Despite their importance in maintaining healthy pregnancy, the control of maternal HCMV infection and spreading by dNK cells is not yet fully understood. Our study is the first to assign a critical role to dNK cells in controlling maternal HCMV infection and in limiting its spreading to fetal tissues through their capacity to acquire potent cytotoxic activity when in contact with infected decidual cells. During normal pregnancy, the majority of dNK cells are CD56brightCD16neg. They secrete a large panel of cytokines and chemokines that are necessary for placental development. We demonstrate that dNK cells undergo phenotypic and cellular changes that allow them to recognize and kill autologous HCMV-infected cells in a FasL- and TRAIL-independent manner.
Immunological synapse formation is a crucial step for the delivery of lethal hits by effector cells. Rapid re-localization of the MTOC is needed for the trafficking and the polarization of lytic granules to the IS 
. We show that although dNK cells recognize and engage IS with HCMV-infected cells very rapidly, they require longer exposure time in order to degranulate and exert the cytotoxic effector function. The delay to unleash dNK cell cytotoxic effector function might correspond to the time necessary for dNK cells to mature and acquire necessary functional changes to exert cytotoxicity. However, we cannot exclude that HCMV-infected fibroblasts provide weak signal to induce fast degranulation or that decidual fibroblasts have an inherent resistance to cytotoxic granule mediated cell death.
Mechanisms that prevent dNK cell cytotoxicity are not completely understood. Even though dNK and pNK cells exhibit similar expression levels of cytotoxicity encoding genes 
, under healthy conditions dNK cells are tolerant to semi-allogeneic fetal trophoblasts. Although mechanisms that control cytotoxicity are not well established, they may include strong interactions of inhibitory receptors with their cognate ligands expressed by fetal trophoblast, production of VEGF-C by dNK cells and/or expression of anti-apoptotic proteins (XIAP) by target cells 
. The lack of dNK cell cytotoxicity can be reversed, at least in vitro
, after exposure to cytokines such as IL-5 and IL-18 or upon engagement of specific activating receptors 
. Here we show that HCMV infection provides the necessary activating signals to trigger dNK cell cytotoxicity. The fact that dNK cells killed heterologous targets from a different donor further emphasizes the intrinsic ability of these cells to kill when they are exposed to the right activating signals. Our observation that dNK cells did not kill semi-allogeneic trophoblasts but killed HCMV-infected autologous fibroblasts highlights their plasticity and their specific ability to respond to HCMV infection.
In contrast to pNK cells, very little is known about dNK cell cytotoxicity as these cells are mainly cytokine and chemokine producers 
. We demonstrated that under HCMV-infectious conditions, a significant fraction of dNK cells that are CD56bright
rapidly dampened down their CD56 expression level and acquired CD16 expression. These changes are most probably due to the acquisition of cytotoxic function. Several NKRs have been involved in pNK cell cytotoxicity 
. For instance, efficient control of HCMV infection involves NKG2D receptor and can be associated with the emergence of NKG2C+
subset that contribute to long term protective immune response 
. Exposure of dNK cells to HCMV-infected fibroblasts resulted in an increased NKG2C+
expression without major changes in NKG2A expression. The role of other receptors in NK cell response to HCMV is not completely understood. HCMV is able to decrease a plethora of key receptor-ligand interactions that are involved in NK-cell response. By contrast to changes in pNK cell repertoire 
, opposite effects were observed for NKp44 and NKp46 receptors while no changes were observed for NKp30 receptor. These observations further highlight differences between dNK and pNK cells modi operandi
during HCMV infection.
Since the nature of HCMV-induced cellular ligands is not known, we took advantage of NKR-Fc chimeric receptors to analyze the expression of NKR ligands on decidual fibroblasts. Although some variations were observed amongst different decidua basalis, we found that decidual fibroblasts constitutively express ligands for NKp44 and NKG2D while they barely express ligands for NKp30 or NKp46. HCMV infection induced NKp30L and resulted in significant decreases of NKp44L and NKG2DL but did not affect the expression of NKp46L. These findings suggest that HCMV infection interferes with the expression level of activating receptor ligands even if some of them are of cellular rather than virally induced.
Using chimeric proteins, we demonstrated that NKp44 receptor plays an inhibitory function in dNK cell cytotoxicity. dNK cells might express an inhibitory isoform of NKp44 receptor as a result of NCR2
alternative splicing as it has been recently demonstrated for NCR3
. Alternatively, NKp44L expressed on decidual fibroblasts might participate to uncoupling of activating adaptor molecules thus promoting an inhibitory profile. However, the expression of an inhibitory isoform is the most likely explanation since dNK cells constitutively express the NKp44 receptor. It has been clearly demonstrated that HCMV maintains an inhibitory status either by preventing the cell-surface expression of NKG2D activating ligands 
or by UL40-mediated up-regulation of HLA-E or MHC-I like surrogates molecules expression. Although, there are some discrepancies between our two observations, namely decreases of NKG2DL and acquired cytotoxicity through NKG2D receptor, it is possible that decreases in NKG2DL are selective resulting in the expression of high affinity ligands. Alternatively, co-engagement of other activating receptors is sufficient even if there is less NKG2D ligands.
Further studies are needed to identify NKG2DL that are expressed on decidual cells. Discovery of such ligand and the characterization of specific receptor-ligand interactions that mediates dNK cellular cytotoxicity will help uncover potential therapeutic target that, when activated in vivo, can limit viral spreading and/or prevent congenital disease.
Previous investigations demonstrated that both classical and non-classical MHC-I molecules have been targeted by HCMV evasion strategies. By contrast to human fetal foreskin fibroblasts and fibroblastic cell lines 
, HCMV-infection resulted in decreased cell surface HLA-E molecules without affecting the total amount of proteins in decidual fibroblasts. The difference between the two cellular systems might reside in the fact that decidual fibroblasts express substantial amounts of HLA-E at the steady state. In decidual fibroblasts, HCMV might interfere with the stability of cell surface HLA-E molecules by impairing rapid protein export or by increasing intra-cellular retention. The inhibitory profile observed upon blockade of HLA-E in HCMV-infected fibroblasts further support the involvement of CD94/NKG2C or possibly CD94/NKG2E activating receptors, both greatly expressed by dNK cells 
. In this context, HCMV peptides might play a critical role in promoting the recognition of HLA-E by activating members of CD94/NKG2C and CD94/NKG2E receptors thus increasing susceptibility of decidual fibroblasts to dNK cell cytotoxicity at early times of infection as it has been shown previously for pNK cells 
. It will be very interesting to investigate whether late HCMV infection is responsible for similar changes and whether specific HCMV peptides play roles in the sequential changes in dNK cell function.
In parallel to these changes in NK cell receptor, dNK cells acquire de novo
expression of MHC-II DR molecules. This potential acquisition of an APC-like phenotype during the course of HCMV immune response might play a crucial role in initiating a cross-talk with neighboring immune cells, including CD4+
T cells. Indeed, within the fetal-maternal interface, dNK cells are in close proximity with decidual CD4+
T cells. Expression of MHC-II DR antigens might be necessary for dNK cell activation and for shaping up the adaptive immunity 
. However, further investigations are needed to demonstrate whether the expression of MHC-II molecules is associated with the acquisition of APC capabilities and HCMV antigen presentation.
It is very intriguing that only few cytokines and chemokines varied in the presence of HCMV infected fibroblasts. HCMV infection induces IL-6 secretion most probably through the expression of the viral-encoded chemokine receptor US28 and the activation of the IL6/STAT3 signaling pathway 
. Interestingly, IL-6 was further increased when dNK cells were in contact with HCMV-infected fibroblasts, most probably through a paracrine effect on dNK cells. sICAM-1 was also increased under HCMV conditions. Previous reports suggested that IL-6 down-regulates the production of several soluble factors 
, while sICAM-1 increases have been correlated to HCMV reactivation 
. Both IL-8 and IP-10 are necessary for trophoblast migration as these cells express a panel of receptors allowing them to respond to these chemokines 
. By lowering the level of IL-8 and IP-10, dNK cells might reduce trophoblast invasion and prevent viral spreading from decidual stroma to fetal tissue or be partially responsible for fetal damages. Remarkably and in sharp contrast with pNK cell response to viral infection 
, there were no changes in secretion levels of cytokines such IL-12, IL15, type I IFN, TNFα or IFN-γ that are all known to regulate NK cell function. Moreover, it is possible that changes in dNK cell secretome create the necessary inflammatory environment that will favor the recruitment and the initiation of anti-HCMV adaptive immune response.
We demonstrate that during HCMV infection, there is a bias of the inflammatory environment in the decidua basalis
. dNK cells seem to lose their “decidual status” and become killers in order to limit viral infection. Exposure to HCMV infection can imprint dNK cell receptor repertoire towards killer activity. We demonstrate that NKG2D, NKG2C/E activating receptors play a crucial role in dNK cell cytotoxic response against HCMV-infected fibroblasts. The fact that dNK cells are able to infiltrate HCMV-infected tissue in vitro
and engage immunological synapse-like structures within the infected placentas in situ
strongly suggest that dNK cells are key players in controlling HCMV infection and spreading during pregnancy. To our knowledge, we provide for the first time evidence for the involvement of dNK cells in clearing HCMV infection. In fact, we clearly show that dNK cells that are present only in the decidua basalis
during healthy pregnancy are in contact with HCMV-infected fetal tissue in vivo
. It is possible that upon activation there is an increased dynamic of dNK cells allowing them to rich fetal site, which is normally devoid of maternal immune cells, and kill HCMV-infected cells. Recent reports have clearly linked the ability of NK cells in controlling HCV replication and liver fibrosis to specific soluble factor secretion and/or specific activating receptor expression 
. Future studies, with large cohort of placentas from medical termination of pregnancy due to congenital HCMV infection, will be necessary to clarify the dynamic of dNK cell activation in vivo
as well as the pivotal role of soluble factor secretion in mounting proper anti-HCMV responses and limiting virus spreading.
In conclusion, our data shed new light on the plasticity of dNK cells and provide evidence for a correlation between phenotypic changes and functional anti-viral response. We have demonstrated the ability of dNK cells to exert anti-viral effector functions in vitro and to infiltrate HCMV infected tissues both ex-vivo and form immune synapse like-structures in vivo. Careful investigations of dNK cell status in vivo in larger cohorts of HCMV+ termination of pregnancy will be required to see whether this predicts clinical outcome. Understanding mechanisms that regulate switch in dNK cell immune tolerance will help us discover key factors/pathways that are involved in the immunopathology of HCMV infection during pregnancy and design strategies to limit congenital infection.