Since the discovery of "null" killer cells more than 30 years ago, there has been a large body of evidence implying the participation of NK cells in the recognition and lysis of tumor cells [37–39]. It is currently known that the activity of NK cells is delicately controlled by a balance between inhibitory and activating receptors . NKG2D, one of the best characterized NK cell activating receptors, can promote tumor lysis upon recognition of MICA, MICB or the ULBPs. [41–43].
Up-regulated expression of MICA/B on the tumor cell surface has been considered a "danger signal" in order to activate NKG2D-expressing cells and promote anti-tumor immunity [4, 13, 42] However, numerous studies have demonstrated tumor evasion through metalloprotease-induced proteolytic release of MICA and MICB from the cell surface, which provokes down-regulation of NKG2D in NK and T cells [44–47]. Recently, we demonstrated increased soluble MICA levels in sera from patients with cervical cancer and precursor lesions as compared with healthy donors . However, in that study we did not test a correlation between soluble MICA amounts and MICA expression in cervical tissues. Therefore, in the present study we performed a systematic analysis of the MICA and MICB expression in human cervical cancer cell lines. The majority of malignant cervical tumors are associated with infection by HPV-16 and HPV-18 . Therefore, we chose SiHa and HeLa cells for our study, which are infected with these virus types. We also included an HPV-negative cervical cancer cell line (C33-A), as well as an HPV-negative, spontaneously immortalized human keratinocyte cell line (HaCaT) with a highly preserved differentiation capacity comparable to that of normal keratinocytes . In this study we used different clones of antibodies against MICA/B (clone 6D4), MICA (clone 159227) and MICB (clone 236511). Interestingly, we observed starkly different MFI with the different antibodies. Of particular interest is the difference between the MICA/B MFI spread between SiHa and HeLa (189.54:116.71) that is not at all seen in the MICA staining of those cells, which were essentially identical (325.44:368.39). From this, we conclude that the common anti-MICA/B antibody must be binding different epitopes with different affinities than the individual antibodies. It is tempting to imagine that the individual MICA antibody binds an epitope that is conserved between HeLa and SiHa cells, while the MICA/B antibody binds an epitope that might be slightly different between the two cell lines, leading to different binding affinities. Our data on the preferential cell surface expression of MICA over MICB in SiHa cells are in agreement with a recent study examining activating NK ligands in various cell lines (including SiHa) and in tumor cells from patients with cervical cancer .
Due to the fact that the tumor cell lines tested in our study showed a significant cell surface expression of MIC molecules, the question is still open as to whether or not these tumor cells are able to escape from the normally efficient NK cell activation through the MIC/NKG2D pathway. For instance, sustained expression of NKG2D ligands has been shown to promote downregulation of the NK cell functions in transgenic mice constitutively expressing Rae-1ε . One mechanism by which the persistent MIC ligand expression could affect NK cell activity is by shedding MIC molecules from the tumor cell surface. In other words, the constitutive MIC expression on the tumor cell surface might be a continuous source of soluble MIC molecules, which are thought to engage the NKG2D receptor and cause its internalization and subsequent lysosomal degradation . The high levels of soluble MICB found in the present study suggest that this ligand might play a different biological role than MICA. In support of this is the fact that elevated soluble MICB correlated with disease activity in patients with multiple sclerosis during relapses while soluble MICA did not show any association with the disease; rather, the levels of this ligand were similar to healthy controls . Therefore, an issue of vital importance will be to address if both soluble MICA and MICB share the same biological function in patients with HPV-associated tumors and to determine whether MICB, which was found in a higher concentration than MICA in the supernatants of cervical cancer cell lines, could promote NKG2D downmodulation in NK cells of cervical cancer patients. It will be also interesting to investigate if MICA and MICB show different mechanisms of posttranslational control; for instance, Agüera-González et al., showed recently that MICB has a short time of residence at the plasma membrane and demonstrated that MICB shedding was one of the mechanisms that contributes to the rapid loss of ligand from the cell surface . Thus, if high levels of soluble MICB are present in the serum of cervical cancer patients, and this contributes to tumor immune escape, targeting of this molecule could be a promising alternative to improve tumor immunosurveillance in these patients.
The Real-Time RT-PCR results for MICB are especially interesting in light of the flow cytometry and ELISA data. While MICB transcripts were slightly lower than those for MICA in the HPV-positive cell lines, C33-A cells showed the opposite. The same finding could be observed by cell surface staining (with apparent MFIs of 84.60 versus 24.67 and 40.06 for SiHa and HeLa, respectively) and also in the ELISA experiments for soluble MICB in the culture supernatant. Despite the fact that MICB appears to be less polymorphic than MICA, polymorphisms in the MICB promoter, with important variations in transcription rates, have been described . If this is the case here, then these polymorphisms might partly explain the higher transcriptional rate that we observed in the C33-A cell line.
More importantly, the results with etoposide demonstrated that while MICB mRNA level rose slightly after treatment, MICA mRNA was strikingly upregulated, especially in non-tumorigenic HaCaT cells (68-fold increase). To follow up on this result, we also measured cell surface expression of MICA and MICB in HaCaT and HeLa cells after 4 h of etoposide treatment, but no significant difference was observed (data not shown). A possible explanation for this discrepancy might be the short time frame of the experiment. However, the quantitative differences observed in the mRNA levels of MICA and MICB genes in response to etoposide treatment suggest that both genes could be regulated in different ways. Indeed, a recent study that analyzed the architecture and function of the promoter of the MIC genes provided evidence for differential regulation of MICA and MICB .
Taken together, our results suggest that in spite of sharing a high degree of homology at both genomic and structural levels, MICA and MICB are differentially regulated at the transcriptional level and by cleavage at the cell surface in response to varying danger signals. Additionally, our data might point to roles for both ligands in the escape from immunosurveillance by tumor cells: sustained over-expression of MICA at the cell surface of HPV-positive cells, which could promote downregulation of the NK cell functions; and the shedding of MICB in HPV-negative cervical cancer cells, which could bind to NKG2D receptors and outcompete cell-surface activating ligands. Modulating the cell surface expression or targeting the proteases that mediate shedding of NKG2D ligands may open a new approach for the treatment of cervical cancer.