specified, day-4 MDMs were polarised for 48 h using IL-4 (25 ng/mL, Miltenyi Biotec, UK, #130–095-373), IL-6 (25 ng/mL, Miltenyi Biotec, UK, #130–095-365) or IFN-γ (25 ng/mL, Miltenyi Biotec, UK, #130–096-873) and LPS (10 ng/mL, Sigma-Aldrich, UK). To generate FRβ^high MDMs, HS was omitted and monocytes were instead differentiated with recombinant M-CSF (50 ng/mL, Miltenyi Biotec, UK, #130–096-491). Where specified, monocytes were differentiated in the presence of ascites fluid (50% v/v).

Ascites samples were acquired with informed consent from routine drainage of cancer patients at the Churchill Hospital, Oxford, UK. Ascites were centrifuged (400 g for 10 min at room temperature) to separate the cellular and fluid components. The fluid was stored at − 20 °C until required. The cellular fraction was treated with red blood cell lysis buffer (Qiagen, UK, #158904) and cryopreserved until further use. For characterisation, cells were stained with Live/Dead Fixable Near IR stain (Invitrogen, UK, #L10119) and antibodies targeting CD4 (OKT4, Biolegend, UK, #317416), CD8 (HIT8a, Biolegend, UK, #300912), EpCAM (9C4, Biolegend, UK, #324206), FAP (427,819, R&D Systems UK, #MAB3715), PD-L1 (MIH3, Biolegend, UK, #374512), CD11b (ICRF44, Biolegend, UK, #301310), CD206 (15–2, Biolegend, UK, #321106) and folate receptor β (94b/FOLR2, Biolegend, UK, #391704), then analysed by flow cytometry using an Attune™ NxT Flow Cytometer (Thermo Fisher, UK).

BiTEs were generated by joining, with a glycine-serine linker, a single chain variable fragment (scFv) specific for CD3ε (L2K, patent #WO2004/106380) to a CD206-targeting nanobody (NbhMMRm3.1, patent #WO2014/140376Al) or a folate receptor β-targeting scFv (m923, patent #US2016/0207999A1). TriTEs were engineered similarly, with an anti-CD28 scFv (Clone 9.3, publically available on the ENA database, #AJ507107.1) or a second anti-CD3 scFv (L2K, patent #WO2004/106380) added with a glycine-serine linker to the N-terminus of the parental BiTE. Control BiTEs/TriTEs contained antibody fragments targeting irrelevant antigens. BiTEs/TriTEs contained an immunoglobulin signal peptide at the N-terminus for mammalian secretion and a deca-histidine (His) tag at the C-terminus for detection/ quantification. Using HiFi Master Mix (NEB, UK) to perform Gibson assembly [28], DNA fragments were inserted into an expression vector (pSF-CMV-Amp, Oxford Genetics Ltd., UK), under the control of a CMV promoter. Transgene insertion was confirmed by restriction digest and Sanger sequencing (Eurofins Genomics, Germany).

BiTE/TriTE-containing supernatants were produced through transfection of HEK293A using Lipofectamine 2000 (Invitrogen, UK) at a DNA:Lipofectamine ratio of 1:3 (w/v). Supernatants were harvested 48 h post-transfection and centrifuged to remove cellular components (400 g, 10 min, RT), then concentrated with Amicon Ultra-15 Centrifugal Filter Units (Merck, UK). Concentrated BiTE/TriTE-containing supernatants were aliquoted and stored at − 80 °C.

BiTEs/TriTE-encoding transgenes were inserted into a parental EnAd plasmid (EnAd2.4) using HiFi Master Mix (NEB, UK). Transgene cassettes contained a CMV promoter to drive BiTE/TriTE expression and a 3′ polyadenylation sequence. Successful transgene insertion was confirmed by restriction digest/Sanger sequencing (Eurofins Genomics, Germany). EnAd-CMV-BiTE/TriTE constructs were linearised with AscI (NEB, UK) and transfected into HEK293A cells using Lipofectamine 2000 (Invitrogen, UK). Cells and supernatant were harvested upon observation of cytopathic effect. Single virus clones were isolated by plaque-purification, then amplified and purified by double caesium chloride banding [29]. Virus stocks were titred by PicoGreen (Life Technologies, UK), yielding estimates of virus particles (vp)/mL (Additional file 13). For quality control, all viruses were analysed by anion exchange HPLC (Shimadzu Prominence, Japan), using GMP-grade EnAd (provided by PsiOxus Therapeutics, UK) to generate a standard curve.

MDM killing was assessed using Celigo-based image cytometry (Nexcelom Bioscience, USA). Day-6 MDMs were harvested, stained with carboxyfluorescein succinimidyl ester (CFSE, Invitrogen, UK, #C34554) and seeded at 25,000 cells/well into 96-well plates. The next day, MDMs were treated with BiTE/TriTEs, in the presence/absence of autologous lymphocytes (E:T ratio of 10:1, unless otherwise specified). In some experiments, co-cultures included 50% ascites fluid. Four days later, lymphocytes were removed and MDMs stained with propidium iodide (PI; diluted to 1 μg/mL in PBS, Sigma-Aldrich, UK, #P4864), and imaged on a Celigo image cytometer. Live MDMs were identified as being both CFSE-positive and PI-negative. % Live cells was calculated as follows: %Live cells=CFSE+PI−counttestCFSE+PI−countmock×100%\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \% Live\ cells=\frac{\ {CFSE}^{+}{PI}^{-}\ count\ (test)}{CFSE^{+}{PI}^{-}\ count\ (mock)}\times 100\% $$\end{document}

T cell activation was determined by flow cytometric analysis of CD25 expression. Lymphocytes were incubated in 96-well plates with/without autologous target MDMs (E:T ratio of 10:1, unless otherwise specified), and treated with BiTEs/TriTEs. Some experiments were conducted in 50% ascites fluid. After 4 days’ co-culture, lymphocytes were harvested and stained with anti-CD4 (OKT4, Biolegend, UK, #317416), −CD8 (HIT8a, Biolegend, UK, #300912), −CD25 (BC96, Biolegend, UK, #302606), −CD69 (FN50, Biolegend, UK, # 310904), −CD107a (H4A3, Biolegend, UK, #328608) and -HLA-DR (L243, Biolegend, UK, #307604) antibodies.

Unpurified ascites cells were seeded at 200,000 cells/well in 100 μL medium into flat-bottom low-adherence 96-well plates. After resting overnight, cells were treated with BiTEs/TriTEs or viruses, diluted in 100 μL medium or autologous ascites fluid. Five days later, cells were harvested and processed for flow cytometry, using Live/Dead Fixable Near IR stain (Invitrogen, UK, #L10119) and anti-CD11b (ICRF44, Biolegend, UK, #301310), −CD64 (10.1, Biolegend, UK, #305006), −CD86 (IT2.2, Biolegend, UK, #305422), −CD80 (2D10, Biolegend, UK, #305208), −CD4 (OKT4, Biolegend, UK, #317408), −CD8 (HIT8a, Biolegend, UK, #300912) and -CD25 antibodies (BC96, Biolegend, UK, #302606).

ELISAs were performed with commercially-available kits measuring IL-6 (Biolegend, UK, #30504), IL-10 (Biolegend, UK, #430604), IFN-γ (Biolegend, UK, #430104), TGF beta-1 Human/Mouse ELISA Kit (Invitrogen, UK, #88–8350-88) and CD206 (RayBiotech, USA, #ELH-MMR-1), following manufacturers’ instructions. Prior to analysis, samples were centrifuged (400 g, 10 min, RT) to remove cellular components and diluted two-fold (IL-6 and IL-10), five-fold (TGF-β), ten-fold (IFN-γ) or thirty-fold (CD206) in PBS.

Cytokines and chemokines in ascites cell supernatants were quantified using the LEGENDplex Human Macrophage/Microglia panel kit (Biolegend, UK, #740526) and flow cytometric analysis, according to manufacturers’ instructions.

BiTE/TriTEs were detected by immunoblotting with anti-C-terminal-His antibody (3D5, Invitrogen, UK, #R930–25). For western blotting analysis, supernatants were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane. For dot blot analysis (Additional file 1), two-fold serial dilutions of the supernatants were applied directly to a nitrocellulose membrane. To generate a standard curve, a deca-His-tagged standard protein of known concentration was serially-diluted and applied in parallel. Membranes were probed with an anti-C-terminal-His tag (1:5000, clone 3D5, Invitrogen, UK, #46–069) primary antibody, then a horseradish peroxidase (HRP)-conjugated anti-mouse secondary antibody (1:3000, Cell Signalling Technology, UK, #7076). SuperSignal West Dura Extended Duration Substrate (Thermo Fisher, UK, #34075) was applied and the membrane exposed to X-ray film, which was developed in an automatic film processor (Agfa CP1000).

Statistical analysis was performed using one- or two-way ANOVA tests, with Dunnett or Bonferroni post hoc analysis, respectively. All data are presented as mean ± SD. The significance levels used were P = 0.01–0.05 (*), 0.001–0.01 (**), 0.0001–0.001 (***). Experiments were performed in biological triplicate, unless otherwise stated.

We first determined the expression of two M2-like macrophage markers, CD206 and FRβ, in clinically-relevant models of human TAMs. Primary ascites cells from five cancer patients were characterised by flow cytometry (Fig. 1a). Ascites-associated macrophages (identified as being CD11b⁺CD64⁺) expressed CD206 (4/5 patients) and FRβ (5/5 patients) at higher levels than M1-polarised monocyte-derived macrophages (MDMs; derived from healthy peripheral blood mononuclear cells (PBMCs)) (Fig. 1b and c). In an alternative approach, we cultured PBMC-derived human monocytes with cell-free malignant ascites fluid, which reportedly generates MDMs that resemble human TAMs [30]. Using ascites fluid from 11 cancer patients (Additional file 15), we observed significant up-regulation of CD206 (11/11 patients) and FRβ (6/11 patients) on human PBMC-derived MDMs, as compared to M1-polarised MDMs (Fig. 1d).Fig1Fig. 1

CD206 and folate receptor (FR) β are markers of human tumour-associated macrophages. a Representative flow cytometry plots showing CD206 and FRβ expression on primary ascites cells after gating for CD11b⁺/CD64⁺ double positivity. b, c, Percentage positivity for (b) and expression levels of (c) CD206 (PE/Cy7) and FRβ (PE) on primary CD11b⁺/CD64⁺ ascites cells from five different cancer patients, and on monocyte-derived macrophages (from healthy donors) polarised with 10 ng/mL LPS and 25 ng/mL IFN-γ (“M(IFN-γ/LPS)”), as determined by flow cytometric analysis. d Primary human monocytes from healthy donors were differentiated into macrophages through 6 days’ culture in medium containing 1% human serum. Where indicated, culture medium contained 10 ng/mL LPS and 25 ng/mL IFN-γ (“M(IFN-γ/LPS)”, added on Day 4), or 50% acellular ascitic supernatant from 11 different patients (“Patient 1″-“Patient 11″, added on Day 0). Expression levels of CD206 (PE) and FRβ (PE) were determined by flow cytometry. c, d Data show mean ± SD of biological triplicates. Statistical analysis was performed by one-way ANOVA with Dunnett’s post-hoc analysis compared with “M(IFN-γ/LPS)” (*, P < 0.05; **, P < 0.01; ***, P < 0.001)

Human TAM-targeting BiTEs were engineered by joining, with a glycine-serine (GGGGS) linker, a single chain variable fragment (scFv) specific for CD3ε to domains specific for CD206 and FRβ (Fig. 2a). The CD206-binding domain was a nanobody, whilst the FRβ-binding domain was a scFv. Antibody fragments were selected based on their similarly high affinities (K_D of 3.4 nM and 2.48 nM for CD206-binding and FRβ-binding fragments, respectively [31], patent #WO2014/140376Al). Control (Ctrl) BiTEs, with the same CD3ε-binding domain and a nanobody or scFv recognising irrelevant antigens (rabies virus protein for the CD206 BiTE and filamentous hemagglutinin adhesin of Bordetella pertussis for the FRβ BiTE, respectively), were also generated. BiTEs contained a signal peptide for secretion and a deca-histidine tag for detection. BiTE constructs were cloned into expression vectors under the control of the cytomegalovirus immediate early (CMV) promoter. All BiTEs were expressed and secreted following transfection of HEK293A cells (Fig. 2b).Fig2Fig. 2

CD206- and FRβ-targeting BiTEs activate primary human T cells to kill autologous M2-polarised macrophages. a Schematic representations of CD206- and FRβ-targeting BiTEs. b, Western blot analysis of supernatants from HEK293A cells 48 h after transfection with BiTE expression plasmids. Blots were probed with a mouse anti-His primary antibody, followed by an HRP-conjugated anti-mouse secondary antibody. c Human MDMs were polarised as indicated, stained with CFSE, and treated with T cells (10:1 E:T ratio) and increasing concentrations of BiTEs. Macrophage killing was assessed 96 h later by propidium iodide staining and Celigo image cytometry. d MDMs were stained with CFSE and treated with the indicated concentrations of BiTE in the presence or absence of T cells (10:1 E:T ratio). 96 h later, cytotoxicity was assessed by propidium iodide staining and analysis with a Celigo image cytometer. e T cell activation in the presence or absence of target cells was assessed by flow cytometric measurement of CD25 expression 96 h after BiTE addition. Data show mean ± SD of biological triplicates (c, d and e). Statistical analysis was performed by two-way ANOVA with Bonferroni post-hoc tests comparing with the relevant “Mock” condition (d and e) (*, P < 0.05; **, P < 0.01; ***, P < 0.001)

We next asked whether the TAM-targeting BiTEs would retain their activity in acellular malignant ascites, which is rich in soluble immunoregulatory factors [33]. Using human MDMs and autologous lymphocytes from healthy peripheral blood, we performed BiTE cytotoxicity assays in the presence of ascites fluid (50% v/v) from three cancer patients (Fig. 3a and b). FRβ BiTE activity was largely unaffected, triggering robust T cell activation and cytotoxicity (Fig. 3a and b). The efficacy of the CD206 BiTE, however, was greatly diminished, with little or no T cell activity observed in ascites fluid (Fig. 3a and b). Elevated levels of three prominent immunomodulatory factors, IL-6, IL-10 and TGF-β, were observed in all ascites samples (Fig. 3c), relative to pooled healthy human serum. Furthermore, soluble CD206, which may block BiTE binding to membrane-bound CD206, was detected at high levels in most ascites fluids (Fig. 3d). Interestingly, the ascites sample with the greatest inhibitory effect (“Patient 1”) contained high levels of IL-10, TGF-β and soluble CD206, perhaps indicating important roles for these factors in limiting CD206 BiTE activity.Fig3Fig. 3

Human malignant ascites supernatant suppresses CD206 BiTE activity but not FRβ BiTE activity. a CFSE-stained MDMs were co-cultured with T cells (10:1 E:T ratio) and the indicated BiTEs, in the presence or absence of 50% ascitic supernatant from three patients (Patients 1, 2 and 5). 96 h later, percentage live MDMs was determined with propidium iodide staining and a Celigo image cytometer. b T cell activation was assessed by flow cytometric analysis of CD25 expression after 96 h of co-culture with MDMs and the indicated BiTEs, in the presence or absence of 50% ascitic supernatant from three patients (Patients 1, 2 and 5). c Quantities of IL-6, IL-10 and total (active and latent) TGF-β in malignant ascites fluid from six different patients, as determined by enzyme-linked immunosorbent assay. Normal serum (NS) pooled from three healthy donors was included as a control. d Quantities of soluble CD206 in malignant ascites fluid from nine different patients was determined by enzyme-linked immunosorbent assay. Pooled NS was used as a control. Each condition was measured in biological triplicate and represented as mean ± SD (a-d). Statistical significance was assessed by one-way ANOVA with Dunnett’s post-hoc analysis compared with “Pooled NS” (c, d), or two-way ANOVA followed by Bonferroni post-hoc analysis, with each treatment being compared to the “Mock” condition within the relevant group (a and b). (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, non-significant)

T cell activation is determined by a balance between stimulatory and inhibitory signals. We asked whether adding a second T cell-activating domain to the N-terminus of the parental CD206 BiTE would push the balance towards T cell activation in ascites fluid. Two trivalent T cell engagers (TriTEs) were engineered; one contained a second anti-CD3-binding domain (referred to as “3–206-3”), whilst the other contained an anti-CD28 binding domain, providing T cell co-stimulation (“28–206-3”) (Fig. 4a). Matched Ctrl TriTEs were generated in parallel (“3-Ctrl-3”, “28-Ctrl-3”) (Fig. 4a). All four TriTEs were secreted by transfected HEK293A cells (Fig. 4b). CD206 binding was not compromised by the additional T cell-binding domains (Fig. 4c).Fig4Fig. 4

Addition of a second T cell-binding domain to the parental CD206 BiTE. a Schematic representations of CD206-targeting TriTEs. b Western blotting analysis of supernatants from HEK293A cells transfected with TriTE expression plasmids 48 h previously. Blots were probed with a mouse anti-His primary antibody and an HRP-conjugated anti-mouse secondary antibody. c BiTE and TriTE binding to recombinant CD206 protein, as determined by ELISA using a mouse anti-His primary antibody and an HRP-conjugated anti-mouse secondary antibody. d-f T cells were cultured for 96 h with the indicated BiTEs/TriTEs at a dose of 50 nM in the presence or absence of target cells. T cell activation was assessed by measuring CD25 expression by flow cytometry, with representative histograms displayed in (d) and geometric MFI values displayed in (e). IFN-γ levels in the supernatants were quantified by ELISA (f). Data show mean ± SD of biological triplicates (c, e and f). Statistical significance was assessed by two-way ANOVA followed by Bonferroni post-hoc analysis, with each treatment being compared to the “Mock” condition within the relevant group (e and f) (*, P < 0.05; **, P < 0.01; ***, P < 0.001)

Given the lack of antigen dependency of the 28–206-3 TriTE, we proceeded only with the 3–206-3 TriTE (hereafter termed “CD206 TriTE”). We next determined its selectivity for M2 (IL-4-polarised) over M1 (IFN-γ/LPS-polarised) MDMs (Additional file 2). A checkerboard approach was employed, with MDMs subjected to increasing BiTE/TriTE concentrations in the presence of increasing numbers of T cells (i.e. greater effector:target (E:T) ratios). Cytotoxicity towards M2-polarised macrophages was observed even at E:T ratios (< 2:1) and BiTE/TriTE concentrations (< 10 nM) (Fig. 5a). Conversely, M1-polarised MDMs were killed only when E:T ratio and TriTE concentration were simultaneously high (E:T ratio of 10:1 and TriTE concentration of 50 nM, Fig. 5a). Similarly, the Ctrl TriTE only elicited macrophage cytotoxicity at high E:T and TriTE concentrations (Fig. 5a). Notably, the CD206 TriTE induced cytotoxicity of M2-polarised MDMs at lower concentrations and more physiologically-relevant E:T ratios than the CD206 BiTE (Fig. 5a). For instance, at a concentration of 2 nM and an E:T ratio of 2:1, the CD206 TriTE triggered a marked decrease in % live macrophages to 10.6%, whilst the CD206 BiTE was completely ineffective (Fig. 5a; Additional file 5).Fig5Fig. 5

A CD206 TriTE with bivalent CD3 binding retains specificity for M2 macrophages and overcomes ascites suppression. a Monocyte-derived macrophages were polarised as indicated, CFSE-stained, and co-cultured for 96 h with T cells at increasing E:T ratios and BiTE/TriTE concentrations. % Live cells were calculated with propidium iodide staining and Celigo image cytometry, with values displayed as a heat map. b T cells were co-cultured with monocyte-derived macrophages and BiTEs/TriTEs in the presence or absence of 50% ascites fluid from seven different patients (Patients 1, 2, 3, 4, 6, 7 and 8). 96 h later, CD25 expression was determined by flow cytometry. C, CFSE-stained monocyte-derived macrophages were treated with T cells (10:1 E:T ratio) and BiTEs/TriTEs in medium alone or 50% ascites supernatant from seven different patients (Patients 1, 2, 3, 4, 6, 7 and 8). 96 h later, cells were stained with proprodium iodide and analysed with a Celigo image cytometer to calculate % live cells. Data show mean ± SD of biological triplicates (b, c). Statistical significance was assessed by two-way ANOVA followed by Bonferroni post-hoc analysis, with each treatment being compared to the relevant “Mock” condition (b, c) (*, P < 0.05; **, P < 0.01; ***, P < 0.001)

We then explored whether the FRβ-targeting BiTE may also be improved upon by molecular engineering. Unexpectedly, we found that reversing the order of scFv domains along the single chain (to N-αCD3-αFRβ-C) increased efficacy, without compromising target cell selectivity (Additional file 6). The new BiTE (“3FR BiTE”) exhibited an EC₅₀ value of 10.63 nM – ≈6 times lower than that of the original BiTE (“FR3 BiTE”; EC₅₀ of 61.22 nM) (Additional file 6). Both FRβ BiTE orientations were compared in the remaining experiments.

Malignant ascites contains a mixture of tumour cells and cancer-associated fibroblasts, lymphocytes, myeloid-derived suppressor cells and M2-like macrophages, making it a valuable tumour-like model for studying BiTE efficacy. Ascites cells from five patients (characterisation in Additional files 7 and 14) were treated with the TAM-targeting T cell engagers in the presence or absence of autologous ascites fluid. The FRβ-targeting BiTEs triggered a marked depletion of ascites macrophages; in the presence of ascites fluid, % residual CD11b⁺CD64⁺ cells decreased to an average of 37.9 and 26.4% for the FR3 and 3FR BiTEs, respectively (Fig. 6a). CD206 TriTE treatment, however, induced killing of CD11b⁺CD64⁺ cells only in the absence of ascites fluids (Fig. 6a); furthermore, this activity was not entirely dependent on the target antigen, as similar effects were observed with the Ctrl TriTE (Fig. 6a). Due to the non-selective nature of the CD206 TriTE in this model, subsequent efforts were directed towards evaluating the FRβ-targeting BiTEs.Fig6Fig. 6

CD206- and FRβ-targeting T cell engagers activate endogenous ascites T cells to kill ascites macrophages. a-e Total unpurified ascites cells from five different patients were cultured for five days with 50 nM BiTEs/TriTEs in medium only or 50% ascites supernatant from the same patient sample. a, b Cells were stained with anti-CD11b, anti-CD64, anti-CD80 and anti-CD86 antibodies, as well as a LIVE/DEAD fixable stain, then analysed by flow cytometry. a % Live residual CD11b⁺CD64⁺ cells were calculated relative to “Mock”-treated samples. b Fold-changes in geometric MFI values of CD64, CD80 and CD86 on live CD11b⁺CD64⁺ ascites cells were calculated relative to “Mock”-treated samples for each patient sample. c Activation of endogenous CD4⁺ and CD8⁺ ascites T cells was assessed by flow cytometric measurement of CD25 expression. d IFN-γ levels in the culture supernatants were determined by ELISA. e Numbers of CD4⁺ and CD8⁺ cells were determined through addition of counting beads to samples immediately prior to antibody staining. Fold-changes in CD4⁺ and CD8⁺ cell count were calculated relative to “Mock”-treated samples. Data show the grand mean ± SD of five individual patient means (calculated from biological triplicate). Statistical significance was assessed by two-way ANOVA followed by Bonferroni post-hoc analysis, with each treatment being compared to the relevant “Mock” condition (a, c-e) (*, P < 0.05; **, P < 0.01; ***, P < 0.001)

EnAd, a chimeric group B oncolytic adenovirus in Phase I/II clinical trials, demonstrates cancer cell-restricted replication and favourable pharmacokinetics following the systemic delivery of three repeated doses [34–36]. EnAd can be engineered to encode biologics that are expressed and secreted by infected tumour cells as the virus replicates, offering a multipronged and tumour-targeted therapeutic strategy [20, 22, 37]. To explore the feasibility of this approach in the context of TAM-targeting T cell engagers, we inserted the FRβ and Ctrl BiTE sequences (in both orientations) downstream of the fibre gene under the control of CMV promoter (Additional file 9). All four BiTE-armed viruses demonstrated comparable oncolytic activity to parental EnAd (Fig. 7a), whilst mediating BiTE secretion into the cell supernatants (Fig. 7b). Indeed, incubation of infected cell supernatants with co-cultures of human lymphocytes and autologous MDMs triggered robust T cell activation and macrophage cytotoxicity (Additional file 10). Supernatants from cells infected with 3FR-BiTE armed EnAd (EnAd-3FR) were particularly effective, achieving significant MDM killing at dilutions as low as 1:1000 (Additional file 10).Fig7Fig. 7

EnAd expressing FRβ-targeting BiTEs activates endogenous ascites T cells to kill ascites macrophages. a DLD-1 cells were infected with parental or armed EnAd, and viability assessed four days later by MTT assay. % Live cells was calculated relative to “Mock”-treated cells. b DLD-1 cells were infected with armed EnAd at 100 vp/cell. Supernatants were harvested 72 h later and analysed for BiTE expression by western blotting using anti-His primary antibody and an HRP-conjugated anti-mouse secondary antibody. c-g Total unpurified ascites cells from five patients were infected with 100 vp/cell parental or BiTE-expressing EnAd for five days with or without autologous fluid. c Activation of endogenous CD4⁺ and CD8⁺ ascites T cells was assessed by flow cytometric measurement of CD25 expression. d CD4⁺ and CD8⁺ cell numbers were determined by adding counting beads to samples immediately prior to antibody staining. Fold-changes in CD4⁺ and CD8⁺ cell count were calculated relative to “Mock”-treated samples. e, g Cells were stained with anti-CD11b, anti-CD64, anti-CD80 and anti-CD86 antibodies, and a LIVE/DEAD fixable stain, then analysed by flow cytometry. e % Live residual CD11b⁺CD64⁺ cells were calculated relative to “Mock”-treated samples. f IFN-γ levels in the supernatants were determined by ELISA. g Fold-changes in geometric MFI values of CD64, CD80 and CD86 on live CD11b⁺CD64⁺ ascites cells were calculated relative to “Mock”-treated samples for each patient sample. c-f Data show the grand mean ± SD of five individual patient means (calculated from biological triplicate). c-f Statistical significance was assessed by two-way ANOVA followed by Bonferroni post-hoc analysis, with each treatment being compared to the relevant “Mock” condition (c-f) (*, P < 0.05; **, P < 0.01; ***, P < 0.001). (c-g) (*, P < 0.05; **, P < 0.01; ***, P < 0.001)