It is well accepted that NK cells play a key role in the innate immune system's surveillance against tumors [35, 36]. Tumor surveillance requires the interaction between NK cell triggering receptors and their ligands. NK cells express major histocompatibility complex (MHC) class I-specific receptors, such as killer cell immunoglobulin-like receptors (KIRs), which may activate or inhibit NK-cell effector functions . Inhibitory KIR receptors allow NK cells to recognize cells that express self-MHC-I molecules and subsequently inhibit NK cell effector functions . With the loss of these inhibitory interactions; however, target cells become susceptible to NK cell activity via the triggering of NK cell-activating receptors [39, 40]. One of these receptors is NKG2D, which recognizes two different groups of MHC-I-like molecules, the MIC and the ULBP/RAET1 families. These ligands have been found to be expressed in a wide range of different solid tumors and in some normal tissues [15–18, 41]. The recognition of MICs or ULBPs by NKG2D induces NK cell activation; however, recent studies have shown that NKG2D-ligands are also capable of inducing the down-modulation of NKG2D, and subsequent loss of NK activation, via persistent cell contact between NKG2D and NKG2D-ligands .
In this study we have shown that cervical cancer cell lines differentially express both families of NKG2D-ligands. That is, MICA (data not shown) and ULBP2 are preferentially expressed by HPV-positive cell lines (HeLa and SiHa), whereas the HPV-negative cell line, C33A, preferentially expresses MICB (data not shown). Interestingly, ULBP4 was only expressed by the non-tumorigenic HaCaT keratinocytes. This differential expression of the NKG2D-ligands raises the question of whether HPV infection could play a role in their regulation. Several studies have shown that some viruses, such as human cytomegalovirus and herpesvirus, are capable of modulating the expression of NKG2D-ligands as an immune evasion mechanism. Both viruses produce immunoevasins that target the NKG2D-ligands MICA, MICB, and ULBP1-3. The intracellular retention of these ligands inhibits their cell surface expression [43–45]. At this time, it is unknown whether HPV possesses similar mechanisms for the evasion of NKG2D-mediated NK responses.
The oncoproteins E6 and E7 of high-risk HPV are capable of interfering with the binding of certain immunoproteins, such as transcriptional factors (IRF3, IRF1) involved in innate immune responses [46–48]. A recent study showed that the HPV16 E5 oncoprotein intracellularly binds the CD1 molecule and inhibits its expression in a HPV16 E5-transfected cervical cancer cell line . This finding raises the question as to whether HPV-derived proteins might also be capable of binding other MHC class I-like molecules, such as NKG2D-ligands. Therefore, it will be interesting to design co-localization experiments to determine if any of the HPV oncoproteins could also intracellularly sequester NKG2D-ligands, leading to functional NK cell and T cell defects.
It has been previously demonstrated that several tumors expressing NKG2D-ligands induce the down-modulation of NKG2D on NK cells and T cells. In the present study, we co-cultured different cervical cancer cell lines and non-tumorigenic HaCaT keratinocytes with an NK cell line, NKL, and observed a very strong down-modulation of NKG2D expression after contact with all cervical cancer cell lines (SiHa, HeLa, and C33A). Like NKL cells, fresh NK cells were also susceptible to down-modulate NKG2D after the contact with tumor cells. This reduction in cell surface expression of NKG2D is similar to recent reports showing that this phenomenon can be induced by other tumor cells .
The down-modulation of NKG2D, by binding to either soluble or transmembranal forms, has been shown to have functional implications for NK cells and T cells, leading to impaired NK cell activity and poor cytotoxicity [21, 42]. In this study, we evaluated whether the negative regulation of NKG2D expression on NKL cells had functional effects on the cytotoxic activity of these cells. We found that NKL cell cytotoxic activity against target cells (K562) diminished after contact with the HPV-positive cervical cancer cell lines, SiHa and HeLa, but not after contact with HPV-negative C33A cells and non-tumorigenic HaCaT keratinocytes. Due to the fact that C33A cells caused NKG2D down-modulation, we would have expected to find impairments in NK cell activity. However, co-culture of NKL cells with C33A cells led to increased cytotoxic activity against K562 cells when compared against both the control and HPV-infected cells. At this point, it is important to note that C33A cells expressed a different pattern of NKG2D-ligands than the other cell lines. Notably, MICA (data not shown) and ULBP2 were lower than SiHa and HeLa cells, while MICB levels were higher in C33A cells (data not shown). A recent study showed that transmembranal MICB is capable of inducing NK cell activation via the NKG2D receptor and that this ligand does not participate in negative modulation of NK cell activity . Thus, it is logical to speculate that the differences in the patterns of NKG2D-ligand expression, specifically the higher levels of MICB, could be responsible for changes in the cytotoxicity of the NKL cells against the K562 target cells. However, we cannot disregard the possible role of soluble ligand expression. We have recently found that soluble MICB is also greatly increased in C33A cells compared with SiHa and HeLa .
One important further experiment will be to knock-down NKG2D expression in NKL cells and to verify whether the loss of this activating receptor explains the loss of NKL-mediated cytotoxicity against K562 targets. This experiment will be important in order to clarify whether other interactions between the tumor cells and NKL cells might also be responsible for the loss of cytotoxicity observed in our experimental setting.
In order to clarify the role of NKG2D in NKL cytotoxic activity, we performed experiments using a blocking anti-NKG2D monoclonal antibody. Intriguingly, the addition of the blocking antibody to NKL cells previously exposed to SiHa cells resulted in a significant diminution of the previously observed cytotoxicity. This could be due to the fact that this cell line is unique (among the cell lines in this study) in its strong expression of the NKG2D ligand ULBP2. While the antibody blockade of NKG2D also significantly inhibited cytotoxicity in unexposed and in NKL cells previously exposed to the other cancer cell lines, this result was not nearly as marked as the result seen with SiHa pre-exposure. Certainly, this inhibition was only partial, which indicates that other activating receptors, such as NKp30 or NKp46 (moderately expressed by our NKL cells) might also be contributing to the tumor cell lysis or that the numbers of anti-NKG2D antibody (at the blocking concentrations used) were not sufficient to block all the receptors remaining on the cell membrane. In summary, it will be important to conduct experiments with blocking antibodies against other important activating receptors, including NKp30, NKp46, and DNAM-1, which might lead to tumor cell-NK cell interactions as well.