Introduction

Current primary care for patients with colorectal carcinomas (CRC) typically involves surgery in combination with chemo- and/or radiotherapy. Unfortunately, for many patients these treatments are not curative, underscoring the need for improved or novel therapeutic strategies. In recent years, various immunotherapeutic strategies including checkpoint inhibition and adoptive transfer of CAR T cells, have become valuable additions to anti-cancer treatment. Additionally, monoclonal antibodies (mAbs) targeting the anti-epidermal growth factor receptor (EGFR) are used to treat patients with advanced colorectal and head and neck carcinomas1,2,3. While anti-EGFR mAbs are primarily recognized for their direct effect of blocking binding of EGF to its receptor, they also exert indirect effects, including complement-dependent cytotoxicity (CDC) and the recruitment of immune effector cells, which are consistent with their role as immunotherapeutic agents4.

Most clinical mAbs are of the IgG isotype, which benefits from a long serum half-life and the ability to activate the classical complement pathway through C1q binding, leading to CDC5. Additionally, IgG mAbs can bridge tumor cells to immune cells that express Fcɣ receptors (FcɣRs), resulting in tumor cell killing by macrophages and natural killer (NK) cells6,7. However, IgG mAbs are less effective at recruiting neutrophils as effector cells in cancer, which constitute up to 70% of all circulating immune cells and possess potent cytotoxic ability. Neutrophils express high levels of FcγRIIIb, which lacks intracellular signaling capability and presumably acts as decoy receptor, limiting neutrophil activation in IgG-based cancer therapies8,9.

Furthermore, neutrophils express an Fc receptor for IgA, Fcα receptor I (FcαRI, CD89). Crosslinking of FcαRI induces stronger intracellular signaling and increased pro-inflammatory effector functions compared to FcγR crosslinking8,9,10. The neutrophil arsenal of immune defense mechanisms includes phagocytosis, degranulation, and production of reactive oxygen species (ROS) production11,12. Neutrophil-derived ROS, particularly hydrogen peroxide (H2O2), has been shown to mediate killing of metastatic breast cancer cells in the pre-metastatic lung, thereby impeding metastatic outgrowth13,14. Additionally, neutrophils secrete fibrous web-like chromatin structures, referred to as neutrophil extracellular traps (NETs), which contribute to trapping pathogens15.

Additionally, trogocytosis, a mechanical disruption of the tumor cell plasma membrane, was identified as a prominent mechanism through which neutrophils kill tumor cells16,17.

Even though tumor killing is increased in the presence of IgA anti-tumor mAbs compared to IgG mAbs, the shorter serum half-life challenges clinical development10,18. To address this, we previously developed IgG bi-specific antibodies (BsAbs) that target both FcαRI and a tumor associated antigen10. Neutrophils effectively killed tumor cells in the presence of FcαRI BsAb in vitro, and tumor growth in preclinical models was decreased. However, tumors were not completely eradicated, which may be due to an immunosuppressive tumor microenvironment (TME) that hampers effective immunotherapy.

The TME contains various immunosuppressive cell populations, including regulatory T cells, myeloid-derived suppressor cells (MDSCs), tumor-associated macrophages (TAMs) and tumor-associated neutrophils (TANs)19,20,21. MDSCs suppress anti-tumor immunity by various mechanisms, such as amino acid depletion immune checkpoint expression and secretion of anti-inflammatory cytokines22,23. Depletion of MDSCs inhibited tumor growth and metastasis24,25. Macrophages and neutrophil often display dual roles. Pro-inflammatory (M1-like) macrophages and classical neutrophils secrete pro-inflammatory cytokines and exert anti-tumor effects, whereas TAMs and TANs often adopt an anti-inflammatory phenotype in the TME, marked by the secretion of anti-inflammatory cytokines, such as TGF-β and IL-1010,14,26,27,28. These phenotypes are commonly referred to as respectively M1/N1-like or M2/N2-like macrophages and neutrophils, although this is an oversimplification due to many subvariants.

Introducing pro-inflammatory cytokines like TNF-α, into the TME may help overcome immunosuppression. TNF-α sensitizes macrophages to further stimulation, and in combination with LPS or IFN-γ, can promote M1-like polarization29,30,31. Exposure to TNF-α induced a metabolic switch to aerobic glycolysis in NK cells and enhanced their IFN-γ production upon IL-18 stimulation32. Knockout of the TNF Receptor 2 (TNFR2) in NK cells impaired glycolysis and proliferation, and led to lower numbers of IFN-γ-producing NK cells33. TNF-α upregulates the surface expression of FcαRI on neutrophils, and modulates their survival by altering apoptotic gene expression34. As such, TNF-α may potentially enhance FcαRI BsAb-mediated anti-tumor immunity. Systemic use of TNF-α is, however, limited by toxicity at effective doses35,36. It has been demonstrated that TNF-α can be targeted by fusing it to an anti-tumor antibody, also referred to as an immunocytokine. Currently, systemic treatment with TNF-α-immunocytokines targeting the EDB domain of fibronectin are tested in Phase I clinical studies in patients with advanced solid tumors and soft tissue sarcoma (NCT02076620, NCT04650984)37.

In this study we investigated whether fusing TNF-α to an FcαRI BsAb increased anti-tumor responses by cytotoxic effectors cells.

Methods

Animals

Human FcαRI-transgenic, LysM-EGFP positive mice were generated previously10. Mice were heterozygous for human FcαRI and homozygous for LysM-EGFP. LysM-EGFP positive, FcαRI-negative littermates were used as controls. Mice were bred and maintained at the Central Animal Facility of the Amsterdam University Medical Center location VUmc (Amsterdam, the Netherlands). Regular chow and water was provided ad libitum. For all experiments, age-matched (8–12 weeks old) male or female mice were used. Experiments were performed according to institutional and national guidelines. The animal ethical committee of the VU University Medical Center approved all experiments (AVD1140020173844). The findings and experimental setup of experiments using experimental animals have been reported in this manuscript in accordance with the ARRIVE 2.0 guidelines.

Cell culture

A431, HCT116 and HT29 tumor cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, 11965092) supplemented with 10% fetal calf serum (FCS, Biowest, 35-079-CV), 2mM L-glutamine (Gibco, A2916801) and 1% Penicillin-Streptomycin (Lonza Bioscience, DE17-602E), referred to as Complete DMEM.

Murine EGFR was mutated at six sites to allow binding of the anti-human EGFR mAb cetuximab (the anti-human EGFR binding arm of FcαRI BsAb, creating chimeric EGFR (cEGFR). MC38-cEGFR cells were generated through the lentiviral transduction of the MC38 cell line (RRID: CVCL_B288) with cEGFR. Third generation lentiviral particles were produced by HEK293T cells (RRID: CVCL_0063) which had been transfected with VSV, GAG/POL and REV expressing vectors, as well as the pHAGE-cEGFR transfer vector using Trans IT (Mirus, MIR2300). The pHAGE-cEGFR plasmid was created using pHAGE-EGFR, which was a gift from Gordon Mills & Kenneth Scott (Addgene plasmid #116731; http://n2t.net/addgene:116731 ; RRID: Addgene_116731)38. Lentiviral particles were concentrated to 50 times the original volume using Lenti-X (Takara, 631231). Wildtype MC38 cells were treated overnight with lentiviral particles in a ratio of 1:15 by volume.

MC38 cells bearing cEGFR (MC38-cEGFR) were cultured in Complete DMEM supplemented with 5 µg/mL puromycin (Sigma-Aldrich, P8833) at humidified conditions, 37 °C and 5% CO2. High protein-yield FreeStyle HEK293 cells (ThermoFisher, R79007) were cultured in suspension with modified serum-free FreeStyle HEK293 expression medium (Gibco, 12338-018) supplemented with 1% Penicillin-Streptomycin. Neutrophils were isolated from healthy donor blood obtained from Sanquin after informed consent had been obtained (Amsterdam) using a Lymphoprep density gradient (Stemcell Technologies, 07851) according to the manufacturer’s instructions. Neutrophils were cultured in Roswell Park Memorial Institute (RPMI) culturing medium (Gibco, 31870-025) supplemented with 10% FCS, 2mM L-glutamine and 1% Penicillin-Streptomycin (referred to as complete RPMI) at humidified conditions, 37 °C and 5% CO2. Monocytes were isolated from PBMCs using CD14 microbeads (Miltenyi, 130-050-201) according to manufacturer’s instructions. Monocytes were cultured in complete RPMI supplemented with 50ng/mL recombinant human M-CSF (Miltenyi, 130-096-491) for 6 days to induce differentiation into macrophages. NK cells were isolated from PBMCs using a human NK cell isolation kit (Miltenyi, 130-092-657) according to manufacturer’s instructions. NK cells were cultured in complete RPMI.

Generation of bi-specific immunocytokine

Freestyle HEK293 cells were transfected with DNA constructs encoding for the heavy and light chains of either a Cetuximab-based anti-EGFR IgG1 or of an anti-FcαRI IgG1 using linear Polyethylenimine (Polysciences, 23966-1). The heavy chain of the anti-FcαRI IgG1 had a single chain murine TNF-α trimer fused to its C-terminus via flexible linkers (SSGGGGSGGGGS). To allow in vivo studies murine TNF-α was used, also because murine TNF-α has been reported to interact with human TNF receptors in a similar fashion as human TNF-α39,40. Concordantly, murine TNF-α (hereafter referred to as TNF-α) was also used in all in vitro assays. After 6 days, supernatants were collected. Parental antibodies were harvested and purified using a Hi-trap protein A column (Cytiva, 17–0402) and the ÄKTAprime system (Amersam Phamacia Biotech). Bi-specific immunocytokines were produced using Duobody technology as described by Labrijn et al.41. Briefly, 25mM 2-Mercaptoethylamine (Sigma, M9768-5G) was used to achieve controlled Fabarm exchange by mixing equimolar amounts of parental anti-EGFR (F405L mutated) and parental anti-FcαRI TNF fusion (K409R mutated) antibodies. After 4 h incubation at 37 °C, the reducing agent was exchanged for PBS using Zeba spin desalting columns (ThermoFisher Scientific, 89892). Antibodies were stored at 4 °C overnight to allow reoxidation of the disulfide bonds and the formation of bi-specific immunocytokines (FcαRI BsAb).

ADCP assay and macrophage phenotype markers

Macrophages were incubated overnight at 37 °C with 100ng/mL LPS (Sigma-Aldrich, L4391) and 1 µg/mL rhIFN-γ (Peprotech, 300-02) to induce skewing towards an M1-like phenotype. For skewing towards an M2-like phenotype, macrophages were incubated overnight at 37 °C with 20ng/mL rhIL-4 (Immunotools, 11340045). Macrophages (1,5 × 105) were incubated with 1,875 µg/mL DiO (ThermoFisher, D275) and plated in 24-well flat bottom plates 24 h prior to experiments. Medium was refreshed the next day. FcαRI BsAb, TNF-α (Peprotech, 315–01 A) or FcαRI-TNF were added in a concentration of 6,67 × 10− 9mmol/mL. Tumor cells were incubated with 5µM eFluor670, and added in an effector to target (E: T) ratio of 15:1. After 24 h incubation at 37 °C, supernatants were harvested and cells trypsinized. A cell scraper was used to remove cells from 24-well plates. Cells were transferred to 96-well V-bottom plates. After washing with 0,5% Bovine Serum Albumin (BSA) in 1X PBS (PBSA), cells were stained for 15 min at 4 °C with a viability dye (eBioscience, 65-0865-14), followed by incubation for 30 min at 4 °C with directly labeled antibodies: BV785 anti-CD40 (BioLegend, 334339), APC anti-CD80 (BioLegend, 305220), PE/Dazzle594 anti-CD200R (BioLegend, 329309) or FITC anti-CD206 (BioLegend, 321103). Following staining, cells were washed and fixed with 4% PFA for 10 min. Fluorescent signal was measured using the BD LSRFortessa Cell Analyzer.

Trogocytosis assay

Tumor cells (8,0 × 103) were stained with 1,875 µg/mL DiO and seeded in 96-well flat bottom plates 24 h prior to experiments. Medium was refreshed the next day. FcαRI BsAb, TNF-α or FcαRI-TNF were added in a concentration of 6,67 × 10− 10mmol/mL. Neutrophils were stained with 5µM eFluor670 (eBioscience, 65-0840-85) and added in an E: T ratio of 50:1. After 4 h incubation at 37 °C, supernatants and cells were harvested. For analysis with flow cytometry, cells were washed with 0,5% PBSA and stained with a viability dye for 15 min at 4 °C. Afterwards, they were incubated for 1 h at 4 °C with directly-labeled antibodies: BV605 anti-CD11b (BioLegend, 101257) and PE anti-CD66b (BioLegend, 305106). Following staining, cells were washed and fixed with 4% PFA for 10 min. Fluorescent signal was measured using the BD LSRFortessa Cell Analyzer.

Tumor killing assay

Tumor cells (8,0 × 103) were seeded in 96-well flat bottom plates 24 h prior to experiments. Medium was refreshed the next day. FcαRI BsAb, TNF-α or FcαRI-TNF were added in a concentration of 6,67 × 10− 10mmol/mL. DNAse (Merck, 11284932001) and sodium pyruvate (Gibco, 11360039) were added in concentrations of 40U/mL and 20mM respectively. Neutrophils or NK cells were added in an E: T ratio of 50:1 or 5:1 respectively. After incubation of 4 h for neutrophils and 24 h for NK cells at 37 °C, effector cells were washed away with PBS. Tumor cells were incubated with CellTiter Blue (Promega, G8081) for 1 h at 37 °C C to determine metabolic activity of remaining cells, which is indirectly taken as indication for tumor cell killing42. Fluorescent signal was measured using the BioTek Synergy HTX.

Supernatant cytokine ELISA

The levels of various cytokines were quantified in supernatants of ADCP, ADCC and trogocytosis assays using ELISA kits for hTNF-α (Invitrogen, 88-7346-88), hIL-12p70 (eBioscience, 14-7128-68), hTGF-β (Invitrogen, 58.168.09), hIL-10 (eBioscience, 14-7108-85), hIL-8 (Invitrogen, 88-8086-88), Lactoferrin (Sigma-Aldrich, L3262-1VL) and hIFN-γ (Invitrogen, KHC4021) according to manufacturer’s instructions. Absorbance was measured using the BioTek Synergy HTX.

SDS-PAGE

Antibody size was analyzed by SDS-PAGE on a reducing 12% Acrylamide gel (BioRad, 1610175). Antibodies (250 µg/mL) were diluted in 1x Tricine sample buffer (BioRad, 161–0739) and DTT (100mM). Afterwards, samples were denatured at 95 °C for 10 min, vortexed and shortly centrifuged. Samples were loaded and ran for 2,5 h at 70 V, after which gels were stained with InstantBlue Coomassie Protein Stain (Abcam, ab119211) and imaged using the Azur Biosystems c200.

Surface plasmon resonance

Surface plasmon resonance (SPR) measurements were performed on an IBIS MX96 (IBIS technologies) device as previously described by Dekkers et al.43.

To assess affinity for human EGFR (hEGFR), antibodies were spotted onto a single SensEye G Easy2Spot sensor (Ssens, 1-09-04-006) in three-fold dilutions using a Continuous Flow Microspotter, ranging from 30 nM to 1 nM, with 10 mM acetate buffer supplemented with 0,075% Tween-80 (Merck, P4780) pH 4,5 as activation buffer. The sensor was deactivated by flowing 100mM ethanol amine, pH 8,8 for 7 min. Soluble hEGFR was injected over the sensor in an 8 times dilution series, starting at 0,78nM until 100nM, in PBS + 0,075% Tween-80 pH 7,4 (PBST). After every hEGFR injection, regeneration was carried out with 100 mM H3PO4, pH1,7.

To determine affinities for the FcɣRs and FcαRI, biotinylated FcɣRs and biotinylated FcαRI were spotted onto a single SensEye G-streptavadin sensor. Receptors were spotted at concentrations of 30, 10, 3,3 and 1,1nM in PBST, except for FcɣRIIIa 158 V, which was spotted at 100, 33, 11 and 3,7nM. Antibodies were injected over the sensor in an 8 times dilutions series, starting at 7,8nM until 1000nM in PBST, and regenerated after every injection with 10mM Glycine-HCl, pH 2,0.

Calculation of the dissociation constant (KD) was performed by equilibrium fitting to Rmax= 300 for the hEGFR affinity and Rmax= 500 for the FcɣRs and FcαRI affinities. Analysis and calculation of all binding data was carried out with Scrubber software (version 2) and Microsoft Excel (version 2402).

Cellular binding assay

Tumor and immune cells (1,5 × 105) were incubated with FcαRI BsAb or FcαRI-TNF for 1 h at 4 °C. After washing with 0,5% PBSA, cells were incubated with a directly labeled goat anti-human IgG secondary antibody (Jackson ImmunoResearch, 109-605-003) for 30 min at 4 °C, after which cells were washed and fixed in 4% PFA for 10 min at RT. Fluorescent signal was measured with a BD LSRFortessa Cell Analyzer.

Intravital imaging

Animals were subcutaneously injected with (2,0 × 105) MC38-cEGFR cells under 3,5% isoflurane anesthesia. Animals were treated intratumorally when individual tumor load had reached 100mm3. They were imaged 24 h later, at a tumor size of 120mm3. After a minimum of 2 h acclimatization, mice were injected intraperitoneally with a mix of 80 mg/kg Ketamine and 8 mg/kg Xylazine as primary anesthesia/analgesia. Following complete anesthesia, confirmed by the paw reflex, a cannula was placed in the retro-orbital sinus. Through this cannula, a maintenance mix of 150 µg/10 minutes Ketamine and 15 µg/10 minutes Xylazine was administered. Subcutaneous tumors were prepared by folding a skin flap and fixated on an imaging window of the microscopy platform of a Leica SP8 Confocal Microscope. Tumors were imaged up to 70 μm deep using a 40x oil objective. Following imaging, mice were sacrificed and the tumors frozen for immunohistochemistry. Cellular metrics were quantified using Imaris 9.9 XT with the LabKit extension for cell segregation.

Fluorescent immunohistochemistry.

Frozen tumor tissue was sectioned at a thickness of 5 μm and stained with 1 µg/mL DAPI (Invitrogen, D1306) for 5 min and 1 µg/mL directly labeled anti-Ly6G antibody (Biolegend, 123116) for 2 h at RT. Sections were imaged using the Olympus VS200 Research Slide Scanner (Olympus) with a 20x air objective. Images were then quantified using Imaris 9.9 XT with the LabKit extension for cell segregation via machine learning.

Statistical analysis

All statistical analyses were performed using Graphpad Prism 10.2.0. Statistical tests that were used are the unpaired t test to compare two conditions with a single variable; the 1-way ANOVA with Tukey’s multiple-comparison correction to compare more than two conditions with a single variable; and the 2-way ANOVA with Dunnet’s multiple-comparison correction to compare more than two conditions with more than one variable. Significance was accepted from p < 0.05 onwards.

Results

FcαRI BsAb and TNF-α co-treatment does not alter ADCP or macrophage phenotypes

First, it was investigated if TNF-α enhances ADCP by macrophages in the presence of FcαRI BsAb. Given that tumor-infiltrating macrophages often exhibit an M2-like, pro-tumor phenotype, macrophages were polarized toward either an M1- or M2-like phenotype prior to ADCP experiments. Successful phenotype skewing was confirmed with flow cytometry. CD40 has been established as a specific marker for M1-like macrophages, whereas CD200R serves as a marker for M2-like macrophages44,45,46,47. M1-like macrophages had a CD40high/CD200Rlow profile, while M2-like macrophages displayed a CD40low/CD200Rhigh expression pattern (Supplementary Fig. 1).

Incubation of macrophages and A431 tumor cells with TNF-α alone did not induce tumor killing (Fig. 1A). In contrast, FcαRI BsAb effectively triggered macrophage-mediated phagocytosis of tumor cells, which was not increased by the addition of TNF-α. Similar results were observed with the colon carcinoma cell line HT29, but HCT116 cells were resistant to ADCP (Supplemental Fig. 2), which was also previously reported48. M1-like macrophages showed similar ADCP activity compared to unskewed M0 macrophages, whereas M2-like macrophages were less effective. Incubation of either macrophages or A431 tumor cells alone with FcαRI BsAb, TNF-α or their combination did not alter cell viability (Supplementary Fig. 3A-B).

To determine whether ADCP modulated macrophage phenotype, surface expression of CD40 and CD200R post-treatment was measured. CD40 expression was elevated on M1-like macrophages compared to M2-like and M0 macrophages, which did not change following incubation with FcαRI BsAb, TNF-α or the combination (Fig. 1B). CD200R expression was lower on M1-like macrophages and increased on M2-like macrophages compared to M0 macrophage, and treatments did not affect surface expression of CD200R (Fig. 1C). Supernatants of tumor cell and macrophage co-cultures were assessed for cytokine concentrations, as different macrophage phenotypes secrete different cytokines49. Levels of human TNF-α, TGF-β and IL-10 were below detection limit (data not shown). IL-12p70 concentration was increased in supernatants of M1-like macrophages compared to M2-like and M0 macrophages, with no significant differences between treatments (Fig. 1D). Thus, although M1-like macrophages exhibited increased tumoricidal activity and IL-12p70 production compared to M2-like macrophages, treatment with FcαRI BsAb alone, or in combination with TNF-α did not influence macrophage phenotype or function.

Fig. 1
figure 1

Incubation with FcαRI BsAb and TNF-α does not alter ADCP or macrophage phenotypes. (A) ADCP of A431 tumor cells by pre-skewed macrophages after 24 h of co-culture in the presence of FcαRI BsAb, TNF-α or the combination, standardized against the untreated control. (B, C) Surface expression of macrophage phenotype markers CD40 (B) and CD200R (C) following ADCP assays of Fig. 1A. Measured by flow cytometry and displayed as the geometric mean fluorescent intensity. (D) Concentration of IL-12p70 in supernatants of ADCP assays from Fig. 1A. Unskewed macrophages are indicated with M0. M1-like or M2-like macrophages are indicated with M1 or M2. Graph A is representative for N = 3 experiments. Error bars indicate standard deviation. Graphs B-D show means of N = 3 experiments. Error bars indicate standard error. *P < 0.05; **P < 0.005; ***P < 0.0005 by 2-way ANOVA with Dunnet’s multiple-comparison correction.

Full size image

Incubation with FcαRI BsAb and TNF-α does not alter tumor cell killing or IFN-γ secretion by NK cells

Treatment with TNF-α alone did not induce A431 tumor cell killing by NK cells (Fig. 2A). In contrast, incubation with FcαRI BsAb led to efficient tumor killing. However, combining FcαRI BsAb with TNF-α did not further enhance cytotoxicity compared to FcαRI BsAb alone. These findings were confirmed with the HCT116 and HT29 colon carcinoma tumor cell lines, although HCT116 proved sensitive to TNF-α alone (Supplementary Fig. 4).

To evaluate NK cell activation, IFN-γ levels were measured in the supernatants of the A431 tumor killing assays. The pattern of IFN-γ secretion by NK cells mirrored that of tumor cell killing. FcαRI BsAb induced IFN-γ release, which was not further increased by the addition of TNF-α. TNF-α alone did not induce secretion of IFN-γ (Fig. 2B).

Fig. 2
figure 2

Incubation with FcαRI BsAb and TNF-α does not alter tumor cell killing or IFN-γ secretion by NK cells. (A) A431 tumor killing by NK cells after 24 h of co-culture in the presence of FcαRI BsAb, TNF-α or the combination, measured by CellTiter Blue assay and standardized against untreated controls. (B) Secretion of interferon gamma in supernatants of tumor killing assays of Fig. 2A as measured with ELISA. Graphs show means of N = 4 experiments. Error bars indicate standard error. By 1-way ANOVA with Tukey’s multiple-comparison correction.

Full size image

The combination of TNF-α and FcαRI BsAb potentiates neutrophil tumor killing and degranulation in vitro

To study trogocytosis, tumor cells were stained with a fluorescent membrane dye (DiO). Since trogocytosis physically disrupts the tumor cell membrane, after which it is internalized by neutrophils, it can be measured with flow cytometry. Co-incubation of A431 tumor cells with neutrophils and FcαRI BsAb resulted in neutrophils that were positive for the DiO tumor membrane staining, which was increased when TNF-α was added (Fig. 3A,B). Incubation with TNF-α alone did not induce trogocytosis of A431 tumor cells by neutrophils. Additionally, incubation of neutrophils or A431 tumor cells with FcαRI BsAb, TNF-α or the combination did not alter cell viability (Supplementary Fig. 3C-D).

Tumor cell killing was also assessed using the CellTiter Blue assay, measuring tumor cell viability. Treatment with TNF-α or with the combination of FcαRI BsAb and TNF-α increased neutrophil-mediated killing of A431, HCT116 and HT29 tumor cells (Fig. 3C, Supplementary Fig. 5). In the absence of neutrophils, no A431 tumor cell killing was observed in the presence of FcαRI BsAb, TNF-α or the combination (Supplementary Fig. 6).

To investigate neutrophil activation and degranulation, surface expression of CD11b and CD66b on neutrophils was assessed after 4 h of co-incubation. CD11b surface expression on neutrophils did not change following co-incubation with FcαRI BsAb, TNF-α or the combination of the two (Fig. 3D). Treatment with the combination of FcαRI BsAb and TNF-α increased surface expression of CD66b compared to FcαRI BsAb alone (Fig. 3E). Secretion of lactoferrin into the supernatant following co-incubation followed a similar pattern as surface expression of CD66b. Treated with a combination of FcαRI BsAb and TNF-α, neutrophils secreted more lactoferrin when compared to FcαRI BsAb alone (Fig. 3F). Furthermore, A431 tumor cell killing by neutrophils was assessed in the presence of NETosis or ROS inhibitors (DNAse or sodium pyruvate, respectively). Neither treatment affected tumor cell killing (Supplementary Fig. 7).

Fig. 3
figure 3

Combination of TNF-α and FcαRI BsAb potentiates neutrophil A431 tumor killing and degranulation in vitro. (A) Representative dot plots of neutrophils after 4 h of co-culture with DiO-labeled A431 tumor cells in the presence of FcαRI BsAb, TNF-α or the combination. (B) Quantified trogocytosis of DiO-labeled A431 tumor cells by neutrophils in the experiments of Fig. 3A. Measured by flow cytometry and displayed as the percentage DiO+ neutrophils. (C) Neutrophil-mediated A431 tumor killing following 4 h of co-culture in the presence of FcαRI BsAb, TNF-α or the combination. Measured by CellTiter Blue assay and standardized against untreated controls. (D,E) Surface expression of neutrophil markers CD11b (D) and CD66b (E) following trogocytosis assays of Fig. 3A. Measured by flow cytometry and displayed as the geometric mean fluorescent intensity. (F) Secretion of lactoferrin in supernatants of trogocytosis assays of 2 A. Measured with ELISA. Graphs show means of N ≥ 3 experiments. Error bars indicate standard error. *P < 0.05; ****P < 0.00005 by 1-way ANOVA with Tukey’s multiple-comparison correction.

Full size image

Fusion of TNF-α to FcαRI BsAb does not affect antibody binding affinity

To generate a multifunctional immunocytokine, a single-chain TNF-α trimer was genetically fused to the C-terminus of the FcαRI-binding heavy chain of the FcαRI BsAb (Fig. 4A). SDS-PAGE was performed to verify the efficiency of Fab arm exchange and assess the structural integrity of the resulting constructs. For cetuximab, distinct bands of approximately 25 and 50 kDa were observed, corresponding to the light and heavy chains, respectively (Fig. 4B). Both FcαRI BsAb and FcαRI-TNF showed two bands of ± 25 kDa were found, consistent with the presence of two variable light chain regions, owing to their bi-specific nature. In the FcαRI-TNF lane, in addition to the heavy chain band of the EGFR-binding arm at ± 50 kDa, an extra band was observed at ± 100 kDa. This band corresponds to the TNF-α-bearing heavy chain of the FcαRI-binding arm, which constitutes a heavy chain (50 kDa) and the TNF-α trimer payload (51 kDa).

Next, binding affinities to hEGFR and Fc receptors were determined. SPR was used to determine the equilibrium dissociation constants (KD) of FcαRI-TNF for human EGFR, FcαRI and both the low and medium affinity variants of FcɣRIIIa (158 F and 158 V, respectively). Minimal differences were observed in the KD for hEGFR, FcαRI, FcɣRIIIa 158 F and FcɣRIIIa 158 V between FcαRI BsAb and FcαRI-TNF (Fig. 4C-F), supporting that the fusion of TNF-α did not compromise receptor or antigen binding.

Fig. 4
figure 4

Fusion of TNF-α to FcαRI BsAb does not interfere with antibody binding affinity. (A) Schematic representation of the FcαRI-TNF protein structure. The light gray sphere indicates the F405L Duobody mutation, while the dark gray sphere indicates the K409R mutation. Created in BioRender. Van Egmond, M. (2024) https://BioRender.com/q55n353. (B) SDS-PAGE of Cetuximab, FcαRI BsAb and FcαRI-TNF on a reducing 12% Acrylamide gel. The uncropped gel is presented in Supplementary Fig. 8. (C–F) Dissociation constants of FcαRI BsAb and FcαRI-TNF for hEGFR (C), FcαRI (D), FcɣRIIIa 158 F (E) and FcɣRIIIa 158 V (F) as determined by Surface Plasmon Resonance. The kinetic chromatographs, underlying these data points, are presented in Supplementary Fig. 9. Graphs C-F show means of N = 3 experiments. Error bars indicate standard error. *P < 0.05; ***P < 0.0005; ****P < 0.00005 by unpaired t test.

Full size image

FcαRI-TNF fusion protein increases tumor cell killing by neutrophils in vitro

The functional activity of the FcαRI-TNF fusion protein was also assessed. First, its binding ability to target and effector cells was evaluated with flow cytometry. Binding to A431 tumor cells and FcɣRIII-expressing macrophages did not significantly deviate from binding of the parental FcαRI BsAb (Fig. 5A-B). Binding of FcαRI-TNF to FcαRI-expressing neutrophils was slightly decreased (Fig. 5C).

Finally, the ability of FcαRI-TNF to induce ADCP or neutrophil-mediated tumor cell killing was compared to incubation with FcαRI BsAb alone or the combination of FcαRI BsAb and TNF-α. Incubation with FcαRI-TNF induced similar ADCP by macrophages in the presence of FcαRI BsAb or the combination of FcαRI BsAb and TNF-α (Fig. 5D). In contrast, FcαRI-TNF significantly enhanced neutrophil-mediated A431 tumor killing, mirroring the increased cytotoxicity that was observed with the FcαRI BsAb and TNF-α combination. This tumoricidal activity was markedly greater compared to incubation with FcαRI BsAb alone (Fig. 5E).

Fig. 5
figure 5

FcαRI-TNF fusion protein increases tumor cell killing by neutrophils in vitro. (A–C) Cellular binding of FcαRI BsAb and FcαRI-TNF to A431 tumor cells (A), macrophages (B) and neutrophils (C), measured by flow cytometry and displayed as the geometric mean fluorescent intensity. (D) ADCP of A431 tumor cells by macrophages in the presence of FcαRI BsAb, the combination of FcαRI BsAb and TNF-α or FcαRI-TNF after 24 h of co-culture, standardized against the untreated control. (E) Neutrophil-mediated A431 tumor killing following 4 h of co-culture in the presence of FcαRI BsAb, the combination of FcαRI BsAb and TNF-α or FcαRI-TNF, measured by Cell Titer Blue assay and standardized against the untreated controls. Graphs (A–C) show the mean of N = 3 experiments. Error bars indicate standard error. *P < 0.05; ***P < 0.0005 by 2-way ANOVA with Dunnet’s multiple-comparison correction. Graphs D-E are representative for N = 3 experiments. Error bars indicate standard deviation. By 1-way ANOVA with Tukey’s multiple-comparison correction.

Full size image

FcαRI-TNF increases neutrophil infiltration and activation in tumors in vivo

The EGFR-binding arm of FcαRI-TNF is derived from cetuximab, which does not bind to murine EGFR. Previously, a murine EGFR variant has been described, in which six amino acids were substituted to allow binding of cetuximab50. Therefore, an MC38 tumor cell line expressing this chimeric EGFR (cEGFR) was established, which was used in a syngeneic immunocompetent mouse model. FcαRI-TNF selectively bound to MC38-cEGFR cells, but not to the parental MC38 cell line (Supplementary Fig. 10).

To assess the in vivo impact of FcαRI-TNF on neutrophil dynamics within the tumor microenvironment, intravital imaging was performed on MC38-cEGFR tumor-bearing mice that had been treated intratumorally 24 h prior to imaging (Fig. 6A). Treatment with FcαRI BsAb modestly enhanced neutrophil infiltration into tumors compared to treatment with PBS. TNF-α treatment alone also increased neutrophil recruitment, but treatment with FcαRI-TNF resulted in the highest number of tumor-infiltrating neutrophils (Fig. 6B, and Supplementary Fig. 11). Quantitative analysis of neutrophil motility showed that FcαRI-TNF treatment significantly increased mean neutrophil speed relative to treatment with PBS, FcαRI BsAb or TNF-α alone (Fig. 6C). Inverse results were found for path straightness of neutrophils, as treatment with FcαRI-TNF decreased path straightness (Fig. 6D). Neither FcαRI BsAb nor TNF-α alone affected path straightness. Strikingly, only FcαRI-TNF treatment induced neutrophil swarming behavior (Fig. 6A and E). a hallmark of potent immune activation within tissues, further underscoring the enhanced functional activation of neutrophils in vivo.

Fig. 6
figure 6

FcαRI-TNF increases tumor infiltration and activation of neutrophils in vivo. (A) Representative images of intravital tumor imaging. Neutrophils are marked in green. Scale bars represent 60 μm. (B) Number of DAPI/LysM+/+ neutrophils per µm2 in tumor sections of experiments in Fig. 6A. (C) Mean movement speed and (D) straightness of movement path of tumor-infiltrating neutrophils 24 h after intratumoral treatment. Visualized by intravital microscopy. (E) Absolute instances of swarming behavior by tumor-infiltrating neutrophils 24 h after intratumoral treatment. Visualized by intravital microscopy. Graphs in 6B-D show mean values of N ≥ 4 animals. Error bars indicate standard error. *P < 0.05; **P < 0.005 by 1-way ANOVA with Tukey’s multiple-comparison correction.

Full size image

Discussion

mAb-based immunotherapies have shown significant clinical success in treating various cancers. However, their therapeutic potential is often limited by the immunosuppressive TME, which impairs the recruitment and function of immune effector cells. We therefore investigated whether the addition of the pro-inflammatory cytokine TNF-α increased the therapeutic potential of an FcαRI-stimulating BsAb.

FcαRI BsAb effectively induced M1 and M0 macrophage- induced tumor cell killing. Macrophages polarized toward an M2-like phenotype retained phagocytic capacity, albeit at reduced efficiency. It has been described that TNF-α, in combination with LPS or IFN-γ, polarizes macrophages to an M1-like phenotype30,51. However, in our system, neither surface marker expression nor secretion of cytokines was affected by the TNF treatments, which may have been due to absence of co-stimulating factors. Furthermore, combining TNF-α with FcαRI BsAb did not enhance ADCP. Nonetheless, CRC, for which FcαRI BsAb targeting EGFR could be relevant, often exhibits a favorable M1/M2 macrophage ratio, suggesting that macrophages may still contribute to FcαRI BsAb efficacy in CRC, although the addition of TNF-α is unlikely to enhance this effect45,46.

Previous studies have reported that exposure to TNF-α can enhance NK cell activity32,52. Mechanistically, it has been suggested that TNF-α sensitizes NK cells to other stimuli, but does not directly trigger activation on its own33,53. NK cell anti-tumor efficacy may therefore depend on the combination of activating stimuli. In our experiments, stimulation of NK cells with TNF-α and FcαRI BsAb did not enhance tumor cell killing, beyond the effect of FcαRI BsAb alone. Likewise, TNF-α alone did not induce NK cell-mediated tumor cell killing. This pattern was mirrored in IFN-γ secretion. These results suggest that FcγR activation by FcαRI BsAb is not influenced by TNF-α-induced sensitization of NK cells, in contrast to receptors for IL-12 and IL-18, which have been shown to respond to sensitization by TNF-α32,54. According to pathway analyses, FcγRIIIa or IL-18R signaling engage distinct downstream pathways, which may explain the lack of TNF-α effect on FcαRI BsAb-induced tumor killing or IFN-γ production by NK cells55,56.

In contrast, neutrophil-mediated killing was increased by combining FcαRI BsAb with TNF-α. While incubation with TNF-α alone was ineffective, treatment with FcαRI BsAb induced trogocytosis,. Interestingly, when FcαRI BsAb and TNF-α were combined, uptake of tumor material by neutrophils increased. This suggests that although TNF-α does not directly induced trogocytosis, it promoted antibody-mediated trogocytosis. Similar levels of tumor cell killing (as measured with the CellTiter Blue assay) were observed after treatment with TNF-α alone or in combination with FcαRI BsAb, but only in the presence of neutrophils. This effect was not due to direct cytotoxicity of TNF-α on tumor cells, since TNF-α treatment alone did not affect cell viability and tumor cell killing was abolished in the absence of neutrophils. These findings indicate that TNF-α enhances tumor killing in a neutrophil-dependent manner.

TNF-α has been reported to increase neutrophil-mediated tumor cytotoxicity in breast cancer patients by inducing PRKCI, which is involved in the production of ROS57. Additionally, TNF- α enhanced NET formation by neutrophils from rheumatoid arthritis58. However, we found that enhanced tumor cell killing, induced by the combination of FcαRI BsAb and TNF-α, was independent of NETs and ROS, as inhibition of NET formation with DNase and ROS scavenging with sodium pyruvate did not affect tumor cell killing (Supplementary Fig. 7). Mechanistically, TNF-α increased degranulation of specific granules by neutrophils, as evidenced by enhanced CD66b surface expression and release of lactoferrin, which are hallmarks of specific granule mobilization52,53,54. It was previously shown that TNF-α primes neutrophils, lowering the threshold required for degranulation59,60. Thus, our results suggest that TNF-α both primes neutrophils for degranulation and promotes trogocytosis of tumor cells.

Given the systemic toxicity of, we engineered an FcαRI-TNF immunocytokine as localized delivery system35,36. Despite the relatively large TNF-α payload size (51 kDa) compared to the antibody moiety (150 kDa) FcαRI-TNF retained strong binding to hEGFR, FcαRI, FcγRIIIa 158 F and FcγRIIIa 158 V, with only minimal decreases in binding affinity Cellular binding to neutrophils was slightly reduced, but functional readouts such as phagocytosis by macrophages and neutrophil-mediated tumor killing were not altered, indicating that both the antibody and TNF-α moieties of FcαRI-TNF were functional.

Neutrophils in FcαRI-TNF-treated tumors migrated through the tumor at a higher mean speed, albeit in a less straight path. While in vitro and ex vivo experiments with a single chemotactic gradient suggest that increased speed correlates with migratory efficiency, the in vivo tumor context likely includes competing gradients and cellular interactions that modulate this behavior55,56,57. Due to the presence of a tumor-targeting BsAb neutrophils can bind tumor cells via FcαRI-TNF in our model, potentially affecting migration. Additionally, the tumor environment may have multiple chemotactic gradients and other factors, which can also affect neutrophil migration.

A key observation was that only FcαRI-TNF treatment induced neutrophil swarming, which is a classic hallmark of neutrophil activation. Swarming typically occurs at sites of infection or sterile inflammation61,62,63. We previously demonstrated that crosslinking of FcαRI leads to the release of LTB4, which is critical for swarming, as neutrophils deficient for Ltb4r1 lacked a swarming response in a sterile skin injury model63,64,65. Additionally, FcαRI BsAb induced neutrophil swarming in vitro, which we however did not observe in the MC38-cEGFR tumor model10. Possibly, treatment with FcαRI BsAb alone did not attract a sufficient number of neutrophils into the tumor, preventing formation of a chemotactic gradient necessary for swarming, whereas the E: T ratio in vitro was likely much higher. Treatment with TNF-α alone enhanced neutrophil infiltration, but did not induce swarming, due to the absence of FcαRI crosslinking and concomitant LTB4 release. In contrast, FcαRI-TNF provided both signal, necessary for neutrophil swarming. The TNF-α payload facilitated neutrophil infiltration, while the FcαRI BsAb crosslinked FcαRI, triggering LTB4 release.

In conclusion, these findings demonstrate that TNF-α enhances neutrophil-mediated cytotoxicity when combined with an FcαRI BsAb, primarily by increasing trogocytosis and degranulation. Incorporating TNF-α into an FcαRI-targeting immunocytokine (FcαRI-TNF) improved tumor infiltration and in vivo swarming behavior. This dual-function immunocytokine represents a promising strategy to potentiate neutrophil-driven anti-tumor immunity, potentially overcoming limitations imposed by the immunosuppressive TME and improving the efficacy of antibody-based cancer immunotherapy.