A Real-Time Image-Based Efferocytosis Assay for the Discovery of Functionally Inhibitory Anti-MerTK Antibodies

Efferocytosis is a phagocytic process by which apoptotic cells are engulfed by professional and nonprofessional phagocytes. Clearance of apoptotic cells is essential for embryonic and postnatal development, as well as the maintenance of tissue homeostasis in normal physiology (1, 2). Efferocytosis also plays a critical role in resolving inflammation in host defense against microbes and restoring homeostasis after infections (3, 4). During the first step of efferocytosis, apoptotic cells translocate phospholipid phosphatidylserine (PtdSer; the “eat me” signal) from the inner side to the outer leaflet of their plasma membranes. Then specific cell surface receptors on phagocytes engage apoptotic cells through direct or indirect binding of PtdSer. Last, the PtdSer engagement activates the cell surface receptors on phagocytes and trigger the engulfment of apoptotic cells (5, 6). A family of receptor tyrosine kinases called Tyro3, Axl, Mer/MerTK (TAM) receptors on phagocytes function as the main efferocytosis receptors (7–9). They mediate efferocytosis by binding PtdSer indirectly through endogenous vitamin K–dependent ligands, protein S, and growth arrest–specific 6 (Gas6), which act as bridging molecules between apoptotic cells and phagocytes (10).

In tumors, many cancer cells become apoptotic and necrotic under stress conditions (hypoxia, nutrient deficiency, etc.), which leads to the release of tumor Ags that trigger host inflammatory responses (11). However, many tumors exploit the efferocytosis function of tumor-associated macrophages to engulf apoptotic cancer cells, leading to the clearance of associated tumor Ags, which helps the tumor escape the host immune surveillance (11). Tumor-associated macrophages are involved in tumor angiogenesis (12, 13) and display a unique M2-like phenotype characterized by enhanced expression of scavenger receptors CD163 and CD204 (14). Tumor-associated macrophages also produce immunosuppressive cytokines (i.e., IL-10, TGF-β1) that sustain a protumor microenvironment while expressing checkpoint ligands programmed death-ligand 1 (PD-L1) and PD-L2 to inhibit T cell cytotoxicity (15–17). In recent years, tumor-associated macrophages have been reported to contribute to the resistance of checkpoint inhibitors and adaptive T cell immunotherapies (18, 19). So far, tumor-associated macrophages have been linked with poor prognosis in many types of solid tumors (20–24). Therefore, targeting tumor-associated macrophages has become a new approach to treat cancers, especially those that are resistant to immune checkpoint inhibitors (25–28). As one of three tyrosine receptor kinases that are critical to efferocytosis, MerTK is predominantly expressed in tumor-associated macrophages and plays an essential role in the clearance of dying tumor cells and tumor Ags, as well as the maintenance of the immune-suppressive protumor microenvironment (7, 29–31). Thus far, several preclinical studies have shown that inhibiting MerTK can be an effective method for combating tumors as a single agent or in combination with other therapies, such as immune checkpoint inhibitors (32–35).

To screen and characterize efferocytosis antagonists targeting MerTK, a specific, high-throughput, and quantitative efferocytosis assay is desirable. Historically, efferocytosis assays have been performed using fluorescence microscopy. However, this microscope method only works when the phagocytic cells are fixed at the right time (i.e., containing phagosomes with engulfed cargoes). Because it is not high throughput or quantitative, the fluorescence microscopy method is considered only a qualitative assay, which is not suitable for drug screening activities. Another approach uses flow cytometry to study efferocytosis (36–38). Even though this method is high throughput, confocal fluorescence microscopy is usually performed side by side to validate that the fluorescence reading is localized inside the phagocytic cells (rather than adherent to the phagocyte surface), confirming that the measurement is indeed efferocytosis (39, 40). This assay platform (known as “imaging flow cytometry”) combines the single-cell imaging capability of microscopy and the high throughput of flow cytometry. It is capable of differentiating cell-bound from internalized materials without losing the high-throughput robustness (41, 42). Nevertheless, no matter which flow cytometry methods are used, the release of adherent macrophages and costaining of macrophage markers as well as apoptotic cells with different fluorescent dyes are usually part of the experimental methods. Moreover, a time-course study is often required to determine incubation time for efferocytosis, Therefore, the flow cytometry–based efferocytosis assays are considered to be time consuming as well as technically complicated and challenging.

In this report, we describe the development of, to our knowledge, a novel efferocytosis assay using the Incucyte Live-Cell Analysis System, a real-time imaging system for live-cell analysis. When apoptotic cells labeled with pH-sensitive pHrodo dye are engulfed by macrophages, the acidic environment in phagosomes of macrophages leads to increased pHrodo fluorescence, which can be recorded in real time by the Incucyte as efferocytosis readout. By optimizing the preparation of apoptotic cells, we established a quantitative and robust efferocytosis assay in 96-well plates. The established efferocytosis assay can be conveniently performed using phagocytic cell lines or primary macrophages, including freshly isolated tissue-resident macrophages and in vitro differentiated macrophages from different species. Using this newly developed assay, we successfully discovered functional anti-MerTK Abs that block tumor-associated macrophage–mediated efferocytosis for cancer immunotherapy.

Recombinant murine and human MerTK proteins as well as human Gas6-IgG1-Fc fusion protein were made at Genentech. Recombinant human proteins M-CSF (300-25), GM-CSF (300-03), IFN-γ (300-02), IL-10 (200-10), and recombinant mouse protein M-CSF (315-02) were obtained from PeproTech (Rocky Hill, NJ). Staurosporine from Streptomyces sp. (S4400) and LPS from Escherichia coli K-235 (L2018) were obtained from Sigma-Aldrich (St. Louis, MO). Cytochalasin D was acquired from Tocris (Minneapolis, MN). pHrodo Red, succinimidyl ester (pHrodo Red SE; p36600), and propidium iodide (P1304MP) were obtained from Thermo Scientific (Waltham, MA). Polyclonal anti-mouse MerTK Ab AF591 (AF591-SP) and polyclonal anti-human MerTK Ab AF891 (AF891-SP) were obtained from R&D Systems (Minneapolis, MN).

The cell surface receptor expression profiles were analyzed by flow cytometry. Briefly, 1 million cells were incubated for 30 min on ice in PBS containing 2% FBS with 1 µg of relevant Abs and then washed twice with cold PBS. Flow cytometry results were acquired using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA), and the data were analyzed with FlowJo (BD Biosciences). To prevent nonspecific binding to Fcγ receptors on the analyzed cells, 5 µl of TruStain FcX (Fc Receptor Blocking Solution, BioLegend) was added per 1 million cells in 100-µl staining volume, then mixed and incubated at room temperature for 5–10 min prior to staining with the Ab of interest. The following reagents were used in flow cytometry: PE anti-mouse CD16.2 (BioLegend, 149504), PE anti-mouse CD64 (BioLegend, 139304), PE anti-mouse MerTK (BioLegend, 151506), FITC anti-mouse F4/80 (BioLegend, 123108), FITC anti-mouse CD163, PE anti-mouse CD206 (BioLegend, 141707), allophycocyanin anti-mouse CD80 (BioLegend, 104713), allophycocyanin anti-mouse CD68 (BioLegend, 137007), PE anti-mouse CD11c (BioLegend, 117307), PE anti-human MerTK (R&D Systems, FAB8912P), PE anti-human CD68 (BioLegend, 333807), allophycocyanin anti-human CD16 (BioLegend, 302011), PE anti-human CD32 (BioLegend, 303205), allophycocyanin anti-human CD64 (BioLegend, 305013), allophycocyanin anti-human CD163 (BioLegend, 333610), allophycocyanin anti-human CD206 (BioLegend, 321110), allophycocyanin anti-human CD80 (BioLegend, 305219), PE anti-human CD86 (BioLegend, 305405), FITC annexin V (BioLegend, 640905), allophycocyanin anti-human/anti-mouse TREM2 (R&D Systems, FAB17291A).

Anti-MerTK mAbs were generated from New Zealand White rabbits immunized with recombinant murine MerTK protein or human MerTK protein using a protocol based on a previous report (43). Briefly, B cell clones were selected based on binding to recombinant MerTK proteins by ELISA, and to MerTK-expressing cells by FACS. Variable regions of the L chain and H chain of positive clones were amplified by PCR and cloned into expression vectors. Recombinant Abs were expressed in Expi293 cells and purified with protein A.

Thioglycolate-elicited mouse peritoneal macrophages from Cell Biologics (Chicago, IL) were used to measure the blocking activity of anti-murine MerTK Abs. Cells (3.0 × 10⁴ cells/well) were serum starved in RPMI medium for 3 h before preincubation with anti-MerTK Abs for 1 h. The cells were then treated with Gas6-Fc (10 μg/ml) for 20 min, lysed, and immediately stored at −80°C. The following day, the lysates were thawed to measure pAKT activity using the phospho-AKT-1 (Ser473) homogeneous time-resolved fluorescence kit (63ADK078PEG) from Cisbio (Bedford, MA) following the manufacturer’s recommendations. Phospho-AKT activity was quantified by calculating the ratio between the fluorescence signal emitted by the donor at 665 nm (Eu³⁺-cryptate) and the fluorescence signal emitted by the acceptor at 620 nm (d2) on an EnSight PRX500 reader (PerkinElmer). The signal generated by cells treated with Gas6-Fc (10 μg/ml) was designated as the maximum signal (100% activity), and this value was used as a reference to determine the activities (percentages) of anti-MerTK Abs.

Frozen human primary monocytes (CD14⁺) from PBMCs (70035) were obtained from STEMCELL Technologies (Tukwila, WA), and written informed consent was obtained from all subjects according to the vendor. Cynomolgus monkey bone marrow cells (IQB-MnBM1-5) were obtained from iQ Biosciences (Berkeley, CA). To generate M2-like macrophages, human primary monocytes and cynomolgus monkey bone marrow cells (10 million–15 million cells) were cultured in RPMI medium supplemented with 10% heat-inactivated FBS and recombinant human M-CSF (100 ng/ml) in a Nunc UpCell 10-cm dish from Thermo Scientific (174902). The cells were cultured in a 37°C incubator for 4 d to promote cell attachment and differentiation into macrophages. On day 5, the media were changed, M-CSF was replenished, and human IL-10 was added (50 ng/ml) for 3 additional days to promote differentiation into M2-like macrophages. To generate M1-like macrophages, human primary monocytes (10 million–15 million cells) were cultured in RPMI medium supplemented with 10% heat-inactivated FBS and recombinant human GM-CSF (50 ng/ml) in a Nunc UpCell 10-cm dish from Thermo Scientific (174902) for a total of 7 d, with media changed at day 4. On day 7, the cells were treated with 10 ng/ml of IFN-γ and 100 μg/ml of LPS to promote differentiation into M1-like macrophages. Once the cells had been fully differentiated into M1-like and M2-like macrophages, the Nunc UpCell 10-cm dish was brought to room temperature, and the cells were detached by washing with chilled PBS without using dissociation enzymes or physical scraping. Fully differentiated macrophages were then seeded at a density of 2.0 × 10⁴ cells/well on a 96-well, low-evaporation Nunclon Delta Surface plate from Thermo Scientific (143761) in RPMI medium supplemented with 10% heat-inactivated FBS, and they were used to conduct efferocytosis, phagocytosis assays, and other experiments.

Gene expression profiling of in vitro differentiated human M1- and M2-like macrophages was assayed with the QuantiGene Plex 2.0 kit (QP1013) from Thermo Fisher Scientific (Waltham, MA) using branched DNA signal amplification and multianalyte profiling beads (xMAP). The assay was performed following the manufacturer’s instructions, and gene expression was calculated as median fluorescence intensity using a FlexMap 3D System from Luminex (Austin, TX). Cytokine profiling of in vitro differentiated human M1- and M2-like macrophages was carried out using the 10-plex Proinflammatory Panel 1 (human) Kit (K15049D) from Meso Scale Discovery (Rockville, MD), following the manufacturer’s recommendations.

Phagocytic activity was measured using commercially available pHrodo (red) E. coli bioparticles from Incucyte (Sartorius, Bohemia, NY). Differentiated primary macrophages or J774A.1 cells (Genentech cell bank) seeded into 96-well plates were placed inside the Incucyte Zoom/S3 immediately after the addition of pHrodo (red) E. coli bioparticles (25 μg/well). Cell images were obtained in real time with the brightfield program and red laser settings set according to the manufacturer’s protocol. Images were collected every 30 min for a period of 24 h immediately after the addition of the pHrodo (red) E. coli bioparticles. Phagocytic activity was quantified as total red fluorescence intensity (Red Calibrated Unit [RCU] × µm²/Image) using the built-in image analysis tools in the Incucyte Zoom software.

Human Jurkat cells were cultured in RPMI medium supplemented with 10% FBS. Cells from an exponentially growing culture at a density of 1.0 × 10⁶ cells/ml were induced to undergo apoptosis by treatment with 1.0 µM of staurosporine for a period of 4 h. Apoptotic cells were then characterized by flow cytometry with staining of annexin V and propidium iodide. After resuspension in 2% FBS containing PBS at 1.0 × 10⁶ cells/ml, the apoptotic cells were labeled by pHrodo red succinimidyl ester (1.0 µM final concentration) in the dark at room temperature for 1 h. After labeling, the apoptotic cells were washed three times with PBS and resuspended in RPMI medium + 5% serum. These cells were then used in the efferocytosis assays.

The efferocytosis assay was carried out using pHrodo (red)-labeled Jurkat cells that served as “meal” for various phagocytic cells, including mouse and human cell lines (U937, J774A.1, etc.), mouse peritoneal macrophages, and differentiated primary human and cynomolgus monkey macrophages. Macrophages were seeded in a 96-well plate and treated with blocking Abs for 1 h. Then freshly prepared pHrodo-labeled apoptotic cells were added to the macrophages at a density ranging between 2.5 × 10⁵ and 7.5 × 10⁵cells/well. After the cell plates were placed inside the Incucyte system, cell images were obtained in real time using the brightfield program and a red laser, using the instrument settings recommended by the manufacturer’s protocol. Images were collected every 30 min for a period of 8 h (or up to 24 h). Efferocytosis activity was measured as total red fluorescence intensity (Red Calibrated Unit [RCU] × µm²/Image) using the built-in image analysis tools and algorithms built in the Incucyte system software. Macrophage viability was monitored by pretreating cells with 5.0 µM of Incucyte Caspase-3/7 Green (4440) from Sartorius and using the “object count metric” method provided by the Incucyte Basic Analysis Software.

To visualize the intracellular localization of engulfed pHrodo-labeled apoptotic Jurkat cells in macrophages, we obtained live-cell images using the ImageXpress Micro Confocal High-Content Imaging System (Molecular Devices, San Jose, CA). J774A.1 macrophage cells were seeded in a 96-well plate and treated with LysoTracker Green DND-26 (L7526) from Thermo Fisher Scientific at a final concentration of 250 nM by incubating at 37°C for 1.5 h. After incubation, cells were washed twice with PBS to remove excess LysoTracker probe and supplemented with fresh growth media. The macrophages were then challenged with freshly prepared pHrodo red–labeled apoptotic Jurkat cells at a ratio of 1:3 (macrophages to Jurkat cells). Image acquisition was carried out using a 20× Nikon CFI Plan Apo Λ objective lens fitted with FITC (LysoTracker Green) and Texas Red (pHrodo red) filters with adjusted focus, master gain, and pinhole for each of the experiments. Brightfield images were acquired using the same 20× objective. The excitation (ex)/emission (em) spectral peaks for LysoTracker probe are λex 504 nm, λem 511 nm and λex 566 nm, λem 590 nm for pHrodo red. After acquisition, the images were processed using ImageJ software (National Institutes of Health, Bethesda, MD).

While we were looking for a robust and quantitative efferocytosis assay to screen anti-MerTK functional Abs, a live-cell analysis system called Incucyte emerged to catch our attention. The Incucyte live-cell analysis system is a real-time imaging platform that is housed inside a tissue culture incubator, with an optics head that moves around stationary cell culture plates to automatically take images of the cells in real time. Recently, Kapellos et al. reported that the Incucyte system could be used to monitor phagocytosis activities of macrophages using pHrodo-labeled E. coli bioparticles as the phagocytic target (44). The amine-reactive pHrodo fluorogenic dye shows a dramatic increase in fluorescence as the pH of the environment becomes more acidic, making it a powerful tool to monitor the process of cellular engulfment of the labeled cargo. We rationalized that an efferocytosis assay could be set up similarly, using pHrodo-labeled apoptotic cells as the “meal.” First, we wanted to confirm the phagocytosis activity of the mouse macrophage cell line J774A.1 using E. coli bioparticles as the target (44). Briefly, we seeded J774A.1 cells (2.5 × 10⁴ cells/well) and provided them with 25 μg/well of a pHrodo red–labeled E. coli bioparticle meal. As shown in Fig. 1A, J744A.1 cells phagocytosed the E. coli bioparticles, which resulted in light emitted that was captured by the red fluorescence channel and quantified as total red object integrated intensity (the total sum of the objects’ fluorescence intensity in the image). We then prepared apoptotic Jurkat cells and labeled them with pHrodo red dye. After adding labeled Jurkat cells to mouse J744A.1 cells cultured in the 96-well plate, we set the Incucyte to immediately capture images. As shown in Fig. 1B and 1C, J744A.1 cells started to exhibit intracellular red fluorescence from 2 h after apoptotic Jurkat cells were added. The total red object integrated intensity increased at a very fast rate between 3 and 6 h, reached a plateau at 12 h and then started to decline (Fig. 1C). To confirm that the observed red fluorescence indicated the engulfment of apoptotic Jurkat cells by J774A.1 macrophages, we used ImageXpress, a confocal high-content imaging system, to visualize the intracellular localization of red fluorescence after feeding J744A.1 cells with pHrodo red–labeled apoptotic Jurkat cells. Meanwhile, we used LysoTracker (green) to label the lysosomes of J774A.1 macrophages. As shown in Fig. 1D, the red fluorescence (pHrodo-labeled apoptotic cells) only colocalized with the lysosome compartments of J774A.1 cells, stained with LysoTracker (green). These results indicate that the observed red fluorescence indeed showed the engulfment of apoptotic cells into lysosomes of macrophages. Therefore, J774A.1 cells are not only able to engulf bacteria but also capable of eating apoptotic cells (efferocytosis). Moreover, the efferocytosis can be monitored by the Incucyte imaging system, because the acidic condition in lysosomes/phagosomes of macrophages increases the fluorescence intensity of the pHrodo-labeled apoptotic meal.

To test if J774A.1 cells are suitable for screening anti-MerTK functional Abs that inhibit efferocytosis, we performed surface receptor analysis on J774A.1 cells by flow cytometry. The flow analysis results showed that J774A.1 cells do not express much MerTK, but instead express another efferocytosis receptor, Tyro3 (Fig. 1E), which belongs to the same tyrosine kinase receptor family as MerTK. Thus, even though we successfully set up the Incucyte system to monitor efferocytosis, J774A.1 may not be an appropriate cell line for screening or characterizing efferocytosis-blocking anti-MerTK Abs.

To look for suitable cells for discovery of efferocytosis-inhibiting anti-MerTK Abs for in vivo mouse studies, we examined mouse peritoneal macrophages isolated after thioglycolate elicitation. Detailed flow cytometric analyses showed that mouse peritoneal macrophages express the macrophage surface markers F4/80, CD11b, and CD68, as well as Fcγ receptors. Most important, mouse peritoneal macrophages express MerTK (Fig. 2A), which was our target. To study the efferocytosis activity of mouse peritoneal macrophages, we set up the Incucyte-based assay similarly to the way we did for the J774A.1 cell line. Indeed, we found that mouse peritoneal macrophages were able to uptake apoptotic Jurkat cells with kinetics similar to those of J774A.1 cells (Fig. 2B). Next, we tested if we could use peritoneal macrophages and the Incucyte platform to identify antagonistic anti-MerTK Abs. During our mouse MerTK Ab discovery campaign, we identified several anti-MerTK Abs that are capable of inhibiting ligand-induced MerTK signaling as measured by phosphorylation of AKT (Fig. 2C), a downstream component of the MerTK signaling pathway. Among them, a few anti-MerTK Abs showed potent dose-dependent inhibition of pAKT signal (Fig. 2D). In our efferocytosis experiments, two anti-MerTK Abs showed that robust blocking activities in peritoneal macrophages mediated efferocytosis, compared with isotype control Ab (Fig. 2E).

The anti-MerTK Abs that we identified using mouse peritoneal macrophages are specific to mouse MerTK, not cross-reactive to human MerTK. To discover anti-MerTK Abs for potential clinical development, we needed to establish an efferocytosis assay compatible with screening for human MerTK reactive Abs. We first characterized human monocytic/macrophage cell lines. Flow cytometric analysis showed that the common human monocytic cell lines (e.g., THP1 and U937) either do not express MerTK (THP1; data not shown) or express other efferocytosis receptors, such as Tyro3 (Fig. 3A), rendering them unsuitable for our screening. Tumor-associated macrophages are reported to originate from bone marrow or circulating classical monocytes (CD14⁺CD16⁻), then infiltrate into the tumor through CCL2/CCR2 signaling and differentiate in situ into a unique anti-inflammatory M2-like macrophage phenotype (45–48). Thus, we decided to differentiate CD14⁺ monocytes in vitro to see if we could get M2-like macrophages that resemble tumor-associated macrophages. Because IL-10 has often been detected in the tumor microenvironment of various types of cancer, we included IL-10 together with CSF1 (M-CSF) in our M2-like macrophage differentiation protocol. In parallel, we also differentiated M1-like macrophages using CSF2 (GM-CSF) in the presence of LPS and IFN-γ for comparison. Our in vitro differentiated macrophages revealed different expression profiles of efferocytosis receptors on their surfaces. As shown in Fig. 3A, M2 cells express a significant amount of MerTK, but not Axl or Tyro3, whereas M1 cells do not express MerTK and Tyro3, with only marginal expression of Axl. Detailed flow cytometric analyses indicated that M2 cells, similar to tumor-associated macrophages, express not only MerTK but also CD163 (Fig. 3B). Under our in vitro differentiation conditions, 99% of differentiated M2 cells showed CD163 and MerTK double positivity (Fig. 3B). Compared with M1 macrophages, M2 macrophages showed a more stretched morphology when differentiated in the cell culture dishes (Fig. 3B). To study the properties of differentiated macrophages, we also analyzed their gene expression as well as cytokine secretion profiles. The results showed that M1 macrophages exhibit upregulation of proinflammatory cytokine genes and secrete significantly more inflammatory cytokines (TNF-α, IFN-γ, IL-1β, IL-8, IL-12) after LPS treatment compared with M2 cells (Fig. 3C, 3D). Consistent with tumor-associated macrophages, M2 macrophages showed downregulated inflammatory cytokines and upregulated anti-inflammatory cytokines (such as IL-10) at both gene and protein levels. Last, we tested the efferocytosis activity of the differentiated M1 and M2 macrophages using the Incucyte live-cell analysis system. The results demonstrate that M1 macrophages have moderate efferocytosis activity, whereas M2 macrophages possess significantly higher efferocytosis capacity with a much faster engulfing rate (Fig. 3E). Taken together, our results demonstrated that the in vitro differentiated M2 macrophages mostly resemble tumor-associated macrophages, both phenotypically and functionally. Thus, we have established a physiologically relevant primary cell type (in vitro differentiated M2 macrophages) to screen and characterize efferocytosis-inhibiting anti-MerTK Abs for potential clinical development.

After identifying the appropriate cells, we envisioned that optimizing the Incucyte-based efferocytosis assay would be beneficial for the discovery of human-specific anti-MerTK functional Abs. First, we compared different M2 cell numbers as well as M2 cell/apoptotic cell ratio. We found that 2.0–3.0 × 10⁴ M2 cells per well of a 96-well plate and a 1:20 to 1:30 macrophage/apoptotic cell ratio generated robust efferocytosis fluorescence signals (Fig. 4A). As previously observed in mouse macrophages and cell lines, primary human macrophages are able to sustain a fast rate of efferocytosis for a few hours before slowing down (Fig. 4A). Thus, we focused on this period (between 3 and 6 h) when testing the inhibitory activity of a commercial polyclonal anti-MerTK Ab AF891 (R&D Systems). We found that anti-MerTK AF891 was able to inhibit M2 macrophage–mediated efferocytosis in a dose-dependent manner (Fig. 4B, gray). Anti-MerTK AF891 exhibited an inhibitory potency with IC₅₀ of ∼0.1 µg/ml but reached a maximum inhibitory effect of only 60% at the highest concentration.

We realized that cell debris is normally generated during the preparation of apoptotic cells and pHrodo labeling. Some cell debris, specifically that which does not contain PtdSer, will be taken in by macrophages through nonspecific endocytosis instead of efferocytosis. This cell debris, if not removed, could have complicated our image-based efferocytosis readout. Thus, we decided to add a slow-speed centrifugation step (100 × g/10 min, three times), after pHrodo labeling of apoptotic Jurkat cells. The slow-speed centrifugation selectively precipitated the larger apoptotic cells, leaving the smaller cell debris in the supernatant, resulting in the remaining apoptotic meal with higher forward scattering in flow cytometry (Fig. 4B, inlets). We also found that the “clean” apoptotic meal showed a higher percentage of annexin V binding by flow cytometry (data not shown). Because annexin V specifically binds to PtdSer, the extra slow-speed centrifugation step resulted in improved apoptotic meals with increased efferocytosis specificity, which can be more efficiently inhibited by anti-MerTK Abs (Fig. 4B, white). During image analysis, we noticed that the pHrodo red–labeled apoptotic cells were emitting a low intensity of autofluorescence that could be captured by the default image analysis algorithm (Fig. 4C, inlets, top panel), artificially increasing the assay background. Thus, we generated a more sophisticated image analysis algorithm using the top-hat method correction with a size threshold mask (125 nm diameter) to minimize autofluorescence (Fig. 4C, inlets, bottom panel). Implementation of this optimized algorithm minimizes signal background and eliminates edge bleed in images compared with the default settings, which further improved the inhibition assay (Fig. 4C, green). Even though MerTK plays a major role in efferocytosis, we noticed that blocking MerTK (by functionally inhibitory anti-MerTK Ab) alone could not result in 100% inhibition of efferocytosis as cytochalasin D, an inhibitor of cellular actin polymerization (Fig. 4D). Nevertheless, this image-based efferocytosis assay is still very useful for screening functional anti-MerTK Abs.

Next, we tested the robustness of our efferocytosis assay for high-throughput screening. We adopted an interlaced control placement scheme on the 96-well plate with the maximum and negative controls placed in every other column. Although the maximum inhibition control (anti-MerTK Ab AF89) showed consistent blocking of efferocytosis across the plates, we sometimes noticed big variations of the negative control readouts (Fig. 4E, top panel). After investigation, we found that even though the majority of the plate wells had consistent cell seeding across the whole well, some of the images showed uneven localized macrophages within the image-capturing area (Fig. 4E, inlets). The area for each image represents less than 15% of the whole well; thus, it is possible that random seeding variations can have a large impact on this assay. The impact of uneven seeding can be minimized by increasing the number of images captured per well (up to four images per 96-well in the Incucyte Zoom). However, this approach is time consuming because it takes 45 min to read a full 96-well plate. Thus, we decided to normalize the fluorescence signal by the number of macrophages in each image, using the phase-contrast image analysis algorithm. By this approach, the assay showed improved well-to-well consistency with great robustness (Z′ = 0.7) (Fig. 4E, bottom panel). Thus far, we have established a quantitative and robust efferocytosis assay for screening anti-MerTK human Abs.

The in vitro differentiation of monocytes into M2 macrophages, however, is a time-consuming step. Thus, we compared the performance of frozen macrophages differentiated previously and freshly differentiated macrophages to make future screening activities more convenient. As shown in Fig. 4F, the inhibition curves of one anti-MerTK Ab using fresh and frozen M2 cells (differentiated from the same donor) were nearly identical, suggesting that we can use frozen M2 cells for conveniently screening anti-MerTK functional Abs.

A representative Ab screening showed different anti-MerTK Abs exhibiting a range of potencies as well as different maximum inhibition of efferocytosis (Fig. 5A). Moreover, assay results using M2 macrophages that were differentiated from different donors’ CD14⁺ monocytes showed the same rank of inhibitory potency for a subset of clones in efferocytosis assays performed at different times (Fig. 5B, 5C). These results indicate the great reproducibility and robustness of our established efferocytosis assay. Out of many efferocytosis-blocking anti-MerTK Abs, one clone (blue curve in Fig. 5A, 5C) showed the most potency in terms of inhibiting MerTK-mediated efferocytosis.

When selecting a clinical drug candidate for later development, relevant animal species are usually chosen for preclinical pharmacokinetics (PK) study and toxicology study. Nonhuman primates are close to humans in their genetics and anatomy as well as in the way they respond to drugs. Historically, PK studies in monkeys provided useful information to determine dose and dosing regimens in clinical trials. Furthermore, adverse drug reactions are often similar between monkeys and humans. For these reasons, monkeys (usually the cynomolgus monkey) are the most relevant species and are broadly used to conduct preclinical PK studies and test the safety of new Ab drugs. Because our anti-MerTK clinical candidate is not cross-reactive to mouse MerTK, testing its activity in a monkey macrophage-mediated efferocytosis assay became important. It would allow us to investigate if there is a mechanism of action of anti-MerTK mAb in cynomolgus monkeys that is similar to that in humans. Also, comparison of anti-MerTK IC₅₀ numbers derived from human and monkey macrophage-mediated efferocytosis assays would enable us to better understand the safety findings from preclinical studies and future clinical trials. Thus, we decided to establish an efferocytosis assay using primary monkey macrophages differentiated in vitro, similar to the way we did for human macrophages. Primary bone marrow cells from cynomolgus monkeys were put in culture in the presence of human CSF1 and IL-10. After in vitro differentiation for 7 d, the attached cells showed stretched morphology similar to that observed in human M2 macrophages (Fig. 6A). Flow cytometric analysis showed that 98% of differentiated cells were MerTK and CD163 double positive (Fig. 6A). We then used these differentiated monkey macrophages to set up the image-based efferocytosis assay. As shown in Fig. 6B, monkey macrophages exhibit efferocytosis activity with kinetics similar to those observed from human M2 cells. More important, anti-MerTK showed inhibitory activity in monkey macrophage-mediated efferocytosis with potency comparable to that in the human M2 cell–mediated efferocytosis (Fig. 6C). Thus, we established efferocytosis with monkey macrophages, and it could serve as a bridge assay for the development of an anti-MerTK clinical candidate.

We have established a high-throughput quantitative efferocytosis assay using an image-based real-time cell analysis system (Incucyte). Compared with other approaches to studying efferocytosis, our method has several notable advantages, including high specificity and reproducibility, low cellular disturbance, and easier setup. We have also identified a special type of cell culture plate (Nunc UpCell; see Materials and Methods) for differentiating and culturing macrophages, which allowed us to conveniently detach adherent macrophages without harsh enzymatic digestion or other abrasive mechanical procedures. This helped to avoid any damage and unintentional activation of macrophages before each experiment. The Incucyte system is conveniently positioned inside a cell culture incubator, and, unlike other imaging systems, the cell plates are fixed while the optic head automatically moves around to capture live images. Moreover, because there is no cell fixation step, the cells are not perturbed when the process of efferocytosis/phagocytosis is monitored. Thus, our method allows an unbiased measurement of efferocytosis with high fidelity. Together with the built-in automation and multiple (six to eight) positions that are compatible with 96-well plates, the Incucyte system is suitable for high-throughput screening to find agonists and antagonists of efferocytosis, as evidenced by our successful discovery of potent efferocytosis blocking anti-MerTK Abs. When we dosed MC-38 tumor-bearing mice with one of the murine-specific anti-MerTK mAbs identified by our efferocytosis assay, the results showed an increasing number of apoptotic cells in tumors as well as increased levels of circulating tumor DNA in the plasma (49). We found that Ab blockade of MerTK resulted in robust anti-tumor responses when combined with anti-PD-L1 that by itself only exhibited modest anti-tumor activity (49). Altogether, the efferocytosis assay using the Incucyte platform enabled us to discover a potent anti-MerTK Ab that blocks macrophage-mediated efferocytosis both in vitro and in vivo.

In addition, the current method allows hands-off experiments and continuous data acquisition, which enabled us to conveniently study the kinetics of phagocytosis/efferocytosis (Fig. 1A, C). We noticed that J774A.1 macrophages engulfed the bioparticles robustly immediately after the addition of pHrodo-labeled E. coli. On one hand, the fluorescence reached a plateau around 20 h and started to decrease until the bioparticles were exhausted. On the other hand, mouse macrophages (both cell line J774A.1 [Fig. 1C] and mouse peritoneal macrophages [Fig. 2B]) started to engulf apoptotic cells around 2 h after the addition of the apoptotic meal. The robust eating phase was between 3 and 6 h. The fluorescence intensity reached a plateau at 12 h, then started to decrease until all meal was exhausted. The delay of engulfment of apoptotic cells (compared with E. coli bioparticles) is probably due to the longer time required for macrophages to sense and recognize the apoptotic meals and to initiate the receptor-mediated efferocytosis. A similar delay was also observed for efferocytosis mediated by both human and monkey primary macrophages (Figs. 3E, 4A, and 6B). When we monitored the viability of macrophages (pretreating J774A.1 cells with caspase-3/7 dye and imaging by Incucyte), we found that, after a robust engulfing phase, some macrophages became apoptotic and eventually died (similar results observed for primary macrophages; data not shown). Meanwhile, we noticed that most live macrophages showed a decline in the rate of eating after the robust engulfing phase. These observations are consistent with other reports, which indicated that the phagocytic activity of macrophages was substantially impaired when reaching the point of saturation (or “exhaustion”) after a fast engulfment phase (50, 51). These results suggest that continuous efferocytosis/phagocytosis could cause exhaustion (or toxicity) for macrophages. Thus, we focused our analysis on the robust engulfing phase when screening for efferocytosis/phagocytosis agonists/antagonists. Nevertheless, further investigations are required to understand the mechanism of the exhaustion/toxicity caused by efferocytosis/phagocytosis.

It is noteworthy that cell debris, specifically that which does not contain PtdSer, will be taken in by macrophages through nonspecific endocytosis instead of efferocytosis. This cell debris, if not removed from the apoptotic meal, could complicate our image-based efferocytosis readout. Because cell debris is inevitably generated during the process of preparation of apoptotic cells and pHrodo labeling, we introduced an extra slow-speed centrifugation step to clean up the apoptotic cells. This procedure enabled us to get rid of the majority of cell debris from the apoptotic meal, resulting in a more specific efferocytosis readout. Still, we noticed that complete efferocytosis inhibition could not be achieved by anti-MerTK blocking Abs even at high concentrations (Fig. 4D). One possible explanation is that some cell debris (with pHrodo label) coprecipitates with apoptotic cells after the slow-speed centrifugation (as shown in Fig. 4B, inlets). A portion of the coprecipitated cell debris may still contain PtdSer, which would be recognized and taken up by macrophages through efferocytosis. Other cell debris that does not contain PtdSer would be taken in by macrophages through nonspecific endocytosis, which cannot be differentiated from efferocytosis by imaging. This is a limitation of all current efferocytosis methods. However, by cleaning the apoptotic meals, we significantly improved the efferocytosis specificity in our assay. The nonspecific endocytosis is not MerTK mediated; therefore, it cannot be inhibited by anti-MerTK Abs. This nonspecific endocytosis may contribute to the maximum inhibition difference that we observed between anti-MerTK and cytochalasin D (Fig. 4D). Another possible contributing factor is the existence of other efferocytosis receptor(s) that anti-MerTK Abs cannot inhibit. In fact, even though MerTK plays a major role in macrophage-mediated efferocytosis (7), other receptors expressed in macrophages were reported to also mediate efferocytosis (52). For example, TREM2 has been found to be expressed in macrophages/microglial cells. It recognizes and directly binds to PtdSer, the “eat me” signal on apoptotic cells, and thus mediates efferocytosis (53, 54). We analyzed in vitro differentiated human and monkey M2 macrophages and found that those cells were also TREM2 positive by flow cytometry (data not shown). Therefore, MerTK may be the predominant receptor in macrophage-mediated efferocytosis because blocking MerTK inhibits this efferocytosis significantly. However, anti-MerTK inhibitory Ab alone could not achieve 100% inhibition of efferocytosis, likely due to the existence of other efferocytosis receptors (i.e., TREM2). Of course, further mechanistic studies would be required if we wanted to fully understand the observations.

While we were preparing this report, the Incucyte S3 model that can accommodate 384-well plates became available, enabling automated screening with even higher throughput capacity. As a real-time, quantitative, and robust method, this image-based efferocytosis assay should have broad applications in studying the kinetics and molecular mechanisms of efferocytosis/phagocytosis, as well as identifying specific efferocytosis/phagocytosis modulators for potential therapeutic use.

The authors have no financial conflicts of interest.