Complement factor P: promoting the antibacterial activity of natural killer cells

Natural cytotoxicity receptors (NCRs) are expressed on natural killer (NK) cells and other subsets of innate lymphoid cells (ILC). Three NCRs, NKp30, NKp44 and NKp46, are expressed on human NK cells.¹ NKp46 is the only NCR shared by primates and rodents, suggesting an essential role for this receptor during evolution.² While cellular ligands for NKp30 and NKp44 have been identified, a well-characterized non-microbial ligand for NKp46 is still lacking. However, in a recent paper in Science Immunology, Narni-Mancinelli et al.³ reported the identification of complement factor P (CFP, also known as properdin) as a ligand of NKp46. As CFP-deficient individuals are highly vulnerable to Neisseria meningitides (N.m.) infection,⁴ the authors used mice to show that the CFP-dependent protection from N.m. infection is largely dependent on NKp46 and NKp46⁺ ILC. Furthermore, these authors showed that the stimulation of NK cells by CFP resulted in the transcription of a chemokine gene rather than the expected cytotoxic response of an NCR. These findings revealed an unexpected role for the complement system in the NKp46-mediated antibacterial activity of NK cells.

For over two decades, numerous efforts have been reported in the search for ligands of NKp46 on both pathogen-infected and uninfected human and mouse cells. Hemagglutinin (HA) from several viruses binds to and activates NKp46.^{2, 5} As viral HA mediates virus attachment to cells by binding to sialic acid, sialylated NKp46 on NK cells, in this case, serves as a ligand for HA. Other proposed pathogen-derived or pathogen-induced NKp46 ligands include the cytoskeleton protein vimentin expressed on M. tuberculosis-infected macrophages and proteins of the yeast Candida glabrata.^{2, 5} Despite these findings, it is clear that a germ line-encoded receptor such as NKp46 cannot bind a broad array of pathogen-specific antigens. Several studies have shown that heparin and heparan sulfate contribute to NKp46 binding to tumor cells, suggesting that the ligand specificity of NKp46 may be complex.⁶ The difficulty in isolating a ligand or cloning the gene encoding a ligand for NKp46 suggests that its elusive ligand may include a combination of several components.

To screen for natural, cellular ligands that could induce NKp46 activation, Narni-Mancinelli et al. generated a chimeric receptor by fusing the extracellular domain of NKp46 with the T-cell receptor ζ chain and expressed this chimera in reporter DO11.10 mouse T cells, which produce IL-2 upon activation. Strong IL-2 production was observed when the NKp46-ζ reporter cells were incubated with simian virus 40-transformed mouse embryonic B12 fibroblasts, suggesting that a ligand of NKp46 was expressed on the surface of these cells. The authors generated monoclonal antibodies (mAb) against B12 cells and screened for clones that could block IL-2 secretion. Clone 27A1.7 was thus identified as a blocking mAb. Mass spectrometry was subsequently used to identify proteins immunoprecipitated from B12 cell lysates by 27A1.7. Junction adhesion molecule 1 (JAM1) and CFP were identified as potential ligands of NKp46. Given the previously described heterogeneity of NKp46 ligands, these authors conducted rigorous testing of JAM1 and CFP binding to NKp46. Serum-purified human CFP, but not JAM1, bound to a NKp46-Fc fusion protein. Binding was further confirmed after staining NKp46⁺ cell lines and primary human NK cells with a CFP-Fc fusion protein. Notably, JAM1, which bound directly to mAb 27A1.7, did not bind NKp46, and conversely, CFP bound to NKp46 but did not bind to mAb 27A1.7. Moreover, to close the loop of their original screen using cells expressing chimeric NKp46-ζ, Narni-Mancinelli et al. showed that silencing CFP expression in B12 cells prevented the induction of IL-2 expression observed after the coincubation of B12 cells with reporter NKp46-ζ cells.

First discovered in 1954 and named properdin, CFP positively regulates the complement alternative activation pathway by binding to and stabilizing C3- and C5-convertase enzyme complexes and is sufficient for the assembly of convertases in the absence of antibodies.^{7, 8} In addition, CFP acts as a pattern-recognition molecule by binding to certain microbes, glycosaminoglycan (GAG) chains associated with surface proteoglycans and apoptotic cells.⁸ Evidence suggests that the stabilization of convertases is mediated by CFP present in plasma, whereas the pattern-recognition function reflects CFP released from neutrophils.⁸ CFP spontaneously forms dimers, trimers and tetramers, and CFP oligomerization increases its activity in promoting the complement alternative activation pathway.⁸ By separating different CFP oligomers using gel-filtration and comparing their affinity for NKp46-Fc by flow cytometry, Narni-Mancinelli et al. showed that CFP dimers and tetramers bind to NKp46.

As the X-linked CFP deficiency is characterized by high susceptibility to meningococcal infections,⁴ these authors used several systems to examine the resistance of N.m.-infected mice, and characterize how NKp46 may contribute to this resistance. The depletion of NK cells using mAb NK1.1 resulted in higher mortality than observed in control mice. Similar results were obtained using selective depletion of NKp46⁺ NK cells using conditional diphtheria toxin receptor expression in NK cells. The administration of exogenous CFP protects mice from N.m. infection.⁹ However, in Ncr1^GFP/GFP mice lacking NKp46, the efficiency of CFP in clearing N.m. was greatly reduced, thus establishing a key role for NKp46 in CFP-mediated antimicrobial activity.

Typical responses of NK cells to activation signals, such as those triggered by sensitive target cells or the antibody-mediated crosslinking of NKp46, include interferon-γ secretion and the fusion of secretory granules with the plasma membrane (also known as degranulation). However, the stimulation of NK cells with CFP coated onto plates elicited a different response (Figure 1). RNA sequencing revealed the downregulation of a group of genes, some of which are normally activated by NKp46, and the upregulation of a few genes, including Xcl1. The chemokine Xcl1 (also known as lymphotactin) has direct antimicrobial activity.¹⁰ Interestingly, Xcl1 binding to its receptor Xcr1 recruits dendritic cells capable of antigen cross-presentation for CD8 T-cell activation during bacterial and viral infection.^{11, 12} It is likely that stimulation of NK cells by plate-coated CFP provided an incomplete signal, and additional ligands are required for the full activation program. In any case, the fact that different ligands of NKp46 may elicit distinct responses is an important concept. It will be interesting to examine NKp46 signaling and the biological response induced by CFP bound to different microbes. In addition, one of the main challenges is to determine how NKp46-CFP interactions and the resulting signals protect from N.m.-induced meningitis.

Figure 1.

Narni-Mancinelli et al.³ showed that CFP is a self-derived ligand for natural cytotoxicity receptor NKp46. CFP-coated plates activated an alternative signaling pathway, which did not induce degranulation but rather the secretion of chemokine Xcl1. These data suggest that microbes opsonized with CFP stimulate NK cells to release Xcl1, which induces an antibacterial response (left image). These authors also showed that NKp46 was required for the control of Neisseria meningitidis infection in vivo. In a traditional scenario, NKp46 binding to target cells triggers a cytotoxic response through the phosphorylation of its associated ITAM-bearing signal adapter proteins CD3ζ and FcεRIγ (right image). A kinase of the Src family phosphorylates ITAMs, which recruit tyrosine kinases Syk and ZAP70. Reorganization of the cytoskeleton and Ca²⁺-dependent degranulation results in the polarized release of lytic granule content. Reflecting the lack of an identified non-pathogen-derived NKp46 ligand on infected cells or tumor cells, the activation of NK cells by NKp46 has primarily been studied using antibodies to either crosslink NKp46 (right image) or block NKp46-dependent target-cell killing. Selected genes either upregulated (Up) or downregulated (Down) after stimulation with CFP (left image) or anti-NKp46 antibody (right image) are listed. CFP, complement factor P; ITAM, immunoreceptor tyrosine-based activation motif; NK, natural killer.

A major question raised by the discovery of CFP as a ligand of NKp46 is the extent to which CFP participates in tumor cell recognition by NK cells. In this context, CFP may function as a member of a complex with other components to trigger classical NK-cell activation through NKp46, or as a partner in synergy with ligands for other NK cell co-activation receptors.¹³ A potential functional partner of CFP is JAM1, the protein bound by mAb 27A7.1 in a complex with CFP, as JAM1 associates with leukocyte function-associated antigen 1,¹⁴ an important adhesion and activation receptor expressed on all NK cells which promotes the formation of polarized immunological synapses for NK-cell cytotoxicity.¹⁵ In general, serum CFP does not bind to healthy primary cells; however, CFP does bind to certain cell lines, including tumor cell lines.⁸ An alternative source of CFP could be its gene activation in cancer cells and subsequent export to the cell surface for detection by NKp46. The mouse embryonic B12 fibroblast cells used by Narni-Mancinelli et al. as the source of a ligand for NKp46 is one such example. Heparan sulfate GAG on the cell surface, to which CFP binds,⁸ could serve to capture secreted CFP for recognition by NKp46⁺ cells. An analysis of CFP expression in healthy cells, tumor cells and transformed cell lines may provide additional information.

The work highlighted here has elucidated an unexpected role for CFP in activating NKp46 signaling, revealing previously unknown crosstalk between the complement system and the functional regulation of NK cells for antibacterial responses. A fascinating exploration of the role of CFP in stimulating NKp46-dependent NK-cell responses during bacterial infections and in cancer has just begun.