In mice, the CD40-CD40L (CD154) interactions participate in driving the activation of NK cells by LPS- or Lacto-N-fucopentaose III-activated macrophages in vitro (Table 1) (Scott et al., 2004; Atochina and Harn, 2005).

In human, the presence of high doses of LPS changes the phenotype of macrophages by inducing the expression of various ligands of the activating receptor NKG2D: UL16-binding proteins (ULBP1, ULBP2, and ULBP3) and MHC class I-related chain A (MICA) (Nedvetzki et al., 2007). Human NK cells which are in contact with macrophages or LPS-activated macrophages express increased levels of NKG2D. However, NKG2D expression in NK cells seems to be less induced by LPS-activated macrophages than unactivated macrophages (Nedvetzki et al., 2007). This phenomenon could be similar to the observation that prolonged exposure to its ligands induces a downregulation of NKG2D on NK cells (Ogasawara et al., 2003; Hamerman et al., 2004; Coudert et al., 2005; Kloss et al., 2008; Lapaque et al., 2009). These receptor-ligand interactions induce the lysis of macrophages stimulated with high doses of LPS, by NK cells via NKG2D (Nedvetzki et al., 2007). Thus, NK cells specifically kill PBMC-derived macrophages stimulated with high doses of LPS probably to eliminate overstimulated macrophages to avoid endotoxic shock. On the contrary, LPS-stimulated microglia (nerve system macrophages) is less susceptible to NK cell-mediated cytotoxicity compared to resting microglia. The killing of resting microglial cells by NK cells is reduced by NKG2D-specific blocking Abs, meaning that NKG2D is involved in this process. Furthermore, NKG2D ligands are expressed constitutively by microglia cells in vitro. However, in LPS-stimulated microglia this expression is down-regulated. By this mechanism, activated microglial cells which are not lysed by NK cells could be able to present antigens to infiltrating T cells and initiate a limited immune response in the brain (Lunemann et al., 2008). Another study on monocytes described that the killing of MICA-expressing monocytes by autologous NK cells is not observed neither in the presence nor in the absence of LPS (Kloss et al., 2008). In presence of LPS, MICA is up-regulated on monocytes. Differences between these various cell types show that macrophages, microglia, and monocytes are not equally susceptible to NK cell-mediated killing after LPS-stimulation. This could be the result of a variation in the level of inhibitory and activating signals from these target cells to NK cells. This might be explained by the difference of NKG2D engagement and different levels of the counterbalance due to MHC class I binding of inhibitory NK cell receptors. Indeed, in presence of LPS, PBMC-derived macrophages are not associated with an MHC class I up-regulation while this is the case for microglia and monocytes (Nedvetzki et al., 2007; Kloss et al., 2008; Lunemann et al., 2008). NKG2D is also required in other organs but the nature of its ligand differs depending on the type of infection. Thus, poly I:C stimulated uterine macrophages up-regulate MICA and activate IFNγ production of uterine NK cells through the recognition of NKG2D (Basu et al., 2009). Likewise, alveolar macrophages infected by M. tuberculosis are lysed by NK cells via the interaction between NKG2D and ULBP1 (Vankayalapati et al., 2005). Another NKG2D ligand, MICB, is expressed when human macrophages are infected with influenza A or Sendai viruses and the infected macrophages are likely to promote IFN-γ release by NK cells (Siren et al., 2004).

In mice, similar pathways can be found to initiate the activation of NK cells. Thus, peritoneal macrophages treated with poly I:C increase NKG2D expression in NK cells (Zhou et al., 2012). In response to Toll-like receptor (TLR) ligands like LPS, lipopeptide Pam₃CSK₄, zymosan, or poly I:C, retinoic acid early inducible-1 (RAE-1) is induced in macrophages (Hamerman et al., 2004). After contact with RAE-1, the NKG2D receptor triggers activation of NK cell cytotoxicity-related molecules, TNF-related apoptosis-inducing ligand (TRAIL), perforin and FasL and therefore the induction of NK cell cytotoxicity toward tumor cells and the secretion of IFN-γ (Hou et al., 2009; Zhou et al., 2012). RAE-1-NKG2D interactions are also involved to stimulate the release of IFN-γ by NK cells during the crosstalk with monocytic myeloid derived suppressor cells (MDSC) (Nausch et al., 2008). The same RAE-1-NKG2D engagement is observed between bone marrow-derived macrophages and NK cells in response to B. anthracis spores (Klezovich-Benard et al., 2012). Thus, the NKG2D pathway has its importance to confer protection against infections, as it indirectly detects a wide variety of pathogens through the recognition of various NKG2D ligands.

In human, the release of IFN-γ by NK cells can be triggered by the interaction of the receptor 2B4 with the ligand CD48 present at the surface of PBMC-derived macrophages (Nedvetzki et al., 2007; Bellora et al., 2010; Romo et al., 2011). This engagement depends on how macrophages have been prearmed. Macrophages stimulated with high doses of LPS activate NK cytotoxicity rather than cytokine release. When the latter have not been activated or activated with low doses of LPS, the system passes through the 2B4-CD48 pathway. The 2B4-CD48 engagement has other consequences like the increase of the expression of 2B4 and the proliferation of NK cells (Nedvetzki et al., 2007). Thus, after contact with macrophages, 2B4 induces NK cell cytokine secretion and proliferation. However, depending on the cell types, the signaling triggered by 2B4 could also result in activation of NK cytotoxicity, which could be due to a different 2B4 isoform expression (Mathew et al., 2009; Kim et al., 2010). When macrophages are infected by human cytomegalovirus, the expression of CD48 is down-regulated, suggesting the existence of a potential viral immune escape strategy targeting the 2B4 pathway (Romo et al., 2011).

One group has investigated the structure of the immune synapses between macrophages and NK cells in the context of LPS-stimulation (Nedvetzki et al., 2007). Images of the macrophage-NK cell conjugates show an accumulation of F-actin at the site of contact which is more frequent when macrophages are stimulated with high doses of LPS. The accumulation of F-actin at cytolytic synapses between NK cells and LPS-stimulated macrophages happens at the NK cell side, while F-actin accumulates at the macrophage side in the non-cytolytic synapses in absence of LPS. Thus, the NK cell-macrophage communication is different depending on whether macrophages are treated or not with high doses of LPS. Indeed, the activating receptor complex NKG2D/DAP10, the activating adaptor protein CD3ζ as well as ICAM-1 are recruited to the NK cell immune synapse by macrophages stimulated with high doses of LPS and very rarely by unactivated macrophages, showing a key role of these molecules in triggering cytotoxicity. On the other hand, a large fraction of 2B4 clustered to the immune synapse of macrophages and macrophages activated with low doses of LPS, confirming the concept that this receptor is important for the macrophage-mediated NK cell proliferation and IFN-γ secretion. Two distinct NK cell synapses are defined, one activating the proliferation and cytokine secretion, by the recruitment of 2B4, and the other one triggering lysis through the activation of NKG2D (Nedvetzki et al., 2007).

In human, the susceptibility of macrophages to be lysed by autologous NK cells can also be mediated by the activating receptors NKp46 and DNAM-1. Thus, blocking of NKp46 and DNAM-1 decreases NK cell cytotoxicity against cytomegalovirus-infected macrophages and LPS-stimulated macrophages derived from PBMC. NKp46 is also involved in the lysis of M. tuberculosis-infected alveolar macrophages (Vankayalapati et al., 2005; Bellora et al., 2010; Romo et al., 2011). In the case of resting microglial cells, only NKp46 engagement is implicated in the killing (Lunemann et al., 2008). Furthermore, the coculture of NK cells and PBMC-derived macrophages increases the expression of NK activating receptors like NKp30, NKp44, and NKp46 (Nedvetzki et al., 2007). In regard of the different data reported, it seems that NKp46 participates mainly to NK cell cytolytic activity whereas 2B4 is rather involved in the induction of IFN-γ production. DNAM-1 plays quite a dual role, as it seems to participate to the activation of both cytotoxicity and IFN-γ secretion.

In contrast to human, the expression of mouse NKp46 and DNAM-1 is not increased in the presence of peritoneal macrophages treated with poly I:C (Zhou et al., 2012). However, in another context, NCR1 (mouse NKp46) is shown to be involved in the macrophage-NK cell crosstalk, as alveolar macrophages infected in vitro by S. pneumoniae mediate lung NK cell activation (CD107a expression and IFN-γ secretion) in part through the NCR1 receptor pathway (Elhaik-Goldman et al., 2011).

Other molecules, like CD44 and ICAM-1, could be involved in macrophage-NK cell interactions, but their precise role remains to be determined (Nedvetzki et al., 2007).

In human, PBMC-derived macrophages and monocytes express the activation-induced C-type lectin (AICL). Monocytes activated through their TLR by LPS, lipopeptide Pam₃CSK₄, or poly I:C increase the expression of AICL at their surface. AICL is the ligand of the NK cell activating receptor NKp80. The interaction AICL-NKp80 leads to the secretion of IFN-γ by NK cells (Welte et al., 2006). TLR-induced AICL expression may act to eliminate monocytes or macrophages exposed to or infected by microorganisms.

Both activating and inhibitory surface receptors participate in the regulation of NK cell activities. Thus, Qa-1, the ligand of the inhibitory receptor NKG2A is upregulated at the surface of poly I:C stimulated-macrophages. However, the NKG2A level does not change when NK cells are cocultured with these macrophages. The Qa-1-NKG2A interactions allow the macrophages to escape to lysis by NK cells (Zhou et al., 2012).

More recently, the proinflammatory cytokines IL-6, IL-12, IL-18, and TNF-α were observed in cultures of mouse spleen NK cells and peritoneal macrophages (Scott et al., 2004). The cytokine production between single cell-type cultures and cocultures showed no differences except a decrease of the IL-6 secretion by macrophages when they were in contact with NK cells. However, the inhibition of IL-12 secretion from activated macrophages has no influence on NK cell function during bacterial peritonitis (Scott et al., 2004). Thus, the potential activators IL-12 and IL-18 do not play a major role in NK cell activation by macrophages under these conditions. On the contrary, the secretion of IL-12 by CpG DNA or B. anthracis-infected macrophages enhances the release of IFN-γ by mouse NK cells but needs in addition a cell-to-cell contact (Chace et al., 1997; Klezovich-Benard et al., 2012). Moreover, IL-12 and IL-18 are involved in the crosstalk between Kupffer cells (liver resident macrophages) and NK cells after poly I:C treatment in mice (Hou et al., 2009).

The same capacity of IL-12 and IL-18 secreted by Kupffer cells to activate IFN-γ production by liver NK cells is found in human (Tu et al., 2008). The key role of IL-12 and IL-18 in NK cytotoxicity is confirmed in an experiment where PBMC-derived macrophages are infected with Salmonella (Lapaque et al., 2009). Furthermore, LPS, in presence of PBMC-derived macrophages, promotes the expression of IL-18 and consequently the synthesis of IFN-γ by NK cells (Bellora et al., 2012). Finally, S. aureus stimulated-monocytes from PBMC, produce IL-12 which participates to IFN-γ secretion by NK cells (Figure 1) (Haller et al., 2002).

On the other hand, the release of the proinflammatory cytokine IL-15 by mouse poly I:C-stimulated macrophages, also promotes the production of IFN-γ and the expression of CD69 by NK cells (Zhou et al., 2012). Furthermore, macrophages isolated from lungs infected by S. aureus represent a source of IL-15 capable to induce TNF-α production by NK cells (Small et al., 2008). Another team has recently shown that mice infected by L. monocytogenes have inflammatory monocytes and subsets of spleen macrophages which mediate granzyme B and IFN-γ secretion by NK cells in vivo (Soudja et al., 2012). After detection of different classes of microbial pathogens, activated monocytes and macrophages upregulate and trans-present IL-15 to NK cells.

Recently, a study described that IFN-γ–producing NKp46⁺ NK cells were increased in the intestine of patients with Crohn's disease. The production of IFN-γ by these cells is stimulated by the secretion of IL-23 and by cell-to-cell contact with the CD14⁺ lamina propria macrophages in response to stimulation with E. faecalis (Takayama et al., 2010). IL-23 may be involved as a key mediator in gut-specific inflammation.

Monocytes have also the capacity of a net inhibitory effect on NK cell activation through the release of IL-10 after an infection with M. tuberculosis and a depletion of these monocytes will increase the secretion of IFN-γ by NK cells (Schierloh et al., 2005).

The cytokines released by monocytes and macrophages to mobilize NK cells require the recognition of “danger” molecules by innate immune receptors. Then the triggering of pathways which involve NK-cell mediated IFN-γ release will promote the expression of antimicrobial functions of effector cells to protect from pathogens (Figure 1).

The outcome of the interaction between macrophages and NK cells is depending on their tissue localization (Zoller and Matzku, 1982; Raz, 2007). Indeed macrophages derived from PBMC have not the same phenotype than macrophages localized in liver. There is lower expression of HLA class I and greater expression of HLA class II molecules on Kupffer cells (Tu et al., 2008). The same phenomenon is observed for NK cells. Indeed NK cells from the liver are phenotypically different from blood NK cells due to an absence of CD16 expression (Tu et al., 2008). Mouse NK cells from lung and spleen are quite different and react differently in contact with spleen or lung macrophages. Spleen macrophages have the tendency to induce a stronger activation of NK cell cytotoxicity than lung macrophages (Michel et al., 2012). Furthermore, depending on the tissue environment, human macrophages are polarized toward either M1 or M2 subsets which display different functionality (Biswas and Mantovani, 2010). LPS stimulation induces a change of unpolarized (M0) and M2 macrophages toward an M1 phenotype and then leads to a strong activation of autologous NK cells (Bellora et al., 2010). Furthermore, human cytomegalovirus infected M1 macrophages secrete IL-6, IL-12, TNF-α, and IFN-γ whereas M2 macrophages produce only low levels of TNF-α, IFN-γ, and IL-10 (Romo et al., 2011) The emerging complexity of interactions between NK cells and macrophages makes the understanding of the pathways implicated in vivo in steady state or during infection an ongoing challenge.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.