Defective FasL expression is associated with increased resistance to melanoma liver metastases and enhanced natural kill cell activity.

Abstract

Objective:

The objective was to determine if the absence of FasL signaling would affect melanoma liver metastases by influencing the anti-melanoma properties of liver natural killer (NK) cells.

Methods:

Melanoma liver metastases were induced in wild-type C57BL/6 mice and the gld/gld mutant C57BL/6 mouse strain that expresses a defective form of FasL (CD95L) that fails to engage and signal via the Fas receptor (CD95). Liver metastases were produced by intrasplenic injection of B16LS9 melanoma cells. Liver NK cell activity directed against murine B16LS9 melanoma cells was determined in a 24 hr in vitro cytotoxicity assay. Liver NK cells, NK T cells, and the NK cell surface activation marker, NKG2D, were measured by flow cytometry.

Results:

Mice expressing defective FasL displayed reduced, rather than enhanced, melanoma liver metastases that coincided with increased liver NK cell-mediated tumor cell cytotoxicity. Enhanced cytotoxicity was not mediated by perforin, TNF-α, or tumor necrosis-associated apoptosis-inducing ligand (TRAIL) but was closely associated with elevated interferon-γ (IFN-γ) in the tumor-bearing liver. FasL-defective gld/gld mice also displayed reduced numbers of liver NK T cells, which have been previously implicated in suppression on liver NK cell activity.

Conclusions:

The absence of functional FasL in the liver correlates with a heightened, not diminished, resistance to melanoma liver metastases. The resistance to liver metastases coincides with a significant, albeit transient, increase in liver NK cytotoxicity and elevated levels of IFN-γ in the liver.

Introduction

There is a growing awareness about the important role of natural killer (NK) cells in the immune surveillance of various cancers and their potential use for immunotherapy of many malignancies [1,2]. NK cells can exert anti-tumor effects by: a) perforin and granzyme-mediated cytolysis; b) TNF-α-induced apoptosis; c) apoptosis induced by tumor necrosis-related apoptosis-inducing ligand (TRAIL), which is expressed on NK cells; d) FasL-induced apoptosis; and e) through the pleiotropic anti-tumor effects of interferon-gamma (IFN-γ) [1–4]. NK cell activity differs from organ to organ [5–7], with the liver having the highest number of NK cells of any site in the body [8].

Uveal melanoma (UM) is the most common intraocular tumor in adults and half of the patients will eventually develop liver metastases [9]. Currently there is no therapeutic modality that has been shown to significantly improve the five-year survival of UM patients. Patients with liver metastases typically survive less than one year[10]. In human uveal melanoma patients up to 40% of the lymphocytes infiltrating the intraocular tumors express NK cell markers [11,12]. Depletion of NK cells in nude mice, which have a normal NK cell repertoire but do not possess T cell immunity, results in a steep increase in liver metastases of uveal melanoma xenografts [13]. Likewise, elimination of NK cells in euthymic, T cell-competent mice leads to a dramatic increase in liver metastases arising from intraocular melanomas [14], while in vivo stimulation of NK activity by interferon-beta (IFN-β) gene transfer produces a sharp reduction in liver metastases [15]. Thus, the weight of evidence suggests that NK cells play a crucial role in controlling liver metastases arising from intraocular melanomas in humans and mice [16,17].

We have used intrasplenic injection of the murine B16LS9 melanoma cell line to study the immunobiology of NK cell-mediated surveillance of melanoma liver metastases in mice [13,18,19]. Intrasplenic melanoma cell injection leads to a high incidence of liver metastases, which are susceptible to NK cell-mediated immune surveillance, a condition that recapitulates the natural history of human uveal melanoma metastases [17,20]. We have also reported that liver NK cell activity decays as mice age, which results in a proportional increase in liver metastases arising from either intraocular melanomas or by intrasplenic injection of melanoma cells [18].

The present study addressed the question as to whether the expression functional FasL influences the development of melanoma liver metastases. The rationale for this investigation is based on the following observations: a) the liver is the most common site for metastases arising from uveal melanoma in humans and intraocular melanomas in mice [9]; b) intrasplenic injection of melanoma cells consistently produces liver metastases that are susceptible to NK cell-mediated surveillance [13,18,19]; c) human uveal melanoma liver metastases are susceptible to NK cell-mediated cytolysis [13,18,19,21]; d) B16LS9 melanoma cells used in this study and the majority of human uveal melanomas are Fas+ [22] and e) FasL-induced apoptosis is one of the potential mechanisms that NK cells employ to attack Fas+ tumor cells [2,23].

Materials and Methods

Cell lines

The B16LS9 murine melanoma cell line was kindly provided by Hans E. Grossniklaus (Emory University School of Medicine, Atlanta, GA). B16LS9 cells were derived from hepatic metastases following intracameral injection of B16-F1 cutaneous melanoma cells in C57BL/6 mice [24]. Tumor cells were maintained in complete DMEM medium containing 10% FBS (HyClone, Logan UT), 100 U/ml of penicillin, 50 ng of streptomycin, 0.1% Fungizone (BioWhittaker, Walkersville, MD), 2.0 mM glutamine (BioWhittaker), 0.01 M HEPES buffer (BioWhittaker), and 0.5% 2-Mercaptoethanol (Sigma-Aldrich, St. Louis, MO).

Mice

Wild-type (WT) C57BL/6 mice and C57BL/6 mice with the mutated FasL gene (gld/gld) were purchased from The Jackson Laboratories (Bar Harbor, ME). Cells from mice with the gld/gld mutation fail to induce apoptosis of Fas-expressing target cells [25]. Mice were used at 8–10 weeks of age. Animals were housed and cared for in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) of the University of Texas Southwestern Medical Center and the Association for Research in Vision and Ophthalmology (ARVO) statement concerning the Use of Animals in Ophthalmic and Vision Research. All surgical procedures and intrasplenic injections were performed under anesthesia in the form of a cocktail of ketamine and xylazine given intraperitoneally and buprenorphine was given subcutaneously as an analgesic after splenic injections.

Intrasplenic tumor injections

Intrasplenic tumor cell injection is an effective method to produce liver metastases by facilitating the dissemination of tumor cells to the liver via the hepatic portal system [26,27]. Melanoma cells (5×10⁴) were injected beneath the spleen capsule of anesthetized mice. Mice were euthanized at days 4, 7 and 14 and NK cells were isolated from the livers and used for in vitro cytotoxicity assays. In other experiments, livers were removed and metastatic melanoma foci were assessed by counting surface tumor nodules by two masked observers using a dissecting microscope.

NK cell isolation

Liver NK cells were enriched using the EasySep Mouse NK Cell Enrichment Kit (Stemcell technologies, Vancouver, BC, Canada) according to the manufacturer’s instructions. Briefly, the mouse CD49b positive selection kit isolates CD49b⁺ cells from liver mononuclear cells using an anti-CD49b antibody which binds to the CD49b expressed on the surface of NK cells. Antibody-bound cells are retained in the column by magnetic particles present on the captured antibody, and unwanted cells are eluted. NK cells were defined as NK1.1⁺ CD3⁻ populations using anti-mouse PE-NK1.1 and FITC CD3 (BD Biosciences, San Jose, CA).

NK cells were depleted in vivo by intraperitoneal injection of 50 μl of anti-asialo GM1 antiserum (Wako Pure Chemical, Osaka, Japan) on days −3, +1 and +4 relative to intrasplenic tumor cell injection [13].

Isolation of myeloid-derived suppressor cells (MDSC)

MDSC are an assemblage of immature myeloid cells that are at an early stage of differentiation. In mice, two distinct subpopulations of MDSCs have been identified based on their differential expression of the Gr1 molecule. PMN-MDSC are characterized as CD11b⁺Gr1^hi while mononuclear MDSC (M-MDSC) are CD11b⁺Gr1^lo [28,29]. Livers from naive or tumor-bearing mice were removed immediately after being perfusion with 10 ml of HBSS administered through the portal vein using a 10-ml syringe. Nonparenchymal cells from the livers were collected and liver leukocytes were isolated by isotonic Percoll centrifugation. Briefly, non-parenchymal cells, resuspended in 40% Percoll, were gently overlaid onto 70% Percoll and centrifuged at room temperature for 25 min at 900xg. Purified liver leukocytes were collected from the interface for further analysis of hepatic MDSC and NK cells. NK cells were enriched by EasySep Mouse CD49b Positive Selection Kit (Stemcell Technologies, Vancouver, BC, Canada) according to the manufacturer’s instructions. Gr-1^hi CD11b⁺ (PMN-MDSC) and Gr^loCD11b⁺ (M-MDSC) MDSCs were isolated and quantified quantified by FACS analysis [30].

FACS Analysis

Cells were stained by standard protocols using the following antibodies: anti-mouse PE-NK1.1, APC-TCR-β, FITC-NKG2D, FITC CD 3 (BD Biosciences, San Jose, CA), and isotype control monoclonal antibodies. The cells were washed three times and re-suspended in FACS buffer solution and analyzed by flow cytometry using the Attune NxT acoustic focusing cytometer (Applied Biosystems; Life Technologies). The data from flow cytometer was analyzed using FlowJo v10 software (Tree Star, Ashland, OR, USA).

In vitro NK cell cytotoxicity assay

Splenic and liver mononuclear cells were isolated and enriched for NK cells as mentioned above. NK cell-mediated tumor cell killing was determined using an in vitro cytotoxicity assay in which tumor cells were labeled with ³H-thymidine (Methyl −³H-thymidine; Perkin Elmer, Boston MA) and were co-cultured overnight with purified NK cells at an effector to target cell ratio of 100:1. Tumor cell survival was assessed by measuring ³H-thymidine incorporation using a liquid scintillation counter (Perkin-Elmer, Boston, MA). The percentage of killing was calculated using the formula: percent cytotoxicity = [(A-B/A] X100, where A represents counts per minute (cpm) of tumor cells cultured alone, and B represents the cpm in test cultures. Similar ³H-thymidine-based cytotoxicity assays evaluated the susceptibility of B16LS9 melanoma cells to killing by either IFN-gamma (IFN-γ) or tumor necrosis factor-α (TNF-α). A neutralizing antibody to murine TNF-α (10 ng/ml; Cell Signaling Technology, Danvers, MA) was added to in vitro NK cell cytotoxicity assays as a blocking agent to inhibit TNF-α-mediated tumor cell lysis.

Cytokine and perforin quantification by ELISA

Unfractionated liver cells were collected from gld/gld and WT mice on days 4, 7 and 14 post tumor cell injection and immediately homogenized without in vitro cultivation. The whole liver homogenates were assessed for the presence of mouse IFN-gamma (IFN-γ) and perforin by sandwich ELISA (R&D systems, Minneapolis, MN) according to the manufacturer’s instructions.

Statistical analysis:Results for cytotoxicity assays, flow cytometry, liver metastases counts, and ELISAs were evaluated by Student’s t-test. Results are expressed as mean ± SEM. Differences in all experimental groups were considered statistically significant if P < 0.05. Each mouse experiment was performed at least twice with similar results. There ≥ 5 mice/group. Cytotoxicity assay and ELISA samples were tested in quintuplicate.

Results

Mice with defective FasL have reduced liver metastases of Fas+ melanoma cells

Fas signaling is known to induce apoptosis of tumor cells and inhibit tumor growth and metastasis [31–33]. There is a growing body of evidence suggesting a strong correlation between NK cell activity and resistance to the development of liver metastases in human uveal melanoma and in mouse models of intraocular melanoma [14,19,21,34–38]. NK cells have many anti-tumor mechanisms at their disposal including: a) perforin and granzyme-mediated cytolysis; b) tumor-necrosis factor-alpha (TNF-α) induced apoptosis; c) apoptosis induced by either tumor necrosis factor-related apoptosis inducing ligand (TRAIL) or FasL (CD95L); or d) by elaboration of interferon-gamma (IFN-γ) [3]. With this in mind we sought to determine if B16LS9 melanoma cells used in this model of liver metastasis expressed the Fas receptor (CD95) and if mice expressing a defective, non-functional form of FasL (gld/gld) had an increased incidence of liver metastases. Like many other murine tumors, B16LS9 expressed significant quantities of Fas receptor (Figure 1A). However, gld/gld mice displayed a significant reduction in liver metastases that was detected as early as day 7 post tumor cell injection (Figure 1B–1F).

Figure 1.

Fas expression on B16LS9 melanoma cells and numbers of melanoma liver metastases in mutant mice with defective FasL (gld/gld). (A) Expression of Fas (CD95) on B16LS9 melanoma cells were analyzed for the expression of Fas receptor (CD95) by flow cytometry. Liver metastases 14 days following intrasplenic injection of B16LS9 melanoma cells in gld/gld (B) and wild-type (C) mice. Number of liver metastases in the whole liver were counted by two independent, masked observers on day 4 (D), day 7 (E), and day 14 (F) after intrasplenic injection. The results are representative of two independent experiments (N=5/group).

Enhanced liver NK cell activity in FasL-defective mice

Liver cell-mediated cytotoxicity activity was examined in gld/gld and WT mice to determine if the reduction in metastasis in FasL-defective mice was the result of elevated liver NK cell activity. Liver NK cell lysis of melanoma cells was evaluated at days 4, 7, and 14 and revealed that gld/gld mice expressed significantly elevated cytolytic activity 7 and 14 days after melanoma cell injection (Figure 2A–2C). Livers from gld/gld and WT mice were further analyzed to determine if the heightened resistance to metastases was due to an increase in the absolute numbers of NK cells in the livers of gld/gld mice compared to WT mice. The number of NK1.1⁺CD3⁻ lymphocytes in the livers of tumor-bearing gld/gld and WT mice was evaluated 4, 7, and 14 days after melanoma cell injection. Even though gld/gld mice displayed elevated liver NK cell-mediated killing of melanoma cells on days 7 and 14, the number of liver NK cells in gld/gld and WT mice were the same (Figure 2D–2I). The purity of a typical liver NK cell preparation used in the in vitro assays was >90% NK1.1⁺CD3⁻ as determined by flow cytometry (Figure 2J).Moreover, depletion of NK cells with anti-asialo GM1 antiserum eliminated the enhanced resistance to liver metastases in gld/gld mice. WT and gld/gld mice depleted of NK cells displayed the same number (P>0.05) of liver metastases (Figure 2K), which provided further support that the observed effects in gld/gld mice were NK cell-dependent.

Figure 2.

Reduced liver metastases in mutant mice with defective FasL (gld/gld) is associated with enhanced liver NK cell-mediated cytotoxicity and requires an intact NK cell repertoire. In vitro NK cell-mediated cytolysis of B16LS9 melanoma cells on days 4 (A), 7 (B), and 14 (C) post tumor cell injection. The numbers of NK cells per the entire liver were determined by flow cytometry on days 4 (D,G), 7 (E,H,) and 14 (F,I). The purity of the NK cell population (NK1.1⁺CD3⁻) was confirmed by flow cytometry(J). NK cells were depleted by in vivo injection of anti-asialo GM1 antiserum on days −3, −1, and +4 and liver metastases were counted 7 days after intrasplenic tumor cell injection (K). Each set of results is representative of two independent experiments with N=5 mice per group, with the exception of anti-asialo GM1 treatment (K), which was performed once. Cytotoxicity are represented as percentage of liver NK cell cytotoxicity against thymidine labelled B16LS9 cells and metastases results are expressed as total number of metastatic tumor foci per liver. In the in vitro assays, each specimen was analyzed in quintuplicate.

The enhanced liver NK cell cytotoxicity in gld/gld mice might be due to a reduced infiltration of myeloid-derived suppressor cells (MDSC), which can inhibit NK cell activity [39–41]. With this in mind we examined the tumor-bearing livers of gld/gld and WT mice and evaluated the number of CD11b+ Gr-1^lo mononuclear MDSC (M-MDSC) and CD11b+Gr1^hi (PMN-MDSC) [28,29]. The results showed that the livers of tumor-bearing gld/gld mice had the same numbers of PMN-MDSC (Figure 3A–3F) and M-MDSC (Figure 3G–3L) as tumor-bearing WT mice at all three time points examined.

Figure 3.

Numbers of myeloid-derived suppressor cells (MDSC) in livers of gld/gld mice with melanoma liver metastases. B16LS9 melanoma cells were injected intrasplenically and the number of CD11b+Gr1^hi (PMN-MDSC) MDSC was assessed on day 4 (A and D), day 7 (B and E) and day 14 (C and F). The number of CD11b+ Gr-1^lo (M-MDSC) was assessed on day 4 (G and J), day 7 (H and K), and day 14 (I and L) by flow cytometry. These experiments were performed twice with 5 mice per group at each time point.

We have previously reported that mice deficient in NKT cells have a significant decrease in melanoma liver metastases that is closely associated with a commensurate increase in liver NK cytolytic activity [13,19]. Accordingly, we tested the hypothesis that the enhanced liver NK cell activity and reduction in liver metastases in gld/gld mice was due to diminished numbers of liver NKT cells in the gld/gld mice. Flow cytometric analysis of the liver NKT cell population revealed that at days 4 and 7, gld/gld mice displayed a significant reduction in the number of NK1.1⁺, CD3⁺ NKT cells compared to WT mice (Figure 4).

Figure 4.

Mice with defective FasL (gld/gld) display reduced numbers of liver NKT cells. B16LS9 melanoma cells were injected intrasplenically and the number of TCR-β+NK1.1+ liver NKT cells were assessed on days 4 (A), 7 (B), and 14 (C) by flow cytometry. These experiments were performed twice (n =5 mice at each time point) with similar results.

Liver NK cells from FasL-defective mice have elevated cell surface expression of the NKG2D activation receptor

NKG2D is a C-type lectin-like receptor that is widely expressed on NK cells [42] and when engaged with its ligand, it induces potent activation of NK cells [43]. We examined the expression NKG2D on liver NK cells from gld/gld and wild-type mice with the suspicion that the elevated NK cytotoxicity in gld/gld mice might correlate with enhanced NKG2D expression. The results suggested that this was the case as NKG2D expression was over two-fold higher on liver NK cells from gld/gld mice at days 7 and 14 post melanoma cell injection compared to WT control mice (Figure 5). The elevated NKG2D expression coincided with the time points when NK cytotoxicity was elevated in the gld/gld mice (Figure 2).

Figure 5.

Enhanced expression of NK activation receptor NKG2D on liver NK cells in tumor-bearing gld/gld mice compared to wild-type mice. B16LS9 melanoma cells were injected intrasplenically and TCR-β^- NK1.1⁺ liver NK cells were isolated and the expression of the NK activation receptor, NKG2D, was assessed by flow cytometry on days 4 (A), 7 (B), and 14 (C) by flow cytometry. These experiments were performed twice (n =5 mice at each time point) with similar results.

Enhanced liver NK cell killing of melanoma cells in FasL-defective mice correlates with increased interferon-γ production

NK cells possess multiple mechanisms for killing tumor cells including: a) perforin-mediated cytolysis; b) TNF-α-induced apoptosis; c) FasL-induced apoptosis; and d) through the pleiotropic effects of IFN-γ on multiple immune cell populations [3,44]. We have previously found that although B16LS9 melanoma cells express the TRAIL receptor (DR5), they are not susceptible to TRAIL-mediated killing by NK cells. Likewise, the liver NK cells from gld/gld mice fail to express functional FasL and therefore the heightened cytolysis of B16LS9 melanoma cells by these NK cells is FasL-independent. Therefore, we explored the possibility that the enhanced killing of melanoma cells by liver NK cells from tumor-bearing gld/gld mice might be due to perforin, TNF-α, or IFN-γ. Although we have previously shown that B16LS9 melanoma cells are highly susceptible to perforin-mediated cytolysis, the cytoplasmic expression of perforin in liver NK cells from gld/gld mice was not significantly different from that found in WT mice at days 4, 7, or 14 post melanoma cell injection (Figure 6A–6C). We next explored the role of TNF-α in the increased cytolysis of B16LS9 melanoma cells by liver NK cells. Accordingly, melanoma cells were interrogated for their susceptibility to TNF-α-mediated killing in a 24 hr in vitro cytotoxicity assay. The results of multiple experiments showed that TNF-α, even at a dose of 100 ng/ml, produced less than 10% killing of B16LS9 melanoma cells (Figure 6D). To confirm that TNF-α was not crucial for the NK cell killing of B16LS9 melanoma cells, a blocking antibody against TNF-α was added to the NK cell-mediated cytotoxicity assays. NK cell-mediated cytolysis of B16LS9 melanoma cells occurred in the presence of anti-TNF-α antibody, which provided further evidence that TNF-α was not an important mediator of NK cell-mediated killing of melanoma cells (Figure 6E). By contrast, IFN-γ produced 25% and 45% cytotoxicity of B16LS9 melanoma cells at doses of 100 units and 500 units respectively (Figure 6F). The likelihood that IFN-γ played a role in the increased resistance to liver metastases in gld/gld mice was supported by the observation that at day 7, unfractionated liver cell suspensions from gld/gld mice secreted significantly higher levels of IFN-γ than WT mice (Figure 6G–6I). Moreover, removal of NK cells from the liver cell suspensions resulted in a significant diminution in IFN-γ confirming that the bulk of the IFN-γ produced in the livers of tumor-bearing mice was produced by liver NK cells (Figure 6J).

Figure 6.

NK cell-mediated cytotoxicity of melanoma cells correlates with susceptibility to interferon-γ (IFN-γ) and transient increase in IFN-γ levels in the liver. Perforin levels were measured in bulk liver homogenates (n=5 for each time point) by ELISA. The perforin levels were not increased in gld/gld livers 4 days (A), 7 days (B), or 14 days (C) after intrasplenic injection of B16LS9 melanoma cells. B16LS9 melanoma cells were not susceptible to TNF-α-mediated apoptosis (D). Addition of anti-TNF-α antibody (10 ng/ml) to in vitro cytotoxicity assay using liver NK cells (E). In vitro cytotoxicity assays using soluble IFN-γ (F). IFN-γ levels were assessed by ELISA 4 days (G), 7 days (H) and 14 days (I) following intrasplenic tumor injection. IFN-γ levels were also measured in livers collected at day 7 and depleted of NK cells (J). In the in vitro assays, each specimen was analyzed in quintuplicate. All experiments were performed twice with the exception of Figure 6E, which was performed once.

Discussion

There is an overwhelming body of evidence suggesting that NK cells play a crucial role in the immune surveillance of liver metastases arising from uveal melanomas in humans and intraocular melanomas in mice [13–17,19,20,36]. NK cells have multiple mechanisms for eliminating melanoma liver metastases including FasL-induced apoptosis. B16LS9 melanoma cells and most human uveal melanomas express Fas (CD95) and are potentially vulnerable to FasL-induced apoptosis [45]. However, the present results indicate that instead of exacerbating melanoma metastases, the absence of functional FasL on liver NK cells was associated with an increased resistance to metastatic disease. The enhanced resistance to liver metastases in mice expressing defective FasL (i.e., gld/gld mice) was associated with a transient, albeit significant increase in the cytolytic activity of liver NK cells. The elevated cytolytic activity coincided with increased expression of the NKG2D activation receptor on liver NK cells at days 7 and 14 post melanoma cell injection. This is consistent with previous investigations reporting that NKT cell-deficient mice display increased expression of the NKG2D activation receptor on liver NK cells and a commensurate enhancement of liver NK cytolytic activity that correlated with reduced liver metastases [13,19]. The present findings are in keeping with the notion that the higher expression of the NKG2D activation receptor and enhanced NK cell-mediated killing of melanoma cells correlated with resistance to the development of liver melanoma metastases in mice with defective FasL signaling. We confess that we do not have any insights as to the mechanisms whereby expression of defective FasL would lead to increased expression of the NKG2D activation receptor on liver NK cells. Although FasL-induced apoptosis is a mechanism utilized by NK cells to kill tumor cells [4], the present results indicate that NK cells lacking functional FasL expresses higher levels of the NK activation receptor NKG2D, which coincides with heightened cytolytic activity of hepatic NK cells.

Perforin-mediated cytolysis is arguably one of the most important mechanisms for NK cell-mediated elimination of tumors [1–3]. However, perforin expression was not elevated in liver NK cells in gld/gld mice at any of the time points tested. Although NK cells can mediate tumor cytolysis by elaborating TNF-α, our results indicate that the B16LS9 melanoma cell line used in this study is resistant to TNF-α-induced apoptosis. Moreover, blocking soluble and cell-membrane bound TNF-α does not diminish liver NK cell-mediated cytolysis in vitro. As stated earlier, NK cells have five pathways for producing anti-tumor effects: a) perforin-mediated cytolysis; b) FasL-induced apoptosis; c) apoptosis induced by TNF-α; d) TRAIL-induced apoptosis; and e) the pleiotropic effects of IFN-γ. We have systematically ruled out the first four pathways for NK cell-mediated elimination of melanoma metastases, which points to IFN-γ as the most plausible candidate responsible for the enhanced NK cytolytic activity in FasL-defective mice. NK cells are major producers of IFN-γ [46,47]. Moreover, we detected IFN-γ in supernatants from in vitro cultured liver cell suspensions from tumor-bearing donors. However, IFN-γ was no longer detected in these cell suspensions when NK cells were depleted. This strongly implicates NK cells as the sole producers of IFN-γ in the tumor-bearing livers. IFN-γ has multiple properties that can influence tumor growth and metastasis. IFN-γ can directly induce apoptosis of many categories of tumors including melanoma as well as regulate tumor angiogenesis [44,45]. IFN-γ can shape anti-tumor responses by polarizing macrophages to an M1 phenotype which expresses tumoricidal activity [48]. Tumor growth can be inhibited by an IFN-γ-dependent process that requires NK cells [44]. IFN-γ can also directly activate NK cells [49]. However, we are aware that the levels of IFN-γ that produced cytotoxic activity against melanoma cells in vitro in our study may not occur in vivo and thus, the in vitro findings suggesting that IFN-γ directly killed melanoma cells in the liver should be viewed with caution.

The absence of FasL signaling in gld/gld mice might indirectly contribute to resistance to metastasis by blunting the invasiveness of melanoma cells. Stimulation of Fas on colorectal cancer cells leads to enhanced invasiveness and growth in a mouse model of colorectal liver metastasis [50]. Thus, the absence of Fas signaling might blunt the invasiveness of melanoma cells entering the blood vessels draining the spleen and entering the portal vasculature. Similarly, blockade of Fas signaling in 4T1 mammary breast cancer cells reduced tumor growth and inhibited tumor metastasis in a murine xenograft model [51]. FasL released by activated T cells promotes B16 melanoma metastases through a Fas/FasL signaling process that correlates with increased expression of MMP9, which facilitates tumor cell invasiveness [52].

MDSC are a heterogeneous population of immature cells of myeloid origin including macrophages, neutrophils and dendritic cells that can exert inhibitory effects on NK cells [30,40,41,53]. There is evidence that blockade of Fas/FasL reduces malignancy by decreasing the recruitment of MDSC [51]. MDSC accumulate in high frequencies in most cancer patients and experimental animals where they can exert a profound immune suppression and thus, are a significant obstacle to immunotherapy [54]. An important property of MDSC is their capacity to impair NK cell-mediated cytotoxicity [39–41]. MDSC can down-regulate the expression of the activating receptors NKG2D, NKp44, NKp30, and NKp46 on NK cells, thereby making NK cells unresponsive to activation or proliferation signals [39,55,56]. However, our findings indicate that gld/gld mice had same numbers of MDSC as WT mice at all time points examined, which suggests that the presence of MDSC did not affect the expression of the NKG2D activation receptor nor did MDSC impair liver NK cell cytotoxicity. Thus, our findings suggest that the absence of Fas/FasL signaling in gld/gld mice did not result in an increased accumulation of either CD11b+Gr1^hi (PMN-MDSC) or CD11b+ Gr-1^lo (M-MDSC) in the livers of tumor-bearing mice, which indicates that MDSC most likely did not account for the differences in NK cell activity in gld/gld and WT mice.

The sharp reduction in liver metastases that occurs in mice with defective FasL raises the question as to whether blocking Fas/FasL interactions might have application for the management of liver metastases in UM patients. This proposition is based on the following observations: a) B16LS9 melanoma cells and most uveal melanomas express Fas receptor; b) liver NK cell activity correlates with resistance to metastatic disease in UM patients and in mice with intraocular melanoma; and c) NK activity is enhanced when Fas/FasL interactions are impeded (i.e., gld/gld mice).