Proliferating NK cells in response to IL-15 do not upregulate surface B220 in vivo

Natural killer (NK) cells and dendritic cells (DC) are functionally diverse leukocytes. However, a population functionally astride both cell types has been described and termed interferon (IFN)-producing killer DC (IKDC). According to the original descriptions, these cells efficiently mediate both natural cytotoxicity and antigen presentation to CD4 T cells. Other authors have challenged the original observations with ontogeny and functional results, which concluded that these cells merely represent an activated status of NK cells. The CD45 isoform B220 is a surface marker used to distinguish both populations. The putative IKDC has been shown to be induced to proliferate and is activated by interleukin 15 (IL-15) as a recombinant protein or when the liver is gene transduced with an efficiently expressed IL-15 expression cassette. In this letter, we show that there are in vivo conditions of activation of spleen NK cells that do not result in surface expression of B220 on either IL-15 gene transfer to the liver, or when NK cells proliferate in a major histocompatibility complex-mismatched and lymphopenic recipient mouse. These in vivo data indicate a lack of interconversion on activation of B220− conventional NK cells and B220+ IKDC. A hot controversy has been raised on the identity and function of NK1.1+DX5+CD11c+B220+ cells. Such a sub-population has been claimed to perform both NK-like cytotoxicity and DC-like antigen presentation to CD4 T cells.1, 2 Because of their high production of IFNγ, these cells were termed IKDC. The main surface marker distinguishing such a population is B220, a high molecular weight (220 kDa) isoform of CD45 that was originally considered as a selective B-cell marker in mice.1, 2 Other authors contend that these cells represent activated NK cells, as classical NK cells (cNK) (DX5+CD3−NK1.1+B220−) acquire B220 on activation.3, 4, 5 Recently, we reported that productive in vivo transfer of the IL-15 gene into the liver by hydrodynamic injection results in the proliferation and activation of both cNK cells and IKDC.6 Furthermore, in the context of in vivo gene transfer of IL-15 into hepatocytes,6 we made other observations centered on both the function and ontogeny of IKDC, which are relevant to this controversy.1, 2, 3, 4, 5, 7 DX5+ cells from the spleen of B6.SJL-PtprcaPep3b/BoyJ mice (CD45.1 mice) were analyzed by fluorescence-activated cell sorting (FACS) analysis (to 98–99% purity) into DX5+CD3−NK1.1+B220− (cNK) and DX5+CD3−NK1.1+B220+ (IKDC) fractions. Both populations once labeled by CFSE (carboxyfluorescein diacetate succinimidyl ester) were adoptively transferred into congenic C57BL/6J (CD45.2) mice, which had been hydrodynamically injected with the IL-15-encoding (pVKL/IL15IRESNeo) or control (pIRESNeo) plasmid 8 h before adoptive transfer, as detailed in Arina et al.6 Immunostaining of the CD45.1/CD45.2 polymorphism permitted unequivocal gating of the adoptively transferred cells. As depicted in Figure 1a, a fraction of cNK cells proliferated several cycles for 3 days without acquisition of B220 surface expression on the vast majority of cells, whereas the anti-B220 antibody readily stained B cells from recipient mice in the same splenocyte preparation (not shown). In a parallel experiment (Figure 1a , middle dot plots), sorted IKDC showed a slightly higher proliferative response to IL-15 gene transfer and retained a bright B220 expression during the 3-day in vivo experiment. An overlay with colored dot plots from matched mice is given for comparison in Figure 1a (bottom dot plots, blue dots for conventional NK and red dots for IKDC). It could be argued that the in vivo proliferation time was not long enough. However, if the experiment had been extended to 5 days, more proliferation could have been substantiated among the cNK, without acquisition of a detectable B220 surface staining in the majority of proliferating cells. To provide sorted cNK cells with an even stronger proliferative stimulus, they were transferred into Rag-2−/−γc−/− BALB/c mice. In these conditions, a fast and strong proliferation of adoptive transfer of cNK cells is observed as a result of alloreactivity and lymphopenia of recipient mice. Even in this case, there was only a slight tendency to acquire B220 expression by proliferating NK cells. Similar results were found when cells were transferred into Rag-2γc−/− BALB/c mice receiving polyI:C (data not shown). Interleukin-15 has been shown to be a major promoter of activation and proliferation for both NK cells and the putative IKDC cell sub-population.6, 8, 9 The fact that cNK cells do not acquire B220 surface expression even when experiencing intense IL-15-dependent proliferation and activation in vivo supports the existence of differences between cNK and IKDC without interconversion at least under the described conditions of in vivo activation. According to our data, the B220+ cells that proliferate on IL-15 systemic delivery are not conventional B220− NK cells that acquire B220 expression but cells that were expressing B220 before exposure to exogenous IL-15. However, the core of the controversy remains as to whether IKDC present antigen from the debris of the cells that they terminate and thereby prime specific T cells. In this regard, two recent papers provide evidence showing that IKDC are capable of efficient presentation of tumor and infectious antigens to CD4+ and CD8+ T cells.10, 11 Moreover, IKDC show interesting and distinct migratory properties to tumor tissues and lymph nodes.12 The fact that cNK cells do not acquire B220 expression even when experiencing intense IL-15-dependent proliferation and activation further supports the existence of differences between NK and IKDC. The authors declare no conflict of interest.