Ly49 receptors: not always a class I act?

In this issue of Blood, Van Den Broeck and colleagues report a series of experiments that lead them to this startling conclusion: Not only can the plasminogen activator, urokinase, inhibit NK-cell function, but it does so via engagement and triggering of a member of the Ly49 family of MHC class I receptors, Ly49E.

Research over the last 15 years has revealed that the immunologic functions of natural killer (NK) cells are regulated by a series of receptors that deliver inhibitory signals following interaction with appropriate ligands. Prominent among these receptors are those belonging to the KIR family in humans and to the functionally equivalent Ly49 family in mice. These receptors are highly polymorphic, being encoded by multiple genes that exist in multiple allelic forms. Although the ligands for only a proportion of the known KIR and Ly49 receptors have been formally identified, in every case until now, the ligands have been found to be either major histocompatibility complex (MHC) class I molecules or class I–related molecules encoded by pathogens. The interaction of KIR and Ly49 receptors with class I molecules on neighboring cells generally delivers inhibitory signals that serve 2 purposes: to ensure that NK cells do not “accidentally” damage healthy cells and to allow NK cells to detect and destroy diseased cells that have down-regulated expression of their class I molecules (so called missing self recognition).

One member of the Ly49 family, Ly49E, however, has stood out as being different. Ly49E is expressed, in the absence of other Ly49 family members, by fetal NK cells^1,2  and subpopulations of γδ T cells, particularly at epithelial sites.³  By contrast, it is absent from resting adult NK cells but is induced on these cells following activation by IL2 and IL15.⁴  Furthermore, Ly49E is essentially nonpolymorphic, and all attempts to demonstrate interaction with class I molecules have so far failed. To identify the ligand for Ly49E, Van Den Broeck et al transduced a Ly49E construct into reporter cells and obtained clear signals when these were admixed with certain other cells, such as the myeloid line J774. Transfection of a J774 cDNA library into a nonstimulatory cell line generated a series of transfectants, all of which contained a cDNA integrant encoding the urokinase plasminogen activator [uPA]. Plates coated with purified uPA could stimulate reporter cells expressing Ly49E but not other related molecules, and could inhibit IFNγ production by NK-cell lines and granule exocytosis by Ly49E-expressing fetal NK cells. Importantly, in the latter 2 cases, the inhibition was partially reversed by anti-Ly49E antibody or Ly49E shRNA.

The authors concluded from these experiments that uPA is a powerful inhibitor of NK cells and that this inhibition is mediated through inhibitory signals triggered by Ly49E engagement with uPA. If true, this would represent the first demonstration of the binding of a Ly49 receptor to a non–class I ligand. Given the extremely close sequence homology of Ly49E to other Ly49 family members, particularly in the class I–binding region,⁵  this is most unexpected, especially as recent studies have indicated that the much more divergent Ly49 family members, Ly49B and Ly49Q, have class I ligands.^6,7  The authors' failure to detect any direct binding of uPA to Ly49E-expressing cells or of soluble Ly49E to uPA suggests that there may be other explanations, for example, that uPA binding to Ly49E requires a cofactor. Regardless of the exact explanation of the data, the results presented in this paper are highly provocative and will undoubtedly stimulate further research into the precise nature of the Ly49E ligand.