Some of the most important information regarding human NK cell differentiation has been provided by multiparametric immunophenotyping of hematopoietic progenitor culture systems. This approach played a key role in the identification of novel stages of NK lineage and in the creation of NK development models. In a recent overview, Freud and Caligiuri (23) have proposed a sequential model of human NK differentiation, distinguishing five functionally different NK developmental stages (from NK progenitors to mature NK cells) based on surface antigen expression (CD34, CD117, CD94, and CD16).
In the present study, we characterized the CD56bright NK cells generated after 30 days of in vitro culture with an essential cytokine combination (FL plus IL-15). Interestingly, as shown in vivo by Shilling et al., this subset was the first to appear after about one month in hematopoietic stem cell transplants (24). However, the CD56bright cells obtained in culture have shown several immaturity traits as revealed by both the lack of β2-integrin and, characteristic of stage 3 NK cells (23), of CD94. As a matter of fact, low percentages of cells expressing LFA-1 or CD94 antigens have already been described in NK cells developing in vitro (10, 25). In our experiments, we have pointed out that, although the percentage of the CD94−/CD117bright NK subset (stage 3) resembled those of CD56bright/LFA-1− NK cells, a small proportion of CD56bright/LFA-1− cells expressed low density C-type lectin MHC-I inhibitory receptors, a reminiscence of the CD117low/CD94low intermediate population described by Grzywacz et al. (14). This indicates that in vitro LFA-1 integrin is upregulated slightly later than the CD94-CD159a heterodimer. On the basis of this observation, we hypothesized that LFA-1, as the β2 integrin CD11b for mouse NK cells (26), might be a useful marker to distinguish stages of human CD56bright NK cell development in vivo. To this regard, both UCB and PB samples contain a subset of Lin−/CD34− cells expressing LFA-1 at a lower density (LFA-1low) than the CD56bright “mature” cells. Of note, Lin−/CD34−/LFA-1low cells expressed some mature NK cell markers (i.e. CD94, CD11c, CD2, CD56, NKG2D, NKp46, NKp30, CD244, and CD122) at relatively low density, suggesting that they represent an immature stage of differentiation. In particular, a fraction of Lin−/CD34−/LFA-1low cells coexpressed CD56 and CD94 at low density, possibly defining an intermediate LFA- 1low/CD56dim/CD94dim/CD16− stage of the NK cell lineage. In agreement with this hypothesis, it has been recently described that CD56dim/CD16− NK cells can first develop into CD56bright/CD16− NK cells and then into CD56bright/CD16dim NK cells (27). Interestingly, like the perforin-deficient stage 3 human lymph node NK cells (13, 23), Lin−/CD34−LFA-1low are small agranular cells expressing CD7, CD161, CD62L, CD244, CD25, CD117bright, and CD33 antigens. However, differently from stage 3 cells, most of them expressed NKp46 and NKG2D molecules, resembling our in vitro findings. On the other hand, the deficiency of other activatory receptors, such as NKp44 and NKp30, is in line with the inducible nature of the former and the low expression density on CD56bright NK cells of the latter. In substance, most of the functional NK cell markers were similarly expressed in in vivo (PB and UCB) immature NK cells and in in vitro generated ones. This suggests that at least part of Lin−/CD34−/LFA-1low cells could represent the in vivo counterpart of CD161+/NCR+/LFA- 1−/CD94− NK cells, generated in vitro from CD34+ hematopoietic progenitor. Within Lin−/CD34−/LFA-1low cells, about half of them were CD161bright/CD56−/CD16−, an immature phenotype already described in PB and UCB (3) that has been shown to secrete IL-5 and IL-13 (12, 28). Interestingly, IL-13 has been described as being basically produced by CD56bright cells (5, 29). Finally, all these observations suggest that CD56bright NK cells may derive from CD161+/NKG2Ddim/NKp46dim/CD56− cells passing through the CD56dim/CD94dim/CD16− NK cell stage. Summarizing the reported data, we suggest the opportunity to distinguish two distinct maturative steps within the well known stage 3 of CD56bright differentiation, as described in Figure 4.
During normal in vivo differentiation, developing NK cells acquire activating and inhibitory receptors and cytotoxic function in a fashion that prevents NK-mediated auto-aggression against normal cells. There are different hypotheses to explain the NK cell self-tolerance during differentiation (20, 21). One possibility is that the expression of functional inhibitory receptors would precede that of activatory ones. Among human MHC-I inhibitory receptors, CD94-CD159a molecules have already been indicated as an early mechanism for self-tolerance during differentiation (20). Indeed, CD94-CD159a heterodimer is expressed earlier than KIRs and its inhibitory function has been demonstrated in NK cells differentiating in vitro (30, 31). However, our and other reports (4, 13) clearly indicate that activatory molecules are expressed in higher percentages of cells than MHC-I inhibitory receptors on immature NK cells, suggesting that activatory receptors precede CD94-CD159a heterodimer expression. In this regard, the inhibitory function of CD244 on the cytotoxic activity of immature NK cells has been claimed to assure a fine control of fail-safe mechanism (4). Nevertheless, the phenotype (LFA-1low) and the scatter (small agranular) characteristics of immature stage 3 NK cells can lead to further explanations. For example, they could not be still functionally cytotoxic due to the deficiency of β2-integrins and cytotoxic granules, and a complete functionality of NK cells could be reached only after the expression of MHC-I inhibitory receptors.