Overexpression of MERTK Receptor Tyrosine Kinase in Epithelial Cancer Cells Drives Efferocytosis in a Gain-of-Function Capacity*

Background: Overexpression of MERTK tyrosine kinase is observed in many human cancers. Results: MERTK has a potent gain-of-function capacity to stimulate efferocytosis in epithelial cells. Conclusion: When overexpressed, MERTK acts as a preeminent efferocytosis receptor that is targetable by soluble Ig-domain containing TAM receptors. Significance: Our findings provide new mechanistic insight into how MERTK may impinge on tumor progression by enhancing efferocytosis. MERTK, a member of the TAM (TYRO3, AXL, and MERTK) receptor tyrosine kinases, has complex and diverse roles in cell biology. On the one hand, knock-out of MERTK results in age-dependent autoimmunity characterized by failure of apoptotic cell clearance, while on the other, MERTK overexpression in cancer drives classical oncogene pathways leading to cell transformation. To better understand the interplay between cell transformation and efferocytosis, we stably expressed MERTK in human MCF10A cells, a non-tumorigenic breast epithelial cell line devoid of endogenous MERTK. While stable expression of MERTK in MCF10A resulted in enhanced motility and AKT-mediated chemoprotection, MERTK-10A cells did not form stable colonies in soft agar, or enhance proliferation compared with parental MCF10A cells. Concomitant to chemoresistance, MERTK also stimulated efferocytosis in a gain-of-function capacity. However, unlike AXL, MERTK activation was highly dependent on apoptotic cells, suggesting MERTK may preferentially interface with phosphatidylserine. Consistent with this idea, knockdown of MERTK in breast cancer cells MDA-MB 231 reduced efferocytosis, while transient or stable expression of MERTK stimulated apoptotic cell clearance in all cell lines tested. Moreover, human breast cancer cells with higher endogenous MERTK showed higher levels of efferocytosis that could be blocked by soluble TAM receptors. Finally, through MERTK, apoptotic cells induced PD-L1 expression, an immune checkpoint blockade, suggesting that cancer cells may adopt MERTK-driven efferocytosis as an immune suppression mechanism for their advantage. These data collectively identify MERTK as a significant link between cancer progression and efferocytosis, and a potentially unrealized tumor-promoting event when MERTK is overexpressed in epithelial cells.

phorylation of STAT1 acting as negative regulators of pro-inflammatory TLR-signaling and promoting the induction of suppressors of cytokine signaling SOCS-1 and SOCS-3 expression, which may in part explain why TAM(Ϫ/Ϫ) mice have elevated inflammatory cytokines such as IL-6 and TNF-␣ especially upon the activation of TLRs (35)(36)(37).
MERTK up-regulation induced by ACs is limited to early involution while expression is negatively regulated in the mammary gland under physiological conditions (26). However, each of the three TAMs has been implicated in human cancers by virtue of their pathological overexpression and their ability to activate oncogenic and survival signaling pathways (4, 7, 8, 38 -42). Previous studies have shown negative regulation of AXL and MERTK expression by miR-34, miR-199a/b, and miR-126, which are down-regulated in a variety of cancers (43,44). Moreover, in a majority of primary breast carcinomas from patients who relapse, the decreased expression of miR-335, an inhibitor of cancer metastasis predicted to target the 3Ј-UTR of MERTK, is potentially explaining the up-regulation of MERTK in these predisposed tumors (45). Clinically, in independent studies, AXL and MERTK overexpression in solid cancers is associated with more aggressive and invasive malignant behavior, and associated with poorer patient survival outcomes (7, 16, 42, 46 -50). In addition, GAS6 is also concomitantly overexpressed in human cancers, implying that, besides receptor overexpression, TAMs participate in autocrine circuits by overexpressing both receptors and ligands (46,51,52).
Given the versatility of MERTK signaling in the mammary gland, and the seemingly contradictory roles that MERTK play in oncogenic and efferocytic signaling pathways, we set out to explore the biology of MERTK overexpression in mammary epithelial cells. Here we show that MERTK has pleiotropic biological functions in mammary cells. Overexpression of MERTK promotes chemoresistance of epithelial cells and migration, as well as increases their efferocytic activity leading to the induction of Programmed death-ligand 1 (PD-L1), a known blockade of immune checkpoints in cancer. Overall, these data suggest that MERTK is sufficient to be a stand-alone receptor driving epithelial efferocytosis in mammary cells, and, when overexpressed, it acts as a preeminent engulfment receptor that is targetable by soluble TAM receptors.
Human GAS6 Conditioned Medium-3 to 5 g of pSecTag2human GAS6 plasmid (a kind gift from Dr. Peter Carmeliet) and 12 to 15 l of X-tremeGENE mixture was used to transfect into a 10-cm plate of 293TN cells at 70% confluence in the presence of 10 g/ml vitamin K1 (Phytonadione injectable emulsion from Hospira). After overnight incubation, the same 293TN cells were transfected again in the same manner. 6 h post-second transfection, the complete medium with 10% FBS was then changed into 0% serum in the presence of vitamin K1. About 36 h later, the GAS6 conditioned medium was collected. For mock-conditioned medium, GFP was used in place of GAS6 plasmid.
Proliferation Assay-MCF10A-pMSCV and MCF10A-MERTK stable cell lines were starved overnight in medium containing 0.5% HS and full MCF10A supplements. Cells were collected in starvation serum medium (0.5% HS and full MCF10A supplements), after which 50 l of cell suspension containing 2.5 ϫ 10 3 cells was added to 100 l of GAS6 conditioned medium or normal growth medium containing 5% HS and full MCF10A supplements. Real-time proliferation was monitored every 5 h for 160 h by xCELLigence system (ACEA Biosciences).
Colony Formation Assay-In each 35-mm plate, 1.5 ml of 0.5% base agar layer in human GAS6 conditioned medium plus 1% serum (and MCF10A supplements described above for all MCF10A cells). About 5000 cells (MCF10A-pMSCV, MCF10A-MERTK, or MDA-MB 231) were then embedded in 1.5 ml of 0.35% top agar layer with GAS6-conditioned medium plus 1% serum (and other MCF10A supplements as needed). Approximately 0.75 ml of GAS6 conditioned medium plus 1% serum (plus other MCF10A supplements when necessary) was added to each plate, and medium was changed every 3 days. After 21 days, each plate was stained with 0.75 ml of 0.005% crystal violet for 4 h, and plates were photographed, and colonies were counted.
Cell Death Assay-MCF10A-MERTK and MCF10A-pMSCV were first starved for 6 h and then treated with 50 M of camptothecin (Sigma) or 20 M of 5-Fluorouracil (5-FU) (Sigma) in the human GAS6 conditioned medium with or without 50 M of LY-294002 (Cell Signaling). After 24 h, cells were stained with FITC-conjugated Annexin V and propidium iodine (PI) according to the manufacturer's protocol (BioLegend), and cell death (Annexin V and/or PI positive) was evaluated by flow cytometry.
Acini Formation Assay-MCF10A-pMSV and MCF10A-MERTK cells were grown in three-dimensional cell cultures containing growth factor reduced matrigel (BD Biosciences) as previously described (54). Briefly, about 5000 cells were overlaid in each well of an eight-chambered slide that had been pre-coated with 40 l of matrigel, and the three-dimensional cell cultures were cultured for 12 days, with assay medium changed every 4 days. The cells were then fixed in 3.7% paraformaldehyde, permeabilized with 0.5% Triton X-100 in PBS, and stained with Alexa Fluor 594 phalloidin (Invitrogen), DAPI (Sigma), and cleaved caspase-3 (Cell Signaling). Confocal analyses were performed with Nikon A1R confocal microscope.
Migration Assay-MCF10A-MERTK and MCF10A-pMSCV were starved for 6 h and then 5 ϫ 10 4 cells were added to the top chamber of the CIM-Plate 16 pre-coated with 10 g/ml fibronectin. Medium in the top chamber containing 3% HS and one-fifth of the MCF10A supplements described above. On the bottom chamber, human GAS6 conditioned medium in 1% serum or mock-conditioned medium in 1% serum was used to induce the migration of those cells. Real time migration was monitored by xCELLigence system (ACEA Biosciences) every 15 min for 10 h. Each experiment was done in four replicates.
Primary Mammary Epithelial Cell (PMEC) Isolation-Six to eight-week-old female C57BL/6 mice, wild type or knock-out Mertk(Ϫ/Ϫ), were euthanized after which mammary glands were collected into PBS with penicillin and streptomycin on ice. Mammary tissues were then transferred into sterilized 10-cm glass Petri dish and minced with two scalpels. Minced tissues were digested with 2 mg/ml collagenase A (Roche) and 0.02g/ml hyaluronidase (Sigma) in DMEM medium with 5% FBS and penicillin and streptomycin for 6 to 8 h at 37°C on a shaker. Dissociated tissue was then washed three times with DMEM plus 5% FBS, and pellets highly enriched in epithelial organoids were collected into 5 ml of 0.25% trypsin-EDTA. Pipetting was done to break up the cells, and then cells were washed with DMEM medium plus 5% FBS for three times. The pellet was resuspended in 2 ml of 5 mg/ml Dispase (Roche) and 200 l of 1 mg/ml DNase I (Roche), washed with DMEM plus 5% FBS, and filtered through a 40-m cell strainer. Collected cells were culture in DMEM/F12 medium containing 5% FBS, 5 g/ml insulin (Sigma), 1 g/ml hydrocortisone (Sigma), and 5 ng/ml EGF (Invitrogen). Cells were allowed to recover for at least 2 days before any further experiment.
Efferocytosis-CEM or MCF10A were used as apoptotic cells (ACs). When phagocytes were labeled green with PKH67 (Sigma) dye or GFP from mouse Mertk bicistronic vectors or CellTracker Green CFDA (Invitrogen), ACs were labeled red with PKH26 dye (Sigma) according to the manufacturer's protocol. However, if phagocytes were labeled PKH26-red, ACs would be labeled PKH67-green. Apoptosis was induced by irradiation of CEM cells with 25mJ/cm 2 ultraviolet B (cells were then incubated at 37°C for 6 -8 h before co-incubation with phagocytes) or by starvation of MCF10A in medium without HS and other supplements for 24 to 48 h. Apoptosis was verified by Annexin V conjugated FITC and PI staining according to the manufacturer's protocol (BioLegend). Phagocytes were starved for 6 h and co-cultured with ACs at the ratio of 1:5 or 1:10 for 1 to 5 h depending on the experiment in the presence of 10% serum or GAS6 conditioned medium. After co-incubation, cells were washed three times with 10 mM EDTA to minimize binding of ACs to phagocytes, fixed with 1% paraformaldehyde, and analyzed by flow cytometry, Nikon A1R Confocal Microscope System, or Amnis Image Stream. For efferocytosis inhibition assay using TAM soluble receptors, efferocytosis was carried on as described above and ACs were mixed with soluble receptors for 6 h prior to co-incubation of ACs and phagocytes.
Immunohistochemistry-12 breast tumor adjacent normal tissue samples (normal) and 34 Grade 2 moderately differentiated breast tumor samples were collected. Paraffin-embedded breast cancer tissue sections (4 mm) were deparaffinized in xylene, and rehydrated in graded solutions of ethanol. Antigen retrieval was done by heating the sections in 0.01 M sodium citrate buffer (pH 6.0) at 100°C for 15 min. The sections were then incubated in 3% hydrogen peroxide to block the endogenous peroxidase activity at room temperature for 10 min. After blocking with 5% BSA for 20 min at room temperature, the sections were incubated with the rabbit monoclonal anti-MERTK (1:100, Cell Signaling Technology, #4319) overnight at 4°C. The slides were then incubated with biotin labeled goat anti-rabbit IgG for 20 min at room temperature and further incubated with streptavidin peroxidase solution (SABC, SA1022, Boster Biotechnology, Wuhan, China). The staining was visualized by reaction with DAB (AR1022, Boster Biotechnology, Wuhan, China). Slides were counterstained with hematoxylin, washed, dehydrated, and mounted. Five random microscopic fields per slide were photographed using a Nikon Light Microscope. Positive staining was shown in brown color.
Soluble TAM Receptor Production-First, cDNAs corresponding to human Ig1-Ig2 domains, of sTYRO3 (residues 42-223), sAXL (residues 33-220), and sMERTK (residues 95-285) were each cloned into baculovirus transfer vector pIEx (EMD Millipore). Transfection and virus amplification was then performed with Bacmagic-2 (EMD Millipore) in suspension cultures of Sf9 insect cells, after which suspension cultures of insect High Five cells grown in Express Five SFM (Invitrogen) to a density of 2 ϫ 10 6 cells/ml were infected at a multiplicity of infection of ϳ10. About 72 h post-infection, cells were collected in 25 mM Hepes pH 7.5, 250 mM NaCl, 500 mM L-Arg, 10% glycerol, and lysed by sonication (Sonic Dismembrator 500, Fisher Scientific). Soluble TAM were then recovered from cell lysate by affinity chromatography using Ni-IDA resin (Clontech), and further purified by ion-exchange chromatography over a Mono Q 5/50 GL column (GE Healthcare) followed by size exclusion chromatography over a HiLoad 16/60 Superdex 200 prep grade column (GE Healthcare) into a final buffer condition of 25 mM HEPES pH 7.5 with 200 mM NaCl.
Soluble TAM Receptor Binding Assay-Human GAS6-conditioned medium or 100 nM human Protein S (Hematologic Technologies) was incubated with 200 nM of individual purified soluble receptors sTYRO3, sAXL, or sMERTK overnight at 4°C. Those complexes of individual soluble receptors and the GAS6 or Protein S were then pulled down using Ni-NTA-agarose according to the manufacturer's protocol (Qiagen). GAS6 or Protein S alone without any soluble receptors was used as a control (mock). Those pulled down complexes were then resolved in SDS-PAGE and immunoblotted with GAS6 and His-Tag antibodies.
Generation of Chimeric Fc-Mertk Receptor Construct-DNA fragments encoded intracellular domain (ICDs) of mouse Mertk (500 -995) were amplified by PCR using sequence specific primers flanked with restriction enzyme sites at the ends. Statistical Analysis-Statistical analysis was done by GraphPad Prism. Descriptive statistics for quantitative variables were summarized using mean Ϯ S.D. Differences between groups were tested by t test or one-way ANOVA followed by Tukey post-hoc test. p values by t test or Tukey post-hoc test are shown, and p Ͻ 0.05 is considered as significant.

RESULTS
Homozygous deletion of MERTK results in systemic autoimmunity from the failure to clear ACs by myeloid cells in peripheral tissues as well as defects in the post-partum mammary gland involution by failure to clear ACs by mammary epithelial cells (5,7,21,23,26). Given the strong phenotype of MERTK deficiency in primary mammary epithelial cells, in combination with the fact that MERTK is frequently overexpressed in mammary carcinoma (44 -46), we set out to study the effects of MERTK when overexpressed in mammary epithelial cells. First, MERTK was stably expressed by retroviral transduction in MCF10A cells, a non-transformed immortalized mammary epithelial cell lines devoid of endogenous MERTK (Fig. 1A). Despite the fact that MERTK was stably expressed on the cell surface of MCF10A, these cells did not form colonies in soft agar as a marker of cell transformation by primary oncogenes (Fig. 1B), nor did they proliferate faster than the control MCF10A-pMSCV cells (Fig. 1C). Also, MCF10A-MERTK and parental cells both formed hollow luminal structures in a threedimensional acini forming assay where the luminal cells died by factor-deprivation induced anoikis (Fig. 1D). In contrast, MCF10A-MERTK cells were measured to have enhanced motility compared with control MCF10-pMSCV cells toward a serum gradient, and this motility was increased by the addition of GAS6 into the chemotaxis gradient ( Fig. 2A). In addition, MERTK protected MCF10A from camptothecin and 5-flurouracil (5-FU) induced cell death. Annexin V and propidium iodine staining revealed that ϳ45 and 50% of control MCF-10A-pMSCV cells was dying while only 23 and 24% of MCF-10A-MERTK cells underwent cell death when treated with 50 M camptothecin and 20 M 5-FU, respectively (Fig. 2, B and C). Since chemoresistance has been linked to the alteration of AKT signaling, we interrogated AKT activation downstream of MERTK signaling. When MERTK was activated by its classical ligand GAS6, AKT phosphorylation was strongly elevated (Fig. 2D) while ERK1/2 phosphorylation was slightly increased (data not shown). To confirm that AKT mediated the chemoresistance in MCF10A-MERTK, PI3K/AKT inhibitor, LY-294002, was employed and the chemoprotective effect of MERTK on MCF10A was indeed reversible (Fig. 2, B and C). Taken together, these data suggest that overexpression of MERTK in mammary epithelial cells could be an advantage for cancer cells by enhancing cell motility and promoting chemoresistance through AKT.
We next explored the capacity of MCF10A-MERTK cells to promote the clearance of ACs, a process termed efferocytosis in recent years. Previous results have shown that MERTK deficiency results in the inability of mammary epithelial cells to clear ACs in the involuting mammary glands, suggesting that MERTK has a preeminent role in efferocytosis during involution (26). Consistent with these findings, primary mammary epithelial cells (PMECs) from Mertk(Ϫ/Ϫ) mice had only ϳ50% of the efferocytic capacity as normal PMECs (Fig. 3A), while PMECs from Axl(Ϫ/Ϫ) mice were spared such an effect (data not shown). In addition, stable MERTK expression in MCF10A-MERTK cells described above promoted over 2-fold difference in its efferocytosis capacity when compared with control MCF10A-pMSCV cells by flow cytometry analysis, demonstrating that MERTK could act as a gain-of-function receptor for efferocytosis when overexpressed in epithelial cells (Fig. 3, B and  C). Indeed, confocal imaging showed that internalized green fragments of apoptotic CEM cells were readily engulfed by red MCF10A-MERTK while, in contrast, it was quite rare to find efferocytosis by MCF10A-pMSCV cells (Fig. 3D).
To ascertain whether enhanced MERTK-inducible efferocytosis in stable MCF10A-MERTK cells could be recapitulated by transient overexpression to minimize compensatory up-regulation of co-receptors, MERTK was introduced into MCF10A by means of pMSCV retroviral vectors and efferocytosis was again increased 3-fold as compared with pMSCV control cells (Fig. 4A). To further demonstrate the role of MERTK as a preeminent efferocytic receptor, primary immortalized mouse mammary epithelial cells 21 or cervical cancer cells HeLa were transiently transfected with a bi-cistronic vector expressing Mertk and GFP. When 21 and HeLa cells were gated for GFP as a marker for Mertk co-expression by bi-cistronic vectors, efferocytosis was increased 3-and 5-fold, respectively, as compared with GFP control 21 and HeLa cells (Fig. 4, B and C). These data suggested that MERTK was capable of immediately driving efferocytosis as a stand-alone receptor when ectopically overexpressed in different epithelial cells, including non-trans-formed mammary cells and epithelial cells transformed by other oncogenes.
To further implicate the relevance of MERTK directed efferocytosis in mammary epithelial cancer, we investigated the expression status of MERTK in breast cancer. Of 34 breast carcinoma and 12 normal breast samples that we analyzed, 24 of the cancer samples (70%) had highly elevated level of MERTK while all normal samples were below detection. Fig. 5A shows representative samples of breast tumor and normal adjacent tissues immunohistochemically stained with MERTK antibodies. Consistent with the tumor expression data, in several breast cancer cell lines that we surveyed including MCF-7 and MDA-MB-231, MERTK was overexpressed as compared with the non-transformed breast cells MCF10A reported above (Fig.  5B). In addition, while examining the association between endogenous MERTK expression and efferocytic capacity, we found that both breast cancer cell lines MCF-7 and MDA-MB-231 had 2 or 4-fold higher than MCF10A in their efferocytic capacity (Fig. 5C), while the most invasive cancer cell line MDA-MB-231 (with the highest level of endogenous MERTK) had the highest level of efferocytosis, which approached that observed in THP-1 macrophages, a professional phagocyte cell line (Fig. 5C). To confirm that MERTK indeed drove efferocytosis in MDA-MB 231 cells that expressed the highest level of MERTK among tested epithelial cells, we employed shRNA viral transduction to knockdown MERTK in these cells. As a result, the efferocytosis capacity of MERTK knockdown cells was reduced indicating that MERTK facilitated efferocytosis in highly invasive breast cancer cells (Fig. 5D). To explore whether the efferocytosis activity in high-MERTK expressing breast carcinoma cells could be targeted by inhibitors, we employed purified soluble human TAM receptors containing Ig1-Ig2 domain to selectively inhibit MERTK. As shown in Fig. 6A, we used soluble recombinant TAM ectodomains (sTYRO3, sAXL, and sMERTK) comprising the tandem extracellular Ig1-Ig2 repeats as decoys for ligands of MERTK. Initially, when we assessed the ability of the sTAMs to bind GAS6 in vitro, sAXL bound much better to recombinant GAS6, while sMERTK and sTYRO3 bound much weaker (Fig.  6B). However, the related TAM ligand PROS1 preferentially bound sTYRO3 and had much weaker binding to sAXL and sMERTK (Fig. 6C), in agreement with observations by Tsou et al. (68) that PROS1 preferentially activates TYRO-3 and GAS6 favors AXL over the other TAM members. When testing the ability of soluble receptors to inhibit MERTK activation, we found that at the concentration of 0.1 M, sAXL effectively inhibited GAS6-induced MERTK phosphorylation while sMERTK could not (Fig. 6D).
To directly examine the role of ACs in MERTK activation, we pre-mixed ACs with GAS6 or PROS1 before treatment of MCF10A-MERTK or MCF10A-pMSCV cells in the presence or absence of soluble receptors. Consistent with the data of Tsou et al. (68) obtained on TAM reporter cell lines, ACs strongly enhanced GAS6 and PROS1 inducible MERTK phosphorylation but not GAS6 or PROS1 inducible AXL phosphorylation (Fig. 7, A and B). This indicated that ACs enhanced the activity of GAS6 and PROS1 toward MERTK but not AXL suggesting the modes of ligand-dependent interaction by AXL and MERTK were different. However, despite the dramatic effects of ACs on MERTK activation, sAXL and sTYRO3 saturated apoptotic cell surface receptors and successfully blocked GAS6 and PROS1 inducible activation of MERTK as evidence by examining the level of MERTK autophosphorylation (Fig. 7, A  and B). Moreover, not only was sAXL highly effective as a receptor decoy to prevent MERTK activation, but it also completely prevented the efferocytosis enhancing effects of MERTK in MCF10A cells, suggesting it acted as a functional trap to block MERTK-mediated efferocytosis independent of the fibronectin type III domains (Fig. 8A).  (n ϭ 2). B, PKH26-red MCF10A-MERTK and MCF10A-pMSCV stable cells were starved for 6 h and then co-cultured with PKH67-green apoptotic CEM in 10% FBS medium for 3 h, and efferocytosis was determined by flow cytometry. C, efferocytosis was carried out as described in B with human GAS6 conditioned medium, GAS6 (1) or (0.5), instead of 10% serum. GAS6 (0.5) was generated by mixing equal volumes of GAS6 (1) with fresh RPMI. NS ϭ not significant. D, representative confocal imaging of PKH26-red MCF10A-MERTK or MCF10A-pMSCV phagocytes and PKH67-green apoptotic CEM co-cultures. The experiment was performed as described in B in an 8-chambered slide. To translate these findings into the complex biology of epithelial efferocytosis in MERTK-expressing breast cancer cells, we explored the effects of adding sAXL when co-incubated ACs with MDA-MB-231 cells that overexpressed endogenous MERTK. As shown in Fig. 8B, sAXL blocked efferocytosis by over 50%, suggesting that MERTK is a preeminent efferocytosis receptor in cancer mammary epithelial cells that up-regulate MERTK. Finally, Amnis Image Stream which had the capacity to both quantify and capture fluorescent imaging of internalization of ACs by phagocytes strongly confirmed that MERTK indeed enhanced efferocytosis in MCF-10A, which could be blocked by sAXL (Fig. 8C). Also, since apoptotic MCF10A cells were used in our Amnis experiments instead of apoptotic CEM cells, it better mimicked the cancer microenvironment where MERTK-overexpressing cancer cells engulfed surrounding dead cells highly abundant in the core of tumor.
While the above findings indicate that MERTK can enhance efferocytosis in epithelial cells, the full consequences of epithelial efferocytosis are not yet realized. To begin addressing this issue, we examined the effect of MERTK on the expression of the immune checkpoint blockades, Programmed death-ligand 1 (PD-L1) and 2 (PD-L2). PD-L1 and PD-L2 are often overexpressed in poorly immunogenic tumors, and allow cancers to evade host immune responses and promote cancer invasiveness (55)(56)(57)(58). In our study, when Mertk was constitutively activated in 293T cells by means of Fc-Mertk transfection, mRNA transcripts and surface expression of PD-L1 and PD-L2 were increased (Fig. 9, A and B). Furthermore, consistent with the current literatures that highly invasive cancer cells often express PD-L1, we found that the aggressive breast cancer cell line MDA-MB 231 highly expressed PD-L1 and knockdown of  MERTK strongly reduced PD-L1 level as compared with scramble knockdown control (Fig. 9C). More importantly, AC treatment further induces PD-L1 level in MDA-MB 231 suggesting that MERTK driven efferocytosis has the potential to inhibit immune checkpoints, an event that may advance tumor progression (Fig. 9C).

DISCUSSION
In the present study, we provide evidence that overexpression of MERTK in epithelial cells results in increased motility and chemoresistance, as well as endowed epithelial cells with a robust gain-of-function capacity for efferocytosis. We show that MERTK alters the chemoresistance/efferocytosis axis, and  acts as a double-edged sword in epithelial cells; first to spare tumor cells from chemotherapy drug induced cell death by promoting AKT survival pathway and second to drive phosphatidylserine (PS)-dependent interactions and processing of bystander ACs by MERTK-expressing tumor phagocytes in the tumor environment. In addition, epithelial cells with MERTK induce PD-L1 and PD-L2 expression, which could alter the immune balance for tumor tolerance and progression. Finally, MERTK directed efferocytosis may be an attractive target in cancers where soluble TAMs as therapeutics can block receptor activation and prevent PS-mediated efferocytosis and tolerance.
While the best understood roles for MERTK lies in their ability to regulate efferocytosis in myeloid-derived cells, such as macrophages and dendritic cells, where systemic MERTK knock-out leads to age-dependent autoimmunity and a systemic elevation in pro-inflammatory cytokines (1, 10, 21-23, 35, 59), MERTK has equally important roles in clearance in solid tissues that include (i) the clearance of shed rod outer segments by RPEs (24), (ii) the clearance of immature spermatogonia by Sertoli cells (25), and (iii) the efferocytosis of dying mammary epithelial by neighboring epithelial cells during involution (26). In this latter scenario during post-lactational involution, there occurs apoptosis of epithelial cells that require rapid clearance to prevent inflammation and fibrosis. Interestingly, within the mammary gland, ACs are cleared within hours of removing pups from a nursing dam by neighboring epithelial cells prior to the infiltration of macrophages to the mammary gland (26). Recent studies indicate that efferocytic epithelial cells are exquisitely dependent on Mertk, as mouse Mertk(Ϫ/Ϫ) fail to clear ACs in the mammary gland, while transplantation of Mertk (ϩ/ϩ) mammary epithelial cells into the mammary fat pads rescues this phenotype (26,37). It is noteworthy that the loss of Mertk in mammary epithelial cells not only caused impaired clearance, but also leads to fibrosis, inflammation, and impaired mammary regeneration (26,37). This suggests that physiological clearance is associated with the production of homeostatic and regeneration factors within the mammary microenvironment.
While the loss of Mertk function in the mouse mammary gland have strong physiological outcomes that affect post-parturition involution, the consequences of Mertk overexpression and/or its native ligand GAS6 overexpression in epithelial cell efferocytosis especially at the tumor micro-environment is not well understood. In the studies described here, while human MERTK was not detectable in normal breast tissue, about 70% of breast cancer tissues expressed high levels of MERTK (Fig.  5A). Moreover, the MCF10A cells lacking endogenous MERTK have limited efferocytic capacity even when saturated with ACs. However, when MERTK was overexpressed, either ectopically or endogenously, this receptor strongly supported efferocytosis as a "stand-alone" receptor in all cases, even after tran- sient expression where compensatory up-regulation of other receptors was not likely (Figs. 3 and 4). In addition, unpublished data by Tsou et al. (manuscript JBC/2014/569020 submitted to JBC) demonstrated that ACs imposed a strong effect on MERTK but not AXL chimeric receptor activation and this observation was also confirmed in our MCF10A-MERTK stable cell line (Fig. 7). These data suggest that up-regulation of MERTK on epithelial cancer cells, as an isolated entity, promotes a favorable environment whereby efferocytosis and receptor activation will be enhanced. Our studies raise the provocative question as to whether tumor cells have evolved to adopt efferocytosis to promote tumor activities especially since classical efferocytic outcomes involve immune tolerance.
The tumor microenvironment consists of epithelial cells, fibroblast, infiltrating immune cells, soluble, and signaling factors such as chemokines and cytokines, and extracellular matrix, which all support neoplastic transformation, proliferation, invasion, and immune tolerance. Recently, the important role of infiltrating immune cells in cancer has become a central focus of cancer biology. Tumor-associated macrophages have been found to enhance tumor initiation, progression and metastasis, while infiltration of CD8ϩ lymphocytes is associated with tumor regression and favorable survival outcomes (37,60,61). Moreover, M2-polarized tumor-associated macrophages are capable of up-regulating mesenchymal markers, and increasing proliferation and migration of cancer cells (62). While the interaction of apoptotic tumor cells with macrophages and dendritic cells in the tumor microenvironment has been greatly studied in recent years, it will be equally important to study the role of tumor cells themselves as a phagocytic entity.
A future goal will be to explore the consequences of MERTK up-regulation and increased efferocytosis on skewing immune responses. Recently, Cook et al. reported that inhibition of Mertk (knock-out) in tumor leukocytes reduced tumor growth and metastasis (37). In mouse models of breast cancer, melanoma, and colon cancer, all tumors proliferated and metastasized poorly in the context of Mertk(Ϫ/Ϫ) myeloid cells (37,63). At the mechanistic level, Mertk(Ϫ/Ϫ) leukocytes, in response to tumor cells, produced lower levels of IL-10 and GAS6 and higher levels of inflammatory cytokines such as IL-12 and IL-6 as compared with wild-type mice. Intratumoral CD8ϩ T lymphocytes were more abundant in the Mertk(Ϫ/Ϫ) tumor microenvironment and the elimination of them resulted in tumor progression. While this study elegantly demonstrated the importance of Mertk receptor in tumor innate immunity and progression in the context of myeloid cells, the contribution of Mertk expression in epithelial tumors themselves was not investigated, particularly whether epithelial cells have the potential to skew the balance of cytokines and chemokines to influence tumor tolerance and progression.
In our pursuit of identifying human MERTK as a suppressor of anti-tumor immunity, we found that breast cancer cell line MDA-MB 231 highly expressed PD-L1, a critical blockade of immune checkpoint that is well known for regulating the balance between T cell activation, tolerance, and immunopathology (55-58). Binding of PD-L1 to PD1 expressed on immune cells has been reported to induce IL-10 production and inhibi-tory effect on effector T cell responses, which might explain the association of PD-L1 expression in cancer with poorer prognosis (64). In our study, we found that, upon treatment of apoptotic cells, PD-L1 expression was further induced while knocking down of MERTK significantly reduced PD-L1 level in highly invasive breast cancer cells MDA-MB 231 (Fig. 9). In addition, constitutive activation of mouse Mertk in 293TN cells promoted both mRNA transcription and surface expression of PD-L1 and a related molecule, PD-L2. This suggests the potential of MERTK inducible efferocytosis in altering immune checkpoint regulation, which will be investigated more extensively in future studies. Finally, at the functional level, it is attractive to predict that inhibition of MERTK mediated efferocytosis in the tumor microenvironment will have therapeutic value, perhaps in an analogous manner as the PS targeting antibodies Bavituxamab and 2aG4 (65)(66)(67). These latter antibodies target PS on the tumor vasculature and inhibit their signaling, with the net result being the repolarization of tumor-associated macrophages from the M2 to the M1-like phenotype. Collectively, our data demonstrate that MERTK may act as a tumor-promoting receptor tyrosine kinase, which conventionally promotes AKT activation, chemoprotection, and migration while non-canonically drives PS-dependent efferocytosis to enhance its tumorpromoting potential. Since MERTK-driven efferocytosis traditionally tends to release factors that promote tissue remodeling and wound healing as well as immune suppression, it is possible that cancer cells adopt MERTK enhanced efferocytosis as a strategy for tumor tolerance.