The human female breast adenocarcinoma cell line SKBR3 (ATCC, Manasses, Virginia, USA), which overexpresses the human epidermal growth factor receptor 2 (HER2), was maintained in RPMI 1640 Medium GlutaMAX Supplement (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA), supplemented with 10% heat-inactivated (hi) fetal calf serum (FCS; Linaris, Wertheim-Bettingen, GER). Cells were incubated at 37°C in a 5% CO₂ humidified environment.

Trastuzumab (Herceptin) was provided by the institutional pharmacy (Vienna General Hospital, Medical University of Vienna). VPA (Sigma-Aldrich, Saint Louis, Missouri, USA) was dissolved in Dulbecco's phosphate-buffered saline without calcium and magnesium (DPBS −/−; Thermo Fisher Scientific Inc.). Vorinostat (Zolinza), also known as suberanilohydroxamic acid (SAHA; Selleckchem, Munich, GER) was dissolved in dimethyl sulfoxide (DMSO; Thermo Fisher Scientific Inc.).

SKBR3 cells were treated with either VPA or SAHA at different concentrations (online supplementary figure S1a, b), collected after 24 hours (h), washed with DPBS −/− and subsequently stained with the FITC Annexin V Apoptosis Detection Kit I (BD Pharmingen, Franklin Lakes, New Jersey, USA), according to manufacturer’s protocol. Subsequently, cells were measured by a Gallios G flow cytometer (Beckman Coulter Inc., Brea, California, USA), and further analyzed with Kaluza analysis software (V.2.1.1; Beckman Coulter Inc.). Viable cells were determined as Annexin V and propidium iodide (PI) double negative (Annexin V−/PI−), early apoptotic cells as Annexin V positive and PI negative (Annexin V+/PI−) and end-stage apoptotic dead cells as Annexin V and PI double positive (Annexin V+/PI+).

Peripheral blood (27 mL) was drawn into VACUETTE EDTA tubes (Greiner Bio-One International, Kremsmuenster, Upper Austria, AUT) from sex-matched and aged-matched healthy volunteers (n=5). PBMCs were immediately isolated using Ficoll-Paque gradient centrifugation (GE Healthcare Bio-Sciences, Uppsala, Sweden). PBMCs were washed twice with DPBS −/− and further resuspended in RPMI 1640 Medium GlutaMAX supplemented with 10% FCS.

SKBR3 cells were trypsinized (Thermo Fisher Scientific Inc.) and resuspended in DPBS −/− to a concentration of 5×10⁷/mL. The cells were further incubated with 1 µL of 0.5 mM carboxyfluorescein diacetate succinimidyl ester (CFSE; Invitrogen, Eugene, Oregon, USA) per 1×10⁶ cells for 10 min (min) at 37°C in a 5% CO₂ humidified atmosphere. Afterwards cells were washed in RPMI 1640 Medium GlutaMAX supplemented with 10% FCS and resuspended to a concentration of 5×10⁵/mL. The cells (200 µL) were further put into 5 mL tubes (BD Falcon) and treated either with 5 mM VPA, 10 mM VPA, 5 µM SAHA,10 μM SAHA or left untreated for 24 hours at 37°C in a 5% CO₂ humidified environment.

ADCP and ADCC activities were measured as previously described.28–30 Briefly, freshly isolated PBMCs were resuspended in RMPI 1640 supplemented with 10% FCS to the concentration of 12.5×10⁶/mL or 6.25×10⁶/mL. PBMCs (200 µL) were added to the CFSE-labeled and HDACi-pretreated SKBR3 cells, resulting in an effector to target (E:T) ratio of 25:1 or 12.5:1 (total reaction volume 400 µL). The cells were incubated together for 2.5 hours at 37°C in a 5% CO₂ humidified environment in the presence or absence of trastuzumab at a concentration of 2.5 µg/mL (figure 1).

Afterwards, the cells were washed in DPBS −/− and stained using Zombie Violet Fixable Viability Kit (BioLegend, San Diego, California, USA), according to the manufacturer’s protocol. Next, the cells were washed with DPBS −/− supplemented with 2% hi FCS and stained with anti-human CD14 allophycocyanin (APC) for 30 min at 4°C.

Staining events were quantified by a Gallios G flow cytometer (Beckman Coulter Inc.) or Amnis ImageStreamXMark II imaging flow cytometer (IFC; Luminex Corporation, Austin, Texas, USA) and further analyzed with Kaluza Analysis Software V.2.1.1 (Beckman Coulter Inc.) or IDEAS software (Luminex Corporation).

CFSE and CD14 APC double-positive cells (CFSE+/CD14+) are phagocytosed cells by ADCP (figure 2A–C), whereas CFSE and Zombie Violet double-positive cells (CFSE+/Zombie Violet+) were classified as lysed cells by ADCC (online supplementary figure S2a-c). ADCP and ADCC events are given as a percentage of all CFSE-positive tumor cells.

For the VPA or SAHA-induced secretome experiments, SKBR3 cells were treated with VPA or SAHA, for 24 hours. Afterwards PBMCs were stimulated with the VPA or SAHA-induced secretome (supernatant) of SKBR3 cells for 12 hours, following the ADCP/ADCC assay as described previously.

SKBR3 cells were treated with different VPA and SAHA concentrations as described previously. The cells were further incubated with primary antibodies for HER2, CALR, CD47 or MCL1 (online supplementary table S1) for 30 min at 4°C. Unconjugated antibodies were further stained with secondary antibodies for 30 min at 4°C (online supplementary table S1). In the case of an intracellular protein staining, the cells were fixed and permeabilized with the IntraPrep Permeabilizaton Reagent (Beckman Coulter Inc.) according to the manufacturer’s protocol. Different protein expression levels were analyzed by the use of a Gallios G flow cytometer (Beckman Coulter Inc.) and quantified with Kaluza Analysis Software V.2.1.1 (Beckman Coulter Inc.). Histogram overlays were created with FlowJo V.10.6.1 (Becton Dickinson, Franklin Lakes, New Jersey, USA).

FcγRI (CD64), FcγRIIA (CD32A), FcγRIIB (CD32B) and FcγRIII (CD16) expressions on monocytes (CD16+) were measured after the 2.5-hour incubation time of the ADCC/ADCP assay, without prior CFSE labeling of SKBR3 cells and without trastuzumab. As CD32A and CD32B can only be distinguished on their intracellular domain, the cells were fixed and permeabilized with the IntraPrep Permeabilizaton Reagent (Beckman Coulter) according to the manufacturer’s protocol. The cells were incubated with primary antibodies (online supplementary table S1) for 30 min at 4°C. Unconjugated antibodies were further stained with secondary antibodies for 30 min at 4°C (online supplementary table S1). FcγR expression levels were analyzed using a Gallios G flow cytometer (Beckman Coulter Inc.) and quantified with Kaluza Analysis Software V.2.1.1 (Beckman Coulter Inc.). Histogram overlays were created with FlowJo V.10.6.1 (Becton Dickinson, Franklin Lakes, New Jersey, USA).

SKBR3 cells (1×10⁶) were seeded on 6-well plates. For transfection, the cells were cultured in Opti-MEM Reduced Serum Media (Invitrogen) and the cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. Test groups were transfected with MCL1 small interfering RNA (siRNA; sc-35877; Santa Cruz, Texas, USA), whereas control groups were transfected with AllStars Negative Control siRNA (Quiagen, Hilden, North Rhine-Westphalia, GER). The medium was exchanged to SKBR3 medium 6 hours after transfection. The cells were further cultivated for 24 hours before being used in the ADCP/ADCC assay as described before.

Statistical tests and graphs were performed with GraphPad Prism 7.0a (GraphPad Software, La Jolla, California, USA). The assessment of agreement between FCM and IFC results was done by Bland-Altman plots.31 All other activities or expression data are presented by scatter plots (circles, boxes and triangles demonstrate individual values) and corresponding mean (long horizontal line) and SD or SE of the mean (SEM). Comparisons for ADCP/ADCC and FcγR results were investigated by a repeated-measures (RM) one-way analysis of variance (ANOVA) and Fisher least significant difference (LSD) post hoc tests. All other comparisons were calculated by a one-way ANOVA and Fisher LSD post hoc tests. Two-sided p values ≤0.05 were regarded as statistically significant.

In order to investigate ADCP and ADCC, we used an FCM-based approach, which facilitates high throughput data acquisition (online supplementary figure S1). We first determined the reliability of FCM to measure accurate ADCP and ADCC events, in terms of truly phagocytosed/incorporated cells in our setting. In this line, we compared each sample with IFC and conventional FCM, despite considering that both methods rely on different technologies (imaging vs quantification of optical parameters) and different data output. Analysis of ADCP (CFSE+/CD14+) and ADCC (CFSE+/ Zombie Violet+) images show truly phagocytosed and lysed cells, respectively (figure 1a–c; online supplementary figure S3a-c).

Data agreement between the two different methods (IFC vs FCM) was further investigated by Bland-Altman plots. For ADCP, the bias (average of the differences) between IFC and FCM measured values was 0.983 and corresponding 95% limits of agreement ranged from −2.822 to 4.789 (figure 2d). ADCC comparison resulted in a bias of 0.213 and its associated 95% limits of agreement from −6.396 to 6.823 (online supplementary figure S3d). We further conducted a Pearson’s correlation (scatter plots not shown) for ADCP (r=0.991; p<0.001) and ADCC (r=0.857; p<0.001). Taking together, measurement of ADCP and ADCC by FCM detects true phagocytic events and is comparable with IFC.

In order to investigate a potential additive or synergistic immunomodulatory effect of valproic acid (VPA) or vorinostat (SAHA) on ADCP or ADCC activity, we performed an ADCP/ADCC assay as shown in the online supplementary figure S1.

We intended to induce limited tumor cell stress that would result in approximately 75% tumor cell survival. We applied sublethal doses of 5 and 10 mM VPA and 5 and 10 µM SAHA, respectively, for 24 hours, which resulted in about 25% early apoptotic tumor cells (online supplementary figure S2).

Strikingly, trastuzumab-mediated ADCP activity was significantly improved by VPA (p<0.01; figure 2A) and SAHA (p<0.01; figure 2B) pretreatment. In contrast, VPA or SAHA alone did not enhance phagocytosis compared with untreated tumor cells (online supplementary figure S4a-b). However, the trastuzumab-enhancing effect was not seen at a lower (12:1) E:T ratio (online supplementary figure S4c-f).

In contrast to ADCP, trastuzumab-independent cytotoxicity was significantly enhanced by VPA (p<0.05; figure 3A) and SAHA (p<0.05; figure 3B). The addition of trastuzumab however did not affect ADCC compared with untreated tumor cells (online supplementary figure S5a-b). Corresponding to the ADCP results, there was also no enhancing effect observed at an E:T ratio of 12:1 (online supplementary figure S5c-f).

To examine mechanistic effects of VPA or SAHA tumor cell pretreatment on trastuzumab-enhanced ADCP activity, we first investigated potential expression changes of our target receptor HER2. Interestingly, VPA and SAHA lead to a decreased protein expression level compared with untreated tumor cells (p<0.05; online supplementary figure S6a-b).

Next, we examined the different antibody-meditating receptors on monocytes (CD14+). We performed the ADCP/ADCC assay as described above and stained for FcγRI (CD64), FcγRIIA (CD32A), FcγRIIB (CD32B) and FcγRIII (CD16) expression on CD14 +cells (figure 4A–D). Protein expression analysis using FCM revealed a statistically significant upregulation of the activating FcγRIIA by VPA (p<0.01; figure 4B). All other FcγR were not affected by VPA. SAHA had no impact on any FcγR.

To get a deeper understanding of how VPA and SAHA affect ADCP activity, we examined the major anti-apoptotic molecule MCL1. MCL1 protein expression was significantly downregulated by VPA and SAHA (p<0.001; p<0.001; figure 5A). To investigate whether this effect is MCL1 dependent, we knocked-down MCL1 by siRNA (figure 5B) and repeated the ADCP/ADCC assay in the absence of HDACi pretreatment (figure 5C). Strikingly, ADCP activity was significantly enhanced in comparison with untreated controls (p<0.05). Notably, this effect was only seen in combination with the antibody trastuzumab (online supplementary figure S7a). However, ADCC was not affected by MCL1 knockdown, neither with nor without trastuzumab (online supplementary figure S7b, c).

As VPA and SAHA enhance trastuzumab-independent cytotoxicity, we further investigated if this observation might be explained by the induction of a so-called ICD, which is in part characterized by the surface exposure of CALR, which is counterbalanced by the “do not eat me” signal CD47.27 32

FCM analysis revealed a statistically significant upregulation of CALR surface exposure by VPA (p<0.001; figure 6A) and SAHA (p<0.05; figure 6B), respectively. Moreover, the “do not eat me” signal CD47 was also downregulated on SKBR3 cells by both HDACi, VPA and SAHA (p<0.001; figure 6C; p<0.001, figure 6D).

Next, we investigated the potential immunostimulatory effects of the HDACi-induced secretome. We stimulated effector cells (PBMCs) with VPA or SAHA-induced secretome of tumor cells for 12 hours prior to the ADCP/ADCC assay. However, neither ADCP nor ADCC, with or without trastuzumab showed enhanced ADCP/ADCC activity compared the tumor cell secretome of untreated cells (online supplementary figure S8a-h).