Native Soluble Carcinoembryonic Antigen Is Not Involved in the Impaired Activity of CD56^dim Natural Killer Cells in Malignant Pleural Effusion

Human carcinoembryonic antigen (CEA), a highly glycosylated protein of about 200 kDa, is a well-characterized tumor-associated antigen that has been found overexpressed in a high percentage of human tumors, including 90% of gastrointestinal, colorectal and pancreatic cancers, 70% of non-small cell lung cancer cells and 50% of breast cancers [1,2]. High levels of CEA have been implicated in enhanced metastasis and the development of malignancy [3]. It has been proven to be a suitable target antigen for the detection of primary and metastatic colorectal and some other carcinomas.

Natural killer (NK) cells are lymphocytes of the innate immune system that play a crucial role in tumor immune surveillance because of the ability to recognize and eliminate transformed cells. Human NK cells account for 10-20% of peripheral blood (PB) lymphocytes and are characterized by CD3^- and CD56⁺. NK cells can be divided into two main functional subsets according to the intensity of CD56 expression: CD56^dim NK cells, accounting for 90% of total NK cells in PB, have superior cytotoxic potential, whereas CD56^bright NK cells, comprising 10% of total NK cells in PB, have a greater ability to produce cytokines and are thought to be a regulatory subset [4,5]. Accumulated data indicated that NK cells in the tumor microenvironment often displayed a suppressed function. Stern and his colleagues [6] had demonstrated that both the CEA-Ig recombinant protein and CEA in the context of mammalian cells by using cell transfectants inhibited NK cell killing via interaction with CEA-related cell adhesion molecule 1 (CEACAM1). However, the CEA used in their studies, but not the native CEA, was artificially modified. So far, the effect of native soluble CEA on NK cell killing was rarely reported. The relationship between native soluble CEA from the tumor microenvironment and NK cells is still mostly unknown. Malignant pleural effusion (MPE) is a common and distressing symptom in patients at advanced stage of various malignant diseases. Lung cancer is the most common etiology of MPE, accounting for 37% of all malignant effusions [7]. High levels of CEA are detected in squamous cell cancer and adenocarcinoma of the lung. Thus, MPE offers a good platform to study the interaction of CEA and NK cells in the tumor microenvironment. Early studies have confirmed the low cytolytic activity of NK cells against tumor cells in MPE [8,9].

In this study, the effect of native soluble CEA within MPE on CD56^dim NK cells was investigated. It was observed that NK cells were accumulated in MPE; however, the percentages of total NK cells and the CD56^dim NK subset reduced the accompanying impaired cytotoxic activity in MPE compared with that in paired PB. Cell-free MPE (CFMPE) treatment reduced both the proportion and cytotoxic activity of CD56^dim NK cells in PB from healthy donors. CEA was not involved in the suppression of NK cells. Further, the suppression effects of CFMPE were not based on the inhibitory cytokines interleukin (IL)-10 and transforming growth factor (TGF)-β₁, but were dependent on the factor with a molecular weight >100 kDa.

Informed consent was obtained from patients according to the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, and the study protocol was also approved by this institution. MPE samples were collected from 22 patients (age range 36-81 years, 10 males, 12 females) with newly diagnosed lung cancer. A diagnosis of MPE was established by the showing of malignant cells in pleural effusion and/or on closed pleural biopsy specimens. None of the patients had received any anticancer treatment, corticosteroids or other nonsteroid anti-inflammatory drugs at the time of sample collection. The levels of CEA in MPE ranged from 40.37 to 1,000 ng/ml.

MPE samples were collected from each subject, using a standard thoracocentesis technique. Ten milliliters of venous PB was drawn simultaneously. PB mononuclear cells (PBMCs) from patients and healthy donors were isolated by Ficoll-Hypaque gradient centrifugation. MPE samples were centrifuged, and the supernatant was collected and used as CFMPE. The cell pellets were resuspended in PBS, and mononuclear cells were obtained by Ficoll-Hypaque gradient centrifugation and described as MPE mononuclear cells. CFMPE was centrifuged at 12,000 rpm for 1 h at 4°C with different Amicon Ultra-15 Centrifugal Filter Units (Millipore) to be divided into different parts according to the size. The human erythroleukemia cell line K562 was cultured in RPMI-1640 culture medium supplemented with 10% heat-inactivated FBS.

The following fluorescent-labeled monoclonal antibodies were used. Alexa Flour488 anti-human CD66a/c/e, percp-Cy5.5-anti-CD3 and isotype controls were from eBioscience (San Diego, Calif., USA). FITC-anti-CD3, APC-anti-CD56, FITC-anti-CD107a and neutralizing antibodies anti-IL-10 were from BD Pharmingen (San Jose, Calif., USA). Neutralizing antibody of TGF-β₁ was from R&D Systems (Minneapolis, Minn., USA). Protein-A- and protein-G-conjugated agarose were from Santa Cruz Biotechnology (Santa Cruz, Calif., USA). Anti-CEA polyclonal antibody was from Abcam (Hong Kong). Anti-CEA monoclonal antibody CC4 was a gift from the Institute of Biophysics, Chinese Academy of Sciences, Beijing, China [2].

Immunoprecipitation was performed to remove CEA from CFMPE. 40 µg polyclonal antibody of CEA was mixed with 40 µl protein-A- and 40 µl protein-G-conjugated agarose at 4°C on a shaking platform overnight. Then the mixture was centrifuged and washed with PBS twice as well as incubated with CFMPE for 6 h at 4°C on a shaking platform. After that, the mixture was centrifuged at 3,000 rpm for 30 s at 4°C, and the supernatant was carefully collected for the indicative assay.

PBMCs from healthy donors were adjust to a density of 1 × 10⁶ cells/ml in a 24-well plate and treated with CFMPE in indicated conditions for 24 h and were collected and resuspended in RPMI-1640 medium supplemented with 10% FBS for flow cytometry analysis or cytotoxicity assay.

PB or MPE mononuclear cells (1 × 10⁶) were stained with FITC-anti-CD3 and APC-anti-CD56 for 30 min at 4°C, or 1 × 10⁶ PB mononuclear cells from healthy donors were stained with FITC-anti-CD3, APC-anti-CD56 and Alexa Flour488-anti-human CD66a/c/e for 30 min at 4°C. After that, cells were washed twice in PBS-1% FBS and analyzed by flow cytometry. Labeled isotype controls were also used to evaluate non-specific staining.

PB or MPE mononuclear cells were coincubated with target cells K562 at a ratio of 1:1 in a final volume of 100 µl in round-bottomed 96-well plates for 4 h. Then cells were stained with CD3, CD56 and CD107a antibodies for 30 min at 4°C. After that, cells were washed twice in PBS-1% FBS and analyzed by flow cytometry. Results are expressed as the percentage of CD107a-positive cells in CD56^dim NK cells.

Data are presented as the mean ± SEM. Statistical analyses were performed with GraphPad Software. Paired data comparisons were made using a Wilcoxon signed-rank test, and data from different groups of treatments were analyzed using the t test. A p value <0.05 was considered significant.

Mononuclear cells in 15 pairs of MPE and PB were first assayed by flow cytometry. As shown in figure 1a and b, NK cells appeared in MPE. The average percentage of NK cells (CD3^- CD56⁺) in MPE mononuclear cells was 7.7%, i.e. significantly lower than that in paired PBMCs (15.2%). The proportion of CD56^dim NK cells in MPE was also lower compared with that in paired PB (fig. 1a, c). In contrast, the percentage of CD56^bright NK cells in MPE was relatively higher (fig. 1a, c).

The surface expression of CD107a, which is present in the cytotoxic activity of NK cells, on CD56^dim NK cells was detected by flow cytometry after target K562 cell encounter for 4 h. As illustrated in figure 2, in the MPE group, not only the frequency of CD107a⁺ CD56^dim NK cells (fig. 2a, b) but also the mean fluorescence intensity of CD107a on CD56^dim NK cells (fig. 2c) was significantly lower compared with that in paired PB.

To confirm the association between the impaired amount and function of CD56^dim NK cells and MPE, PBMCs from 1 healthy donor were treated with CFMPE from 2 patients for 72 h. As shown in figure 3a, after treatment, at 24-, 48- and 72-hour time points, both the percentage of NK cells (CD3^- CD56⁺ cells) and the proportion of CD56^dim NK cells were more significantly reduced than those in the medium. Furthermore, treating the same healthy PBMCs with CFMPE from 7 patients depressed the surface expression of CD107a on CD56^dim NK cells (fig. 3b). Treating PBMCs from 3 donors with the same CFMPE got similar results (fig. 3c).

To determine whether native soluble CEA which was dramatically elevated in MPE was the critical factor for suppression of cytotoxicity of CD56^dim NK cells, the blocking experiment was performed. As shown in figure 4a, CD56^dim NK cells in healthy donors express CEA receptors CD66a/c/e. The molecular weight of CEA is about 180 kDa, so the >100-kDa part in CFMPE was separated by ultrafiltration centrifugation to be used in the following experiments. As shown in figure 4b, treatment of the >100-kDa part significantly impaired the cytotoxicity of CD56^dim NK cells from healthy donors, and neutralizing antibody CC4 did not rescue the changes. The effect of treatment of the >100-kDa part on the percentage of NK cells was of marginal significance (p = 0.06, n = 3; fig. 4c). Removing CEA with immunoprecipitation (fig. 4d, e) had no significant effect on the ability of the >100-kDa part to bring about the changes in NK cells. The data suggested that CEA had no significant effects on the changes of NK cells in CFMPE.

To identify the component in CFMPE which impairs the amount and function of CD56^dim NK cells, the effects of TGF-β₁ and IL-10, which are potent immunosuppressive factors and significantly elevated in MPE [10,11], on NK cells were assayed. As shown in figure 4f and g, neither TGF-β₁- nor IL-10-neutralizing antibody resumed the cytotoxicity of CD56^dim NK cells treated with CFMPE. In addition, blocking TGF-β₁ or IL-10 had no effect on the ability of CFMPE to bring about the decrease in the percentage of NK cells (online suppl. fig 1A, B; for all online supplementary material, see www. karger.com/doi/10.1159/000345214).

To identify the component impairing the cytotoxicity of CD56^dim NK cells in MPE, ultrafiltration tubes with several sizes were used to divide CFMPE into different parts. Compared with the medium, the full supernatant and the parts >100 kDa in CFMPE impaired the percentage of NK cells and the cytotoxicity of CD56^dim NK cells (n = 7; fig. 4h, i). However, the parts <100 kDa or <10, <30 and <50 kDa had no impairment on NK cells (online suppl. fig. 1C, D).

MPE is a common and distressing symptom in patients at advanced stage of various malignant diseases. In a study of patients presenting with MPE, lung cancer was the most frequent cause [12]. In this study, a significant reduction in NK cells was observed in MPE compared with paired PB from lung cancer patients. This was consistent with an early study by Okubo et al. [9]. Furthermore, the proportion of the CD56^dim NK subset in MPE was much lower than that in paired PB, while the CD56^bright NK subset was higher. The altered proportion of NK subsets was also described in several other tissues, as well as in some human diseases [13,14,15]. There were various interpretations about this finding, including apoptosis of the CD56^dim NK cell subset or recruitment of the CD56^bright NK subset [14,16,17]. As CD56^dim NK cells is the main subset to kill tumor cells, the decreased cytolytic activity of NK cells in MPE might attribute to the reduced proportion of the CD56^dim NK subset. Therefore, CD56^dim NK cells in MPE were investigated by our team. CD107a is considered a functional marker for the identification of activity of NK cells [18]. Consequently, in this study, the cytotoxic activity of CD56^dim NK cells was evaluated by detecting the surface expression of CD107a after coincubation with target cells. It was found that the cytotoxic activity of CD56^dim NK cells was considerably decreased in MPE compared with paired PB.

Although Stern et al. [6,19] reported that recombinant CEA and CEA transfected in mammalian cells inhibited the cytotoxicity of NK cells via CEACAM1, which is one of the receptors of CEA, in this study, we found that native soluble CEA in MPE did not play a role in the impairment of NK cell cytotoxicity. Native CEA is a highly glycosylated macromolecule, and this glycosylation plays a critical role in its function [20]. That may explain why the effect of native CEA on NK cells is totally different from that of recombinant CEA. In the experiment system of Stern et al. [6], NK cells expressed a high level of CD56; thus, we analyzed the effect of CEA on CD56^bright NK cells in our study as well. As shown in online supplementary figure 2, using CD107a as the marker of degranulation, CFMPE decreased the degranulation of CD56^bright NK cells. CEA blockade or depletion and TGF-β₁ or IL-10 blockade were not able to modulate the decrease. Because Stern et al. [6] used a CEACAM1+ NK cell clone and we used primary NK cells in the experiment system, we concluded that NK cells from different origins may display a different response to CEA stimulation. It suggests that the real effect of native CEA on the killing ability of NK cells needs to be further investigated and reevaluated.

Previous studies compared cytokines and other soluble factors between MPE and paired PB [21], among them, TGF-β₁ and IL-10 which were considered to inhibit the cytotoxic activity of NK cells which was highly increased in MPE compared with paired PB [22,23]. However, our study demonstrated that neither TGF-β₁ nor IL-10 played a role in the impairment of cytotoxic activity of CD56^dim NK in MPE.

Further study indicated that factors in MPE >100 kDa depressed the cytotoxicity of CD56^dim NK cells. The components of MPE are complicated. Both immune cells and abundant tumor cells are infiltrated in MPE although the pathophysiology of MPE is multifactorial and still incompletely understood. Because chemokines and cytokines from immunocytes abnormally accumulated in MPE are usually small molecular proteins [21], we speculated that the factors which impact the cytotoxicity of NK cells and are >100 kDa are released by tumor cells.

In summary, CD56^dim NK cells displayed a series of alteration in MPE compared with paired PB, including the reduced proportion and impairment of cytotoxic activity which appeared to be affected by a component >100 kDa in cell-free pleural effusion from MPE. These results are helpful for understanding the mechanism that may lead to impaired NK cell-mediated tumor rejection in lung cancer and provide clues for immune therapy by NK cells. Components >100 kDa in CFMPE need to be further investigated and identified.

We thank the Department of Clinical Laboratory, 1st Hospital of Wuhan, for measuring the concentration of CEA. This work was supported by grants from the National Natural Sciences Foundation of China (No. 81072466 and 30801378).

The authors have no financial conflicts of interest.

J.Q. and D.L. contributed equally to this paper.