Introduction
HLA-G is a tolerogenic molecule which expression was originally observed and characterized on throphoblasts. Even though it has recently been suggested that it participates in the induction of trophoblast cell fusion [
1,
2], HLA-G is best known for its ability to confer protection to the semi-allogeneic fetus from the maternal immune system [
3,
4]. HLA-G differs from classical MHC class I molecules by its genetic diversity, expression, structure, and functions. It is characterized by a relatively low allelic polymorphism and a highly restricted tissue distribution. HLA-G constitutive expression is mainly restricted to trophoblast cells [
3], and to adult thymic medulla [
5], pancreatic islets [
6], and stem cells [
7,
8]. However, HLA-G can be neo-expressed in pathological conditions such as transplantation [
9], inflammatory and autoimmune diseases [
10], viral infections [
11], and cancers [
12]. HLA-G expression is under the control of epigenetic mechanisms. In most adult tissues, HLA-G gene expression in repressed by methylation, which can be reversed by demethylating agents such as 5-aza-2′-deoxycytidine [
13]. Of note: such demethylating agents are used in cancer therapy; they might therefore have the adverse effect of inducing immune escape through HLA-G expression. Concerning HLA-G expression regulation, it is under the control of polymorphism and microenvironmental factors such as hypoxia and cytokines, and miRNA such as MiR148a and MiR152 (for review, see [
14,
15]).
HLA-G is exclusively immune-inhibitory. Under its membrane-bound and soluble forms, it is able to inhibit NK cells and cytotoxic T lymphocyte cytolytic activity [
4,
16-
19], proliferative T cell responses, T cell and NK cell ongoing proliferation [
20-
22], and dendritic cell maturation [
23,
24]. Recent studies have also shown that HLA-G is capable of inducing the differentiation of regulatory T cells and antigen-presenting cells (APCs), which can then inhibit immune responses themselves [
20,
23,
25-
27].
HLA-G neo-expression has been detected in several human cancers including melanoma, renal cell carcinoma, breast carcinoma, and large cell carcinoma of the lung [
12,
28-
32]. HLA-G expression by tumor cells has been shown to be important for the escape of immune surveillance by host T lymphocytes and NK cells [
12,
28,
29,
33-
35]. HLA-G promoter specificities even allow its expression when that of classical HLA-class-I molecules is downregulated. This is particularly evident in trophoblast cells, but also occurs in tumor cells. Thus, HLA-G expression by malignant cells may prevent tumor immune eradication by inhibiting the activity of tumor infiltrating NK, cytotoxic T lymphocytes (CTL), and APCs. The clinical relevance of HLA-G expression by tumors as a prominent immune escape mechanism was supported by numerous studies and for both solid and liquid tumors (for review, see [
36,
37]). HLA-G is not a tumorigenic molecule per se, but it could contribute to tumorigenesis if expressed by cancer stem cells or precancerous cells, shielding them from immune destruction during their evolution processes. Of particular relevance to our study, HLA-G expression in B cell chronic leukemia correlated with a strong immunodeficiency and poor clinical evolution [
38,
39].
HLA-G acts mainly through two inhibitory receptors: LILRB1/ILT2 and LILRB2/ILT4 that are differentially expressed by NK, T, and B cells, and myeloid APCs. HLA-G was also reported to exert its tumor immune escape functions by the mechanism of trogocytosis [
22,
40].
Trogocytosis is a mechanism of rapid transfer of membranes and membrane-associated proteins between interacting cells. Membrane transfers have been observed mainly between immune cells [
41], and in particular for HLA molecules. Although trogocytic transfers are common and rather easy to observe experimentally, the mechanisms that underlie them and the molecules directly responsible for transfer remain unclear. In particular, trogocytosis was shown to be dependent on MHC-TCR interactions or costimulatory molecules (for review, see [
42]). Our laboratory described the trogocytic transfers of HLA-G and its receptor ILT2 [
22,
40,
43]. In these studies, we could not identify which molecules were responsible for APC-to-T or tumor-to-NK, or tumor-to-APC trogocytosis. However, we did demonstrate that in APC-to-T trogocytosis, multiple molecules transferred to various degrees. Among these, MHC molecules and CD86 transferred the most, a finding that is compatible with the transfer of membrane-bound organized molecular clusters such as immune synapses or lipid rafts [
40,
44]. After trogocytic acquisition, the transferred molecules retain their original function and it is now well established that trogocytosis of the HLA-G molecule can impact the outcome of an immune reaction by conferring protection to the HLA-G acceptor cell, and by conferring sensitivity to inhibition by HLA-G through the transfer of its receptor ILT2 [
22,
40,
43,
45,
46]. Most of these studies have been performed using immune cells as membrane acceptor cells (T cells, B cells, monocytes), and tumor cell lines as membrane-donor cells [
22,
40,
46,
47]. In two studies [
22,
48], we showed that the NK cell line NKL was able to efficiently acquire HLA-G1-containing membranes from an allogeneic HLA-G1-expressing melanoma line through trogocytosis [
22]. These experiments demonstrated that hematological tumor cells may also be trogocytic, although it cannot be assumed that tumor-to-tumor transfers occur by the same mechanisms as APC-to-T cell or tumor-to-effector transfers. Other reports showed that trogocytosis occurred between autologous cells [
43,
46-
48]. Taken together, these data suggest that tumor cells may exchange membranes and proteins by trogocytosis among each other.
In the present report, we demonstrate in vitro and ex vivo that tumor cell lines of immune origin, and tumor cells from malignant hemopathies such as lymphoma or leukemia malignancies, possess trogocytic capabilities: they can acquire membranes and the membrane-bound immune escape molecule HLA-G1 from their surroundings and from each other.
Discussion
Trogocytosis, i.e., the capability to acquire membranes and membrane-associated molecules from a cell by another, has been described for non-tumor immune cells and shown to impact their immunological functions. In this work, we investigated whether tumor cells of immune origin were also trogocytic. Working with hematological cell lines and freshly isolated hematological tumors, we demonstrate here that trogocytic function was indeed preserved in hematological tumors, that membrane exchanges happen in autologous situations, and hence are likely to occur in vivo.
Our previous studies on trogocytosis between tumor cell lines and non-tumoral immune cells showed that effector T and NK cells were capable of acquiring membranes and HLA-G1 protein from HLA-G1-expressing tumor cell lines, and that trogocytosis or HLA-G1 acquisition was not dependent on HLA-G1. Indeed, blocking the HLA-G1/ILT2 interaction did not prevent HLA-G1 from being transferred from tumor to immune cells. Rather, HLA-G1 transferred because it was part of transferred membrane patches whose trogocytosis was dependent on other molecules that could not be identified, even though CD28/CD86 or MHC-TCR interactions were ruled out [
40]. Fixation of the donor cell’s membrane using PFA was previously shown to inhibit HLA-G1 trogocytic transfer from tumor cells to activated T cells [
40], and this was confirmed here for tumor-to-tumor transfers (Additional file
4: Figure S3).
One of the most striking features of trogocytic transfers is that the acceptor cells may display membrane proteins that they do not express themselves. This has been demonstrated for T and NK cells in the context of HLA-G1 trogocytosis [
22,
40]. As a consequence, the acquired proteins (e.g., HLA-G1) are displayed on the acceptor cells for a limited time only. Previous studies from our laboratory showed that HLA-G1 that had been acquired by trogocytosis was no longer present at the cell-surface after 24 h for T and NK acceptor cells, and 6 h for monocytes [
22,
40,
47]. Experiments performed in the context of the present study confirmed that the life time of acquired HLA-G at the surface of acceptor tumor cells is limited, and of the order of 24 h (data not shown).
We previously reported that HLA-G does not direct its own transfer, and seems to transfer passively from cell to cell. This relates to the fact that many proteins are transferred during trogocytosis, including costimulatory molecules such as CD86 or CD54, receptors such as ILT2, and MHC class I and class II molecules [
40]. Many more have not been investigated, and it is likely that most of them also transfer passively. This does not mean that any molecule can transfer by trogocytosis. Indeed, we reported that ILT3 or CD4, CD8, CD14, CD19, and here CD5 are not concerned by trogocytosis, which suggests that trogocytic exchanges are not random processes, but regulated ones.
This begs the question of what the molecules that mediate trogocytic transfers are. Previous work has focused on trogocytosis between antigen-presenting cells and T cells, or between target cells and cytotoxic T or NK cells. In most of these cases, trogocytic transfers were associated with antigen-specificity (T cells) or function (i.e., target recognition, NK cells) [
22,
52]. However, previous studies show that trogocytic transfers are only partially dependent on antigen-specificity: for instance, in case of antigen-specific interaction between non-activated donor and acceptor T cells, the TCR is involved in the transfer process, but when T lymphocytes are activated, trogocytosis occurs independently of TCR engagement, and with a higher efficiency [
40]. In another study, we demonstrated trogocytosis between autologous cells in the absence of foreign antigen [
43], and in the present report, we show that B and monocytic tumor cell lines are capable of trogocytosis in autologous conditions and in the absence of antigen, strengthening the notion that trogocytosis is at least partially independent of antigen-specificity. Nonetheless, the molecules that are responsible for trogocytosis remain elusive.
B-CLL progressively lost their trogocytic capability upon culture, sometimes as fast as in 24 h (data not shown). This indicates that the trogocytic capability of tumor cells is highly dependent on microenvironmental factors, which means that the levels of trogocytosis might be higher in vivo than what we observed ex vivo. As stated above, we do not know which molecules may be responsible for tumor-to-tumor membrane exchanges. It is therefore difficult to predict which microenvironment factors may modulate the expression and/or function of these molecules and, down the line, the extent of trogocytic exchanges. Keeping B-CLL in culture medium supplemented with autologous serum rather than FCS allowed B-CLL cells to retain their trogocytic capabilities longer (not shown). Thus, microenvironmental factors supporting trogocytic capability might be cytokines, growth factors, stress factors, and more generally molecules present within the serum, rather than nutrient deprivation or microenvironmental conditions such as hypoxia for instance.
Trogocytic exchanges between tumor cells
in vivo are relevant because we demonstrated that the transferred proteins might be functional in their new cell hosts. This is even more likely if donor and acceptor cells are cells of the same tumor. We already demonstrated
in vitro that trogocytosis could contribute to tumor escape. Indeed, HLA-G1 acquisition by T cells from HLA-G1-expressing tumor cells rendered them temporarily suppressive. Recently, this hypothesis was strengthened by Brown et al. [
51], who confirmed our previous
in vitro results and demonstrated in an all-autologous
ex vivo system that T cells can acquire HLA-G from autologous multiple myeloma tumor cells, turning them into regulatory cells. Here, we also show that T cells can acquire membranes from autologous B-CLL
ex vivo, and confirm that such transfers remain rare events. However, our results, obtained with the same blood samples, show that in B-CLL, tumor-to-tumor membrane exchanges are prominent. The meanings of such exchanges remain to be explored in depth, but already the interest for the tumor of transferring HLA-G between its constituent cells can be envisioned. For instance, it is known that some but not all B leukemia cells and multiple myeloma cells from the same tumor can express HLA-G [
38,
53-
56], which is capable of protecting them from NK cell cytolysis, and thus constitutes an immune escape mechanism [
57]. Therefore, by transferring HLA-G and possibly other immune-inhibitory molecules, tumor cells may share some of their immune escape strategies, as our
in vitro model showed here. For a tumor whose cellular constituents heterogeneously express immune escape molecules such as HLA-G, trogocytic sharing presents definite advantages, such as (i) conferring extra-protection to those cells which do not express the transferred protein, and (ii) diversifying the array of immune escape strategies that each cell uses. This would strengthen the overall resistance of the whole tumor to anti-tumoral immunity. This also means that blocking trogocytosis might constitute a new way to restore anti-tumor immunity, and emphasizes the need to precisely characterize the molecular mechanisms that are involved in trogocytic exchanges.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JL and JC designed the experiments and wrote the manuscript; MD, MS, VR, and BA performed the experiments; and EDC designed the experiments and wrote the manuscript. All authors read and approved the final manuscript.