what does it take to make a natural killer? - Nature

REVIEWS

WHAT DOES IT TAKE TO MAKE A NATURAL KILLER? Francesco Colucci*, Michael A. Caligiuri‡ and James P. Di Santo* We know how B and T cells develop, what they ‘see’ and the receptors they ‘see with’. By contrast, and despite an unprecedented increase in the number of receptors and ligands known to regulate the activity of natural killer (NK) cells, we still have many questions regarding how these cells develop. Nevertheless, we are beginning to understand the transcriptional programmes of NK-cell maturation and the role of the effector functions of NK cells in the regulation of immune responses. An improved knowledge of NK-cell development in mice and humans might be useful to harness the power of these natural killers in the clinic to fight autoimmune diseases, infection and cancer. HAEMATOPOIETIC STEM CELLS

(HSCs). These are defined functionally as the one cell that is capable of self-renewing and generating all of the cell types present in the blood. Common myeloid progenitors, common lymphoid progenitors, early lymphoid progenitors and natural-killer-cell precursors are committed progenitors that are derived from HSCs.

*Cytokines and Lymphoid Development Unit, Department of Immunology, Pasteur Institute, 25 Rue du Dr Roux, 75724 Paris, France. ‡Department of Internal Medicine, Division of Haematology and Oncology, The James Cancer Hospital and Comprehensive Cancer Centre, The Ohio State University, Columbus, Ohio 43210, USA. Correspondence to J.P.D. e-mail: [email protected] doi:10.1038/nri1088

Innate immunity is evolutionarily older than antigenspecific immunity and is crucial for its effector function1. After several decades of fascination with the generation of diversity in B and T cells involved in adaptive immunity, immunologists have a renewed interest in innate immunity, and are now starting to understand some of the molecular mechanisms that regulate its functions2. Natural killer (NK) cells are one of the cellular mediators of innate defence. They are lymphoid cells that, without the need for immunization or pre-activation, can recognize and kill aberrant cells and rapidly produce soluble factors — chemokines and cytokines — that have antimicrobial effects or prime other cells of the immune system. NK cells have a heterogeneous arsenal of surface receptors that allow them to respond to microbial products, cytokines, stress signals and inducible molecules that are expressed after target-cell transformation3. NK cells are therefore crucial for defence against infectious diseases and cancer. We have also begun to understand the role of cytokines in immune regulation, to elucidate the molecular mechanisms of cytokine signalling and to define the transcriptional programmes of lymphocyte maturation. This review compares and contrasts human and mouse NK-cell development in an attempt to address some unresolved questions (BOX 1) with respect to the physiological roles of NK cells.

NATURE REVIEWS | IMMUNOLOGY

The where and when of making a killer

Lymphocytes develop in anatomically distinct sites in the fetus and the adult; however, the precise sites for NK-cell development are poorly defined. NK cells are derived from HAEMATOPOIETIC STEM CELLS (HSCs). Haematopoietic precursors have been identified in many sites during different phases of intra-uterine or adult life, including the yolk sac, aorta–gonad–mesonephros region and liver in the embryo and fetus, and the bone marrow, thymus, spleen, omentum and liver in adults4. Many of these tissues contain mature NK cells (FIG. 1). Where in the body and when in ontogeny are NK cells produced? In adults, the present consensus states that the bone marrow is the main site of NK-cell generation. The bone-marrrow microenvironment provides a rich source of cytokines and growth factors that can support NK-cell development in vitro and contains stromal cells that are required for full maturation of NK cells. Moreover, NK-CELL PRECURSORS (NKPs) are found in the bone marrow. Indeed, bone marrow ‘ablation’ by oestradiol5 or radio-isotope treatment6 in mice affects NK-cell development more markedly than it affects other haematopoietic lineages, which can develop in extramedullary sites, such as the spleen. Nevertheless, the cytokines, extracellular matrix and other features that make the bone marrow particularly suited for NK-cell development remain to be defined.

VOLUME 3 | MAY 2003 | 4 1 3

© 2003 Nature Publishing Group

REVIEWS

NK-CELL PRECURSORS

(NKPs). Cells that can differentiate into natural killer (NK) cells but no other haematopoietic lineage. The expression of the interleukin-2 receptor β-chain by early lymphoid precursors seems to indicate commitment to the NK-cell lineage. DI GEORGE SYNDROME

A clinical syndrome that is characterized by congenital aplasia or hypoplasia of the thymus and the parathyroid glands, associated with facial and cardiovascular malformations. In 85% of cases, it is caused by a de novo deletion of chromosome 22q11.2 DEVELOPMENTAL INTERMEDIATES

Phenotypically defined cell types that are known to be part of a developmental pathway, and that, on the basis of expression of cell-surface markers or geneexpression profiles, can be placed precisely in that pathway. COMMITMENT OF HAEMATOPOIETIC STEM CELLS

Lineage commitment is a progressive process that results in loss of the ability to differentiate into multiple lineages. This process is associated with a change in the gene-expression profiles and often correlates with the acquisition of expression of specific cell-surface markers, which are useful for the identification of particular stages of development.

Neither the thymus nor the spleen seems to be essential for the generation of NK cells. Congenitally athymic nude mice7, patients with DI GEORGE SYNDROME8, or thymectomized mice9 and humans10 seem to have normal numbers of functional NK cells. Similarly, splenectomy in mice11 or humans12 does not markedly affect NK-cell development. NK cells are also present in the liver, which is an active site for the production of NKPs in fetal life13,14, but whether this is a site for NK-cell development in adults is not known. It remains possible that the thymus, liver and/or spleen might have a role in the NK-cell diversification that results in NK-cell heterogeneity. Extensive phenotypic and functional analysis of NK cells that are generated in the absence of the thymus or spleen might shed light on this issue. The relationship between fetal and adult NK-cell development is also poorly understood. Both the fetal thymus and fetal liver contain haematopoietic precursors that have the potential to become NK cells13–16, and NK cells can develop in the fetal thymic microenvironment13,16−19. Fetal tissue could therefore provide a source of NK cells for use in the clinic. However, the biological importance of fetal NK-cell development is not clear. Do fetal thymus- and fetal liver-derived NK cells have a particular function? Does the thymus continue to make and export NK cells in adult life? One possibility is that thymic NK cells are simply what is left over from nature’s experiments, when primordial lymphoid cells (NK-cell-like) and their precursors made their entry into the primordial thymus. Growing killers in vitro

The identification of the developmental stages in B- and T-cell lymphopoiesis has advanced considerably, whereas our knowledge of NK-cell DEVELOPMENTAL INTERMEDIATES remains in its infancy. Nevertheless, through advances made in the identification of haematopoietic precursors and in defining appropriate culture conditions (FIG. 2), human and mouse NK cells can be reproducibly generated in vitro from fetal thymus, fetal liver, cord blood and bone marrow HSCs13−17,20−24. In early studies, NK-cell development from purified HSCs was shown to be stromal-cell dependent21, later reports identified the cytokines that could allow stromal cellindependent generation of NK cells from HSCs14,22−24. The initial two-step model24 for NK-cell development (FIG. 2) provided a useful starting point for guiding our thinking about NK-cell generation. In this review, we make use of this model and discuss three stages of

Box 1 | Outstanding questions in NK-cell development • What are the signals that dictate natural killer (NK)-cell commitment, and how do transcription factors and cytokines regulate NK-cell development and differentiation? • How do NK cells acquire self-tolerance together with receptor expression and shaping of the NK-cell repertoire? • What is the sequence of acquisition of NK-cell effector functions? • Do the various lymphoid organs influence the differentiation of NK cells, and if so, how? • What do we know about the life-span of NK cells, their homeostasis and recirculation? • Is there a division of labour between the different NK-cell subsets?

414

| MAY 2003 | VOLUME 3

NK-cell development: an initial phase, which involves COMMITMENT OF HAEMATOPOIETIC STEM CELLS to the lymphoid lineage and then to the NK-cell lineage. This stage generates NKPs that have the potential to become NK cells but no other haematopoietic lineage; a secondary phase involving the maturation of NKPs towards cells that have phenotypic and functional characteristics of peripheral NK cells. Immature NK cells are generated in this process. This stage includes the development of the NK-CELL REPERTOIRE, acquisition of SELF-TOLERANCE and the generation of EFFECTOR FUNCTIONS; a final phase that involves the export or ‘seeding’ of mature NK cells to the peripheral organs. This process generates the steadystate NK-cell pool, which can then be modified through homeostatic proliferation and recirculation under diverse pathological conditions. For example, NK-cell populations expand during virus infection25, and are recruited to the sites of infection26 or tumour challenge27. This model of NK-cell development has acquired considerable experimental support and further definition on the basis of analysis of NK cells from various mutant mouse strains and rare human disorders (TABLE 1). Stage 1: commitment to becoming a killer

From HSCs to NK-cell precursors. HSCs give rise to precursors for the different haematopoietic lineages. The first step of HSC commitment generates common myeloid progenitors (CMPS)28 and common lymphoid progenitors (CLPS)29. CLPs and their immediate precursors, the earliest lymphoid progenitors (ELPS)30, further differentiate to give rise to B- and T-cell-restricted precursors. Although existence of a committed NKP was assumed by analogy, definitive evidence for its existence was, until recently, lacking. Some clues were provided from studies of the generation of NK cells from HSCs in vitro (FIG. 2). Early studies showed that NK-cell generation from human CD34+ HSCs required stromal-cell contact, whereas more-differentiated CD7+CD34+ cells could generate NK cells when cultured in interleukin-2 (IL-2) alone21. Similarly, fetal liver CD34+CD38+ cells (but not more-primitive CD34+CD38– HSCs) could generate NK cells in vitro using IL-15 (REF. 14). These studies indicated that once generated, NKPs could be driven to become NK cells under the influence of IL-2 or IL-15 alone. Later, it was shown that the stromal cell requirements could be replaced by early-acting cytokines, such as stem cell factor (SCF); also known as c-KIT ligand, fetal liver kinase 2 ligand (FLK2), also known as FMS-like tyrosine kinase 3 ligand (FLT3L) and IL-7 (REFS 23,24). By culturing HSCs in these cytokines, two groups identified a subpopulation of IL-2 receptor β-chain (IL-2Rβ)-expressing cells that presumably were NKPs23,24. Subsequently, NKPs were identified in mouse fetal thymus17 and adult bone marrow18: the cells were lineagenegative (CD3–CD19–Gr1–Ter119–), expressed IL-2Rβ and lacked the expression of many cell-surface markers of mature NK cells (such as NK1.1, DX5 and Ly49 receptors). These cells could give rise to NK cells at a high frequency, but lacked the potential to become other haematopoietic lineages, including B, T, myeloid and

www.nature.com/reviews/immunol


REVIEWS

Fetal life

Birth

Adult life

Thymus

IL-2Rβ

NK

NKP Thymus

T–NK-precursor

Liver

HSC NK

AGM Liver HSC NK-CELL REPERTOIRE

Bone marrow

The collection of cell-surface receptors expressed by a population of NK cells. Individual NK cells can express combinations of activating and inhibitory NK-cell receptors; the sum total of different NK-cell clones in the organism defines the ‘repertoire’. NK-cell receptors are expressed in a sequential, but stochastic manner during maturation in the bone marrow.

NKP Spleen NK

SELF-TOLERANCE

With reference to NK-cell differentiation, the capacity of NK cells to recognize self-MHC molecules (typically classical and non-classical MHC class I gene products) through inhibitory receptors with the resultant downregulation of NK-cell effector functions. EFFECTOR FUNCTIONS

NK cells were discovered for their ability to spontaneously kill tumour cell lines in vitro. Besides natural cytotoxicity, NK cells can mediate other forms of killing, such as antibody-dependent cellular cytotoxicity. NK cells are also potent producers of interferon-γ, tumour-necrosis factor and granulocyte− macrophage colony-stimulating factor, which prime the adaptive immune system, stimulate myeloid growth and mediate anti-virus and cytotoxic effects. NK cells can also produce chemokines, which attract other cells of the immune system. CMPS, CLPS AND ELPS

A group of haematopoietic precursors that are committed to either the myeloid (common myeloid progenitor), or the lymphoid (common lymphoid progenitor or early lymphoid progenitor) lineage. CMPs give rise to monocytes, granulocytes, erytrocytes, megakaryocytes and mast cells, whereas CLPs/ELPs give rise to B, T and NK cells.

NK NKG2 CD94

KIR (human)

Ly49 (mouse)

Figure 1 | The where and when of NK-cell development. In the embryo, haematopoietic stem cells (HSCs) are generated in the aorta−gonad−mesonephros (AGM) region, which gives rise to all the blood-cell lineages. The intermediate steps from HSCs to natural killer (NK) cells are not completely understood, but precursors that are capable of generating NK cells have been identified in fetal thymus, blood, spleen and liver, such as bipotent T−NK progenitors. At birth, the site of haematopoiesis shifts from the fetal liver to the bone marrow. Here, NK-cell precursors (NKPs), but not bipotent T−NK progenitors, have been identified. NK cells that are found in the adult liver and thymus might be derived from bone marrow, but they might also be a remnant from fetal life. Splenic NK cells at birth do not yet express MHC inhibitory receptors — killer-cell immunoglobulin-like receptors (KIRs) in humans and Ly49 receptors in mice — but express CD94−NKG2 receptors that bind the non-classical MHC class 1 molecules HLA-E (human) and Qa1b (mouse). NKPs and mature NK cells that are present in primary lymphoid organs and in the periphery express a functional receptor for interleukin-15 (IL-15), here indicated by the IL-2Rβ chain.

erythroid cells18. These mouse NKPs were non-lytic and did not produce marked amounts of interferon-γ (IFN-γ); their further characterization indicated that they expressed receptors for growth factors (c-KIT and FLT3) and cytokines (IL-7Rα and IL-15Rα), as well as transcription factors that are known to be crucial for lymphoid development (PU.1, GATA3, ID2 and ETS1)18. Human NKPs have not been fully characterized ex vivo; however, the first IL-2Rβ-expressing cells that emerge following culture of human CD34+ HSCs with c-KIT ligand or FLT3L have been shown to express CD38 and transcripts for IL-15Rα, and are negative for CD7, CD16, CD56, NKG2A and several NK-cell receptors32. Collectively, these observations indicate that acquisition of expression of IL-2Rβ by NKPs denotes NK-cell commitment to the haematopoietic lineage (FIG. 3a). Cytokines that direct the transition of HSCs to NK-cell precursors. The receptor tyrosine kinases c-KIT and FLT3 are expressed by HSCs, ELPs, CLPs and NKPs, but are either not expressed by mature NK cells (FLT3) or


only expressed by a minor subset (c-KIT)32−34. The overlapping expression patterns of c-KIT and FLT3 receptors and the similar biological effects of their ligands indicate functional redundancy during early haematopoiesis33 — a notion that is supported by experimental evidence in mice. In the absence of either receptor, there is a reduction in the number of haematopoietic precursors, including CLPs35,36, which becomes more severe when both c-KIT and FLT3 are absent37. By contrast, deficiency in either c-KIT ligand or FLT3L signalling has only minor effects on the number of peripheral NK cells34,38. So, c-KIT ligand or FLT3L might have an important role in the generation of NKPs from HSCs through CLPs. Previous studies have indicated that c-KIT ligand and FLT3L might directly induce the expression of IL-2Rβ by HSCs or CLPs, thereby turning them into IL-2- or IL-15-responsive NKPs that are committed to the NKcell lineage23,24. The reduced number of NKPs in mice that are deficient in either c-KIT or FLT3 (REF. 34, and C.A. Vosshenrich and J.P.D., unpublished observations) rules out the possibility that either of these cytokines are

VOLUME 3 | MAY 2003 | 4 1 5


REVIEWS

Fetal tissues

Adult bone marrow c-KIT ligand, FLT3L and IL-7

IL-2Rβ Cord blood

Non-lytic committed precursor

NKP

Lineage negative c-KIT+ SCA1+ HSC

Thymus IL-7R+ IL-15 or IL-2 AGM

Liver

CLP

NKR-P1 Pseudomature lytic NK CD94

Stromal cells (cell lines, bone marrow, spleen)

IL-2Rβ+ NKP

IL-15 or IL-2 IL-21?

Ly49 (mouse) NK

Mature NK

KIR (human)

Figure 2 | NK-cell development in vitro. Haematopoietic precursors from embryonic tissues (yolk sac, aorta−gonad−mesonephros (AGM), liver, thymus, cord blood, peripheral blood and spleen) or adult bone marrow can generate natural killer (NK) cells in vitro. A two-step culture system has been described. In the first step, early activating cytokines increase the number of haematopoietic precursors and allow the emergence of commited NK-cell precursors (NKPs), which express the interleukin-2 receptor β-chain (IL-2Rβ). In the second step, IL-15 (or high doses of IL-2) can expand and further differentiate NKP populations to acquire lytic capacity. These ‘pseudomature’ NK cells might not have a corresponding correlate in vivo. Final NK-cell maturation, which is marked by the expression of killer-cell immunoglobulin-like receptors (KIRs) in humans and Ly49 receptors in mice, requires contact with stromal cells. CLP, common lymphoid progenitor; FLT3L, FMS-like tyrosine kinase 3 ligand; HSC, haematopoietic stem cell; NKR-P1, NK-cell receptor protein 1.

HOMEOSTASIS

A self-regulating process by which biological systems maintain stability, while adjusting to conditions that are optimal for survival. When applied to lymphocyte subsets, this refers to the mechanisms that regulate cell numbers in the peripheral pool.

416

essential for NK-cell commitment. c-KIT mutations in humans have been identified, for example in patients with Piebald’s syndrome39, but bone marrow and peripheral NK cells from these patients have not been analysed. If c-KIT ligand and FLT3L have redundant functions in the generation of NKPs, analysis of NK cells in mice that are deficient for both c-KIT and FLT3 should be informative. Finally, c-KIT ligand and FLT3L might promote CLP survival and proliferation, whereas other signals might commit them to the NK-cell lineage. The reduction of NKPs in both c-KIT- and FLT3-deficient mice is also consistent with this possibility. But, what are these other signals? Could they be provided by other cytokines? Because HSCs, ELPs, CLPs and NKPs express common γ-chain (γc)-dependent cytokine receptors18,29,40, IL-2, -4, -7, -9, -15 or -21 could be involved in the generation of NKPs. However, γc-deficient mice have normal numbers of NKPs (C.A.Vosshenrich and J.P.D., unpublished observations), indicating that all γc-cytokines are redundant for NKP generation. So, commitment to the NK-cell lineage might be driven by factors other than early-acting cytokines or γc-cytokines, although these soluble factors might influence subsequent NK-cell progeny.


Transcription factors and NKP generation. Transcription factors regulate numerous developmental processes and are crucial intrinsic regulators of lineage determination in the haematopoietic system41. A subset of transcription factors that either affect development of all lymphocytes or selectively block the development of individual lymphoid lineages have been identified. A null mutation in the gene encoding Ikaros causes marked effects on mouse lymphopoiesis, blocking the development of B cells, NK cells, dendritic cells and fetal T cells, and allowing only a limited postnatal wave of T-cell development42. This gene can be alternatively spliced to generate multiple isoforms, which might have unique roles in haematopoiesis: overexpression of certain Ikaros isoforms in CD34+ HSCs can markedly affect lymphopoiesis (possibly by acting as dominant-negative regulators)43. As bone-marrow haematopoietic progenitors of Ikaros-deficient mice express low levels of FLT3 and c-KIT receptors44, Ikaros might control ELP/CLP HOMEOSTASIS and thereby the development of all lymphoid lineages, through the maintenance of cytokine and growth-factor receptors. Members of the ETS family of transcription factors have been implicated in various aspects of haematopoietic development. The increased apoptosis that has been observed in B- and T-cell populations from ETS1-deficient mice45 might also cause the reduction in NK-cell numbers seen in these mice46. The ETS-family member PU.1 is crucial for the specification of myeloid and lymphoid lineages47. PU.1-deficient fetal liver cells can give rise to normal numbers of HSCs, but generate reduced numbers of NKPs and NK cells48. As the genes that encode c-KIT and IL-7Rα are molecular targets of PU.1 (REF. 49), a deficiency in ELP and CLP maintenance might help to explain the cell-intrinsic defect in the generation of B, T and NK cells in the absence of PU.1. What are the molecular events that indicate the divergence of committed B-, T- and NK-cell progenitors from ELPs/CLPs? Master ‘switches’ have been identified that regulate B- and T-cell development, such as NOTCH1 and PAX5, respectively. These signals seem to function by promoting a particular commitment pathway without regard for other options: NOTCH1 activates a T-cell programme, whereas PAX5 targets the B-cell lineage transcriptome41. According to this model, it is not surprising that mutations in NOTCH1 or PAX5 have little effect on NK-cell development50,51. The basic helix-loop-helix (bHLH) family of DNAbinding proteins are crucial molecular switches in the specification of lymphoid lineages. E12, E47 and HEB are essential for B-cell development and are implicated at many stages of T-cell development52. The transcriptional activity of the bHLH proteins is negatively regulated through hetero-dimerization with a family of inhibitors of DNA binding (ID) proteins. Are ID proteins the master ‘switches’ involved in NK-cell development? Overexpression of ID3 in human CD34+ HSCs in vitro inhibits T-cell development in fetal thymic organ cultures and enhances NK-cell development53. Constitutive expression of both ID2 and ID3 in CD34+ HSCs markedly favours the development of NK cells, whereas the generation of



REVIEWS T cells, B cells and lymphoid-associated dendritic cells is completely inhibited54. Similarly, ID2 deficiency results in a selective block in NK-cell development55, which affects the generation of NKPs56. These data are consistent with a model in which ID proteins promote NK-cell development from ELPs/CLPs through NKPs at the expense of committed T- and B-cell precursors19. The transcription factors that regulate the development of NK cells only partially overlap with those that control B- and T-cell development, indicating that the transcriptional programme for the NK-cell lineage might be established early in the progression from HSCs to NKPs. A unique transcriptional identity of NK cells is perhaps not surprising given the fact that the signals for NK-cell development markedly diverge from those that promote B- and T-cell development3. The identification of NKPs provided an important piece in the puzzle of NK-cell development. Further characterization of NKPs will undoubtedly provide insights into the mechanisms that control their genesis and maintenance. The ability to modify steady-state levels of NKPs could provide a means to enhance immune responses selectively under certain clinical situations.

Stage 2: making a mature killer

Immature NK cells. The molecular and cellular map of the maturation of NKPs to NK cells is largely incomplete. What is known is that during this process, NK cells establish a characteristic cell-surface phenotype and the capacity to elicit effector functions. Most attempts to characterize intermediates in NK-cell development have depended on markers of mature NK cells, including NK1.1, DX5 and Ly49 in mice, and CD161, CD56, CD16 and killer-cell immunoglobulin-like receptors (KIRs) in humans. Careful phenotypic and functional analyses of rare immature NK cells isolated from fetal liver, cord blood and bone marrow, or those that develop in vitro from HSCs14,18,20,21,31,57, support a linear model of NK-cell maturation (FIG. 3a). The markers that are expressed first by immature NK cells belong to the C-type lectin NK-cell receptor protein 1 (NKR-P1) family — detected as NK1.1 in mice and as CD161 in humans18,20,57. Mouse immature NK cells are DX5 negative and their human counterparts are CD56 negative. In addition, immature NK cells in both species fail to express the MHC-specific receptors Ly49 and the KIR family, in mice and humans, respectively. These

Table 1 | Mutations and their effects on human and mouse NK cells Mutated protein

Stage 1: bone marrow

Stage 2: bone marrow

Stage 3: spleen

References

NKP

iNK cells NK cells

NK cells Cytokines Cytotoxicity

IL-2, IL-2Rα

Normal*

Normal*

N.D.

Normal

N.D.

↓↓ (normal)*

68,69

IL-2Rβ

Normal*

↓↓↓*

↓↓↓*

↓↓↓

N.D.

↓↓↓

73,76

IL-15, IL-15Rα

Normal*

↓↓↓*

↓↓↓

↓↓↓

N.D.

↓↓↓

71,72

IL-7, IL-7Rα

Normal*

Normal

Normal*

Normal

N.D.

Normal

γc

Normal*

↓↓↓*

↓↓↓

↓↓↓

↓↓↓

↓↓↓

Cytokines and receptors

67 74,75

c-KIT ligand, c-KIT

↓*

↓*

↓

↓

N.D.

↓

FLT3 ligand, FLT3

N.D.

N.D.

↓

↓↓

N.D.

↓↓

38

IL-21R

N.D.

N.D.

N.D.

Normal

N.D.

Normal

90

LTα, LTβR

N.D.

N.D.

N.D.

↓↓

N.D.

↓↓↓

62,63

Ikaros

N.D.

N.D.

N.D.

↓↓↓

N.D.

↓↓↓

42

ETS1 (I)

N.D.

N.D.

↓↓↓

↓↓

↓↓

↓↓↓

46

PU.1 (I)

↓↓

N.D.

↓↓

↓↓↓

N.D.

Normal

48

ID2 (I)

↓↓

N.D.

N.D.

↓↓↓

N.D.

↓↓

MEF (I)

N.D.

N.D.

N.D.

↓↓

↓↓

↓↓↓

96

MITF (I)

N.D.

N.D.

N.D.

Normal

N.D.

↓↓

95

CEBPγ (I)

N.D.

N.D.

N.D.

Normal

↓↓↓

↓↓

IRF1 (E)

N.D.

N.D.

Normal

↓↓↓

N.D.

↓↓↓

IRF2 (I)

N.D.

N.D.

Normal

↓↓

N.D.

↓↓

94

NEMO (I)

N.D.

N.D.

N.D.

Normal

N.D.

↓↓

91

34

Transcription factors‡

55,56

93 64,65

Deficiencies in the absolute level (or percentage) of NK cells, cytokine production or cytotoxicity are given as: ↓, reduced compared with controls by a factor of less than or equal to two; ↓↓, reduced by a factor of 3–10; ↓↓↓, reduced by a factor of greater than ten or not detectable. *C.A. Vosshenrich and J.P.D, unpublished observations. ‡Transcription factors can act intrinsically (I; cell autonomous) or extrinsically (E; modifying the environment). Most transcription factors function in an intrinsic manner in developing NK cells, except for IRF1, which seems to influence the production of IL-15 by stromal cells. CEBP, CCAAT/enhancer binding protein; FLT3, FMS-like tyrosine kinase 3; γc, common cytokine-receptor γ-chain; ID2, inhibitor of DNA binding 2; IL, interleukin; iNK, immature natural killer; IRF, interferon regulatory factor; LT, lymphotoxin; MEF, myeloid elf-like factor; MITF, microphthalmia-associated transcription factor; N.D., not determined; NEMO, NF-κB essential modulator; NK, natural killer; NKP, NK-cell precursor; R, receptor.


VOLUME 3 | MAY 2003 | 4 1 7


REVIEWS a Developmental intermediates

HSC

CLP

NKP

Pseudomature lytic NK

Immature NK

Mature NK

NKR-P1

IL-2Rβ

DX5 Mouse

CD94 CD2

Ly49

c-KIT Cytokine receptors

FLT3 IL-7R IL-2Rβ NKR-P1

IL-2Rβ

CD56 Human CD94 CD2 Stages

b

KIR

Maturation

Commitment

Division of labour

Recirculation and homeostasis Thymus

Cytokine producers CD56hi IFN-γ TNF

CD25

Killers

c-KIT

CD56low CD94

KIR

GM-CSF

KIR

CD16

CCR7 Lungs Peripheral pool

Liver Spleen

Bone marrow

Figure 3 | A working model for NK-cell development. Natural killer (NK)-cell differentiation occurs as precursors interact with cytokines and stromal cells in the bone marrow. Cell-surface molecules are sequentially expressed by maturing NK cells and can be used as markers of developmental intermediates in mice and humans. Although some markers differ between the two species, many are shared and can be used to delineate common stages of NK-cell differentiation. a | Stage 1 (commitment) is marked by the acquisition of expression of interleukin-2 receptor-β (IL-2Rβ). Stage 2 (maturation) can be subdivided into further steps. Acquisition of expression of NK-cell receptor protein 1 (NKR-P1) molecules (NK1.1 in mice and CD161 in humans) and CD2 identify immature NK cells that are not lytic. DX5 in mice and CD56 in humans are subsequently acquired together with cytolytic potential. Expression of the CD94−NKG2 complex is acquired later, but before the final maturation, which results in the expression of MHC-specific receptors (Ly49 in mice and killer-cell immunoglobulin-like receptors (KIRs) in humans). For clarity, other markers, such as CD38, CD7 and 2B4, are not shown. Cytokine profiles seem to be developmentally regulated in human NK cells89, but, evidence in mice is lacking. b | Mature NK cells are exported to the periphery and are found in the blood, all of the lymphoid organs and some parenchymal tissues (lungs and liver). Peripheral NK cells might recirculate between the different organs and tissues. Evidence for such recirculation is seen in pathological conditions, such as cancer and infection; however mechanisms that regulate NK-cell recirculation under normal conditions are not known. In humans, two distinct subsets of NK cells have been described that are specialized in cytokine production or killing. This ‘division of labour’ has not been clearly shown in mice. CCR7, CC-chemokine receptor 7; CLP, common lymphoid progenitor; FLT3, FMS-like tyrosine kinase 3; GM-CSF, granulocyte−macrophage colony-stimulating factor; HSC, haematopoietic stem cell; IFN-γ, interferon-γ; NKP, NK-cell precursor; TNF, tumour-necrosis factor.

418




REVIEWS immature NK cells fail to lyse perforin-sensitive targets and do not produce IFN-γ18,20,57. Mouse or human immature NK cells can be induced to express other NKcell markers, including CD56 or DX5, and Ly49 or KIRs following in vitro culture with stromal cells18,58–60; these interactions presumably mimic some of the processes that select the repertoire of Ly49/KIR molecules in vivo. Human or mouse NK cells that are generated in vitro with stromal cells and IL-15 are cytolytic and can produce cytokines18,20,31,32,57−60. By contrast, prolonged culture of human CD34+ HSCs or mouse NKPs with IL-15 in the absence of stromal cells generates ‘pseudomature’ lytic NK cells — these cells express several markers of mature NK cells (NK1.1 and DX5 in mice, CD56 in humans and the CD94−NKG2 receptor complex in both species), but not Ly49 receptors or KIRs18,20,23,24,31,32. Whether ‘pseudomature’ NK cells exist in vivo is unknown. These observations indicate that the mechanisms involved in selecting Ly49-receptor and KIR repertoires are not required to produce functional NK cells. The heterogeneity of bone-marrow immature NK cells is considerable. In mice, analysis of integrin expression (CD11b and CD49d) by developing bone-marrow NK1.1+ cells indicated that numerous subsets were present18,61. Expression of some growth-factor receptors (FLT3 and IL-7Rα) decreases as a mouse cell proceeds from NKP to immature NK cell to mature NK cell, whereas expression of IL-2Rβ, CD2 and 2B4 (CD244) increases. Several subsets of human immature NK cells have been identified ex vivo and using in vitro culture systems; the most immature NK cells are CD56−CD161+ and can give rise to CD56+CD161+ cells before acquiring the expression of CD94, CD16, KIRs and natural cytotoxicity receptor (NCR)20,31,32. So in both mice and humans, NK-cell maturation probably proceeds in a dynamic and continuous manner (FIG. 3a), which alters the responsiveness of the developing NK cells as they acquire lytic and cytokine-production capabilities. The identification of immature NK cells provides an essential starting point for addressing several outstanding questions that concern intermediates in NK-cell differentiation, including the nature of the signals that promote and regulate the development of the Ly49/KIR-receptor repertoire and NK-cell effector functions. Cell−cell interactions and NK-cell generation. Interactions between developing NK cells and neighbouring stromal cells are important for proper NK-cell maturation. An intrinsic defect in NK-cell development has been observed in mice that are deficient in lymphotoxin-α (LTα) or the LTβ receptor (LTβR). A model has emerged in which a membrane-bound form of lymphotoxin, LTα1β2, presumably expressed by NKPs, signals through LTβR that is expressed by stromal cells to enhance the production of IL-15 (REF. 62). As exogenous IL-15 promotes in vitro NK-cell development from bone-marrow precursors derived from LTα-deficient mice, it is probable that NKPs develop in this setting, although they fail to receive an IL-15 signal from the stromal microenvironment63. No data on this subject


are available in humans, although these studies indicate possible adverse effects of using LTβR antagonists to treat disease. Similarly, interferon regulatory factor 1 (IRF1)-deficient mice have severely reduced numbers of NK cells owing to a defect in the production of IL-15 by the bone-marrow microenvironment64,65. Collectively, these observations highlight the bidirectionality of signals between developing NKPs and stromal cells in the bone-marrow microenvironment. As indicated earlier, stromal-cell interactions provide signals to developing NK cells that lead to the expression of Ly49 receptors or KIRs. Although the nature of the stromal-cell-derived signals remains unexplained, in vitro studies have shown that different cell types, including bone-marrow stroma and splenic fibroblasts, can function in this capacity18,31,59,60,66. As indicated earlier, IL-15 can potently promote NKcell development from NKPs14,18,20,32. However, early work clearly showed that IL-2 could maintain NK-cell survival and drive their development from committed precursors16,20–22. As IL-2 and IL-15 use shared cytokine receptor chains for signalling (IL-2Rβ−γc), these two cytokines could be redundant for this function. However, several observations indicate that IL-15 (and not IL-2) is crucial for NK-cell generation under physiological conditions. First, IL-2 is mainly a T-cell product, and T-cell-deficient mice and humans have normal NK-cell differentiation7,8,18,67. Second, IL-2 deficiency in mice and humans is not associated with NK-cell deficiency: NK cells are present in an IL-2-deficient patient68, and although this patient, as well as IL-2-deficient mice69, has lower NK-cell-mediated cytotoxicity, IL-2deficient mice suffer from a severe inflammatory disease that is associated with uncontrolled T-cell activation, which could influence NK-cell activity. Indeed, normal numbers of mature, fully functional NK cells are found in recombinase-activating gene 2 (Rag2) and IL-2 doubledeficient mice (C.A. Vosshenrich and J.P.D., unpublished observations). Third, patients with a defect in the expression of CD25 (IL-2Rα)70 have normal NK-cell phenotype and function (C. Roifman, personal communication). By contrast, absence of IL-15 signalling in humans or mice causes a severe reduction in the number of NK cells71−76. Collectively, these observations convincingly support the notion that IL-15 is the main cytokine for NK-cell development (BOX 2). The cellular context in which IL-15 functions is poorly defined. IL-15 was originally thought to bind a trimeric receptor complex (IL-15Rα–IL-2Rβ–γc). However, IL-15-dependent cells from normal mice are lost when transferred into mice that express IL-15 but lack the high-affinity IL-15Rα chain77. Other groups have observed the expression of IL-15 protein on the cell surface of fibroblasts and epithelial cells78. These observations imply a model whereby IL-15 would, in part, be presented (through IL-15Rα) to IL-15-responsive cells. Recently, Dubois and co-workers79 showed that an IL-15Rα−IL-15 complex can associate intracellularly through an endocytic pathway and be presented in trans to cells that express the IL-2Rβ–γc complex. Accordingly, bone-marrow stromal cells should co-express IL-15Rα

VOLUME 3 | MAY 2003 | 4 1 9


REVIEWS

MISSING-SELF HYPOTHESIS

Almost 20 years ago, Klas Kärre predicted the existence of inhibitory receptors expressed by NK cells that recognize selfMHC class I molecules87. The lack of expression of the relevant MHC by target cells (missing self) consequently activates NK cells and results in target-cell elimination.

and IL-15, which would then trigger the maturing NK cells. However, this model79 provokes many new questions: first, what is the nature of the IL-15Rα-expressing cell that delivers the IL-15 signal, as IL-15Rα is broadly expressed80? Second, what is the relevance of the expression of IL-15Rα by NKPs and NK cells themselves18,24,32? And third, what is the functional capacity of soluble IL-15Rα−IL-15 complexes? Answers to these questions could provide new avenues for therapeutic modulation of NK cells. Receptor tuning and NK-cell self-tolerance. Developing NK cells acquire a complex repertoire of activating and inhibitory NK-cell receptors. Because the overall threshold of NK-cell activation is regulated by a fine balance between activation and inhibition3, the sequence in which these receptors are expressed influences NK-cell self-tolerance. For example, if NK cells express activating but not inhibitory receptors, they might kill indiscriminantly. There are several potential solutions to this problem: first, developing NK cells might express inhibitory receptors before activating receptors; and second, NK-cell selftolerance might be achieved through asynchronous expression of activating receptors and their ligands. Ligands for the NKG2D receptors — such as UL16binding protein 1 (ULBP1), ULBP2 and ULBP3, and the MHC class-I-chain-related molecules MICA and MICB in humans, and minor histocompatibility antigen 60 (H-60) and retinoic acid early transcript (Rae1) in mice — are generally expressed early in embryonic life81 before NK cells are even generated. However, when NKG2Dexpressing NK cells are produced early in life, the ligands are no longer expressed, unless a pathological situation arises, such as infection or tumour transformation82. This

phenomenon might be relevant for other activating receptors also. Third, NK cells might only develop effector functions late in their ontogeny. Evidence has accumulated to show that all of these mechanisms might occur during NK-cell maturation82. Some members of the CD94−NKG2 family of receptors can mediate inhibition after contact with non-classical MHC class I molecules, such as human HLA-E83 and mouse Qa1b (REF. 84). As CD94−NKG2 receptors are expressed before Ly49 receptors in NK-cell development, CD94−NKG2 could provide a mechanism for self-tolerance early in ontogeny58,82. So, the sequence of receptor expression seems to be regulated in an orderly manner with successive receptor expression in both humans and mice59,60,85. The benefit of limiting the expression of receptors for self-MHC by a particular NK cell is evident in the light of the MISSING-SELF HYPOTHESIS. An excess of such receptors would make it difficult for NK cells to detect the downregulation of self-MHC molecules, which can occur during virus infection or tumour transformation86. Paradoxically, the 2B4 receptor, which is normally considered to be an activating receptor, might function early in NK-cell ontogeny to downregulate NK-cell activation. 2B4 can recruit the SRC homology 2 (SH2)containing adaptor protein SAP (signalling lymphocyte-activation molecule-associated protein) and allow cell activation, including the production of cytokines and cytotoxicity. However, 2B4 and SAP are not coordinately expressed during NK-cell development31, so that in immature NK cells 2B4 is expressed in the absence of SAP. As a consequence, the engagement of 2B4 recruits SH2-domain-containing protein tyrosine phosphatases, which results in NK-cell inhibition rather than activation. So, 2B4 might provide a fail-safe mechanism that prevents immature NK cells from auto-aggression.

Box 2 | IL-15: fuel for NK-cell generation in vivo Although interleukin-15 (IL-15) is essential for natural killer (NK)-cell development71,72, how does it promote NK-cell differentiation in vivo ? IL-15 mediates its biological activity through the signalling receptor complex IL-2Rβ–γc (common γ-chain). Does IL-15 promote survival, proliferation or maturation of NK-cell precursors (NKPs)? This is a technically challenging question as these cells are infrequent in bone marrow. However, mouse and human NKPs proliferate following triggering of the IL-2Rβ–γc complex, acquire mature NK-cell markers and become lytic. Whether IL-15 induces maturation or simply maintains survival of pre-programmed NKPs is unknown. Analysis of IL-15-deficient mice has provided some insights (C. A. Vosshenrich and J.P.D., unpublished observations): these mice have low numbers of bone-marrow NK1.1+ NK cells that seem to be mature, cytotoxic and produce cytokines. One downstream effector of IL-15 seems to be the anti-apoptotic factor BCL-2: transgenic overexpression of BCL-2 restores normal numbers of NK cells in IL-2Rβ-deficient mice117 and allows peripheral NK cells to persist in IL-15-deficient hosts after adoptive transfer104,105. Collectively, these data indicate that IL-15 functions mainly as a survival and proliferation factor for developing bone-marrow NK1.1+ cells throughout their differentiation. Although IL-2 is not a physiological factor for NK-cell differentiation (see main text), under certain conditions, IL-2 can markedly influence NK-cell homeostasis118. Low doses of exogenously administered IL-2 can selectively expand CD56hi NK-cell populations in humans115, as this subset constitutively expresses CD25 (IL-2Rα). The effect of IL-2 in these conditions is similar to that observed with IL-15 and predominantly involves increases in the number of NK cells, although increases in lytic capacity have also been observed.

420


Shaping the killer’s weapons: cytolysis. NK cells were discovered by virtue of their cytolytic capacity. With time, it has become clear that the ‘killing’ machinery of NK cells can be triggered by several distinct pathways, and that phenotypically distinct NK-cell subsets, which vary in their cytolytic capacity, can be identified when freshly isolated from human blood87. The main pathway of NKcell-mediated cytolysis is dependent on perforin and granzymes; however, other mechanisms of inducing target-cell lysis have been described, including the role of FASL (CD178) and tumour-necrosis factor-related apoptosis-inducing ligand (TRAIL)-dependent receptors88. As stated earlier, human immature NK cells (CD56–CD161+) were unable to lyse classical NK-cell-sensitive targets (through perforin- or CD95L-dependent mechanisms) but were able to mediate TRAIL-dependent cytotoxicity20,57. Subsequent expression of CD56 correlated with the acquisition of perforin- and CD95L-dependent cytotoxicity20−23. Similarly, the perforin-dependent lytic potential of developing mouse NK cells in vivo seems to be a late event: among the NK cells that are present in the bone marrow, only the most mature (Ly49-expressing) cells are cytolytic, and have less activity compared with splenic NK cells18.



REVIEWS

Box 3 | Abnormal NK-cell development in human pathological states There are several reasons to believe that malignant transformation of early myeloid progenitors influences natural killer (NK)-cell development and, likewise, that defects in NK-cell development might lead to malignant transformation. Chronic myeloid leukaemia (CML) is a malignant disorder of stem cells that is caused by the fusion of the breakpoint cluster region (BCR) gene located on chromosome 22 with the v-abl Abelson mouse leukaemia viral oncogene homologue (ABL) on chromosome 9, resulting in a cascade of aberrant cell signalling. Miller and colleagues119 have shown that NK cells from patients with CML late in the disease express the BCR−ABL fusion gene product, whereas T cells and B cells do not, and that the number of NK-cell progenitors is markedly decreased in these patients. Furthermore, transfection of normal CD34+ haematopoietic stem cells (HSCs) with the BCR–ABL fusion gene product resulted in a decline in NK-cell differentiation in vitro, and co-culture of normal CD34+ HSCs with BCR–ABL+ HSCs inhibited NK-cell development in vitro. These studies, therefore, show that the CML stem cell includes an NK-cell progenitor and that the malignant clone has both direct and indirect effects on NK-cell development. Similarly, it is becoming increasingly clear that abnormalities in growth factors or growth-factor receptors that are involved in NK-cell development can lead to malignant transformation. As discussed in the main text, the CD34+FLT3+c-KIT+ cell probably identifies at least one lymphoid cell progenitor population. A somatic mutation in FMS-like tyrosine kinase 3 (FLT3), which results in an internal tandem duplication and constitutive receptor activation, has now been identified in up to 30% of patients with acute myeloid leukaemia (AML) and in some patients with acute lymphoblastic leukaemia (ALL)120. Similarly, activating somatic mutations in c-KIT have been identified in patients with AML121 and patients with large granular NK-cell leukaemia or lymphoma122. Evidence of aberrant IL-15 production or IL-15-receptor signalling that results in NK-cell leukaemia has not been reported. However, in a transgenic mouse model in which the 5′ and 3′ translational regulatory elements were removed for more efficient translation and secretion of mouse IL-15, most mice eventually developed clonal NKT or NK lymphoblastic leukaemia 3 to 5 months after a polyclonal expansion of both NK-cell and CD8+ memory T-cell populations123,124. Hence, constitutive activation of any of these three growth signalling pathways that are involved in NK-cell development and homeostasis potentially has an important role in malignant transformation of either human or experimental acute leukaemia

ANTIBODY-DEPENDENT CELLULAR CYTOTOXICITY (ADCC). Cells that are coated

with antibodies are targeted and destroyed by NK cells, macrophages and granulocytes. On NK cells, ADCC is mediated by the Fc receptor complex (CD16).

NK cells mediate ANTIBODY-DEPENDENT CELLULAR CYTOTOXICITY (ADCC) through an Fc-receptor complex (CD16), which allows NK cells to participate in the elimination of antibody-targeted cells. Most mouse NK cells and most (~90%) human CD56+ NK cells express CD16. What triggers expression of CD16 by developing NK cells is unknown. In vitro experiments have shown that IL-21 can synergize with FLT3L/c-KIT ligand and IL-15 in promoting the generation of NK cells from human CD34+ cells89. Evidence was also provided indicating that NK cells that were generated in vitro with IL-21 had a different phenotype (CD16+CD56+) than previously observed using FLT3L/c-KIT ligand and IL-15, where most cells were CD16–. This indicates that IL-21 might have a role in the later stages of human NK-cell maturation, although this observation awaits independent confirmation. Although IL-21 stimulated the production of cytokines and lytic activity of mature NK cells in mice and humans89, it does not seem to be essential for NK-cell development as IL-21R-deficient mice have normal steady-state numbers of NK cells, which have a normal phenotype90. Studies of human pathologies and analyses of genetargeted mice have provided important new insights into the molecular mechanisms that control NK-cell


cytotoxicity (TABLE 1, BOX 3). Patients with mutations in nuclear factor-κB (NF-κB) essential modulator (NEMO) have defective natural cytotoxicity, despite normal numbers of NK cells and normal ADCC91. One of these patients also suffered from recurrent infections with human cytomegalovirus. These results are consistent with a role for NF-κB in the regulation of perforin expression in vitro92. Mice deficient for CCAAT-enhancer binding protein-γ (CEBPγ), a transcriptional co-regulator, have normal numbers of NK cells, but have impaired NK-cell cytotoxic activity and IFN-γ production93. IRF2, which is a member of the IRF family, generally antagonizes IRF1. IRF2-deficient mice, similar to IRF1-deficient mice, have reduced numbers of NK cells, which are immature and nonlytic. However, the nature of the cell-intrinsic defect in IRF2-deficient mice is unknown94. The bHLH transcription factor microphthalmia-associated transcription factor (MITF) is encoded by the mi allele. mi/mi mice have normal numbers of NK cells, which do not express perforin and therefore cannot kill tumour cells95. The mi allele presumably disrupts the nuclear translocation of transcription factors that are directly involved in activation of the gene that encodes perforin. Similarly, the ETS-family member myeloid elf-like factor (MEF) regulates the perforin gene promoter. MEFdeficient NK cells fail to express perforin and cannot kill tumour cells96. In both mi/mi and MEF-deficient mice, the cytotoxic activity of CD8+ T cells was only mildly affected, and in MEF-deficient mice, IL-2 could restore the expression of perforin by cytotoxic T cells but not by NK cells. So, the regulation of NK-cell lytic function is regulated by many transcription factors and in ways that differ from that of cytotoxic T cells. Shaping the killer’s weapons: cytokines. Several questions remain concerning the acquisition and control of cytokine production by NK cells. When do developing NK cells become competent to produce cytokines? Do cytokine profiles change with NK-cell maturation? In human NK cells, there seems to be a ‘division of labour’ (FIG. 3b). Subsets of NK cells with relatively poor natural cytolytic capacity (CD16−CD56hi and CD16+CD56hi) are potent producers of cytokines, such as IFN-γ, tumour-necrosis factor (TNF), LTβ, granulocyte− macrophage colony-stimulating factor (GM-CSF), IL-10 and IL-13, following selective monokine stimulation97. What is the significance of this heterogeneity in cytokine production? Freshly isolated CD56hi NK cells proliferate in response to IL-2 or IL-15 and are potent producers of both type 1 (IFN-γ) and type 2 (IL-5 and IL-13) cytokines compared with the CD56low NK-cell subset, which does not readily proliferate, produces low levels of cytokines yet shows enhanced natural cytotoxicity87,97. As culture of CD34+ HSCs with IL-15 generates mainly ‘pseudomature’ CD16–CD56hiKIR− NK cells (and few, if any, CD16+CD56low NK cells23), the data indicate that the CD56hi NK-cell subset is less differentiated than the CD56low NK-cell subset. However, the factors that are responsible for this CD56hi to CD56low maturation

VOLUME 3 | MAY 2003 | 4 2 1


REVIEWS

NKT CELLS

T cells that express antigen receptors of limited variability and that share markers and functions with NK cells. NKT cells are potent producers of cytokines and are heterogeneous in terms of target recognition, anatomical localization and functions.

pathway have not been identified. Bendelac and colleagues98 have elegantly carried out in situ studies tracking Vα14+ NKT CELLS using CD1d tetramers, and have shown that these cells undergo a T helper 2 (TH2) to TH1 conversion as they mature in the thymus. Similarly, Loza and Perussia99 developed an in vitro culture system and reported that a type 2 to type 0 to type 1 sequential cytokine conversion of human NK cells occurs as they mature from early (CD56− CD161+) to late (CD56+CD161−) stages, indicating that cytokine profiles of NK cells, similar to NKT cells, might be developmentally controlled. Moreover, this group identified distinct subsets of IL-13- and IFN-γproducing human NK cells in cord blood after stimulation100, indicating that type 2 and type 1 NK-cell subsets were non-overlapping. However, earlier work showed that mature CD56+ NK cells cultured under polarizing conditions could generate NK-cell lines101,102 that have a mixed type 0 profile (IFN-γ, IL-5 and IL-13). In addition, CD56+ NK-cell clones99 have been described, which have either type 1, type 0 or type 2 cytokine profiles. There is no clear evidence for functionally distinct — for example, cytokine-producing versus cytolytic — NK-cell subsets in mice. Moreover, a homologue of the human CD56 molecule does not exist in the mouse genome, so comparative studies have not been carrried out. Bulk cultures of mouse splenic NK cells can produce IFN-γ, IL-10, IL-13, TNF and GM-CSF after stimulation with IL-12, and IL-18 in combination with phorbol 12-myristate 13-acetate (PMA) and ionomycin102,103; however, whether all NK cells produce these cytokines or whether there are distinct subsets with selective cytokine-production is unknown. Careful analysis of NK-cell differentiation in mice that are mutant for those factors known to be involved in establishing cytokine production profiles in lymphocytes might provide important clues in this rapidly evolving field. Stage 3: dispatching killers to the frontline

Life span. The life span of mature NK cells is unknown. When mature splenic NK cells are labelled with fluorescent tracking dyes and adoptively transferred to normal adult mice they can be detected in the circulation for at least 5 weeks104. Interestingly, when these NK cells are adoptively transferred to mice that genetically lack IL-15, labelled NK cells are no longer detectable in blood, bone marrow, liver or spleen within 5 days, indicating that IL-15 is required not only for NK-cell development, but also for NK-cell homeostasis in vivo104,105. Moreover, NK cells seem to compete for peripheral sources of IL-15. Transfer of NK cells into NK-cell-deficient recipients shows that mature NK cells can ‘homeostatically’ proliferate104. By contrast, transfer to NK-cell-sufficient mice blocks this proliferation, which is presumably due to competition for IL-15. The availability of IL-15 might, therefore, be tightly controlled, perhaps through the regulation of expression of IL-15Rα. This is also consistent with the fact that it is difficult to detect soluble IL-15 in culture supernatants or in vivo23,66.

422


Trafficking. Mature NK cells are found in the circulation and in several tissues, including the bone marrow, spleen, liver, lung, omentum, intestine and placenta (FIG. 3b). We know very little about the mechanisms that direct NK cells to these sites under normal conditions, although the expression of various chemokine receptors and adhesion molecules by NK cells indicates a potential mechanism. In humans, the CD56low NK-cell subset shows marked chemotaxis in response to the chemokine CXCL8 (also known as IL-8) and soluble fractalkine, similar to neutrophils, whereas the CD56hi NK-cell subset has highlevel expression of CC-chemokine receptor 7 (CCR7) and L-selectin, which are associated with homing to the lymph nodes106. Consistent with this, Fehniger et al.107 have shown that ten times more CD56hi NK cells are found in the parafollicular (T-cell) regions of normal human lymph nodes than in the blood. Similarly, the CD56hi NK-cell subset is the main NK-cell subset present in the placenta during pregnancy108,109. These data indicate unique migratory properties for CD56hi and CD56low NK cells and, together with the immunoregulatory role of the CD56hi NK-cell subset110, support the notion of distinct development of NK-cell subsets in humans. What remains to be determined is whether the cytokine-producing CD56hi NK-cell subset is the less differentiated precursor of the more cytolytic CD56low NK-cell subset, and so whether the lymph node itself might be the site of this putative NK-cell maturation process. The absence of expression of CCR7 by CD56low NK cells could promote their exit from the lymph node following maturation from the CD56hi to CD56low phenotype. Two additional observations are consistent with this notion. First, only CD56hi NK cells can be generated from CD34+ haematopoietic bone-marrow progenitor cells in vitro, indicating that the CD56low NK-cell subset requires an alternative environment for maturation110. Second, following allogeneic bone-marrow transplantation111 or prolonged in vivo administration of low doses of IL-2 (REFS 112,113), there is an increase in the size of the CD56hi NK-cell population in the blood compared with the CD56low NK-cell population, indicating that the role of the bone marrow in NK-cell differentiation ends after emergence of the CD56hi NK-cell subset. As indicated earlier, evidence for distinct NK-cell subsets in mice is lacking, although it seems plausible that such functional specification takes place. Little is known about the mechanisms that control NK-cell homing in mice, and the observations of distinct expression of chemokine receptors by human NK-cell subsets provide an interesting potential model that could be further analysed in mice. By contrast, we know some of the mechanisms that control trafficking of NK cells in mouse models of virus infection. Biron and colleagues26 have elegantly shown that hepatic production of macrophage inflammatory protein 1α (MIP1α, also known as CCL3), which is induced by IFN-α/β, is required for mouse NK-cell homing to the liver and effective control of mouse cytomegalovirus (MCMV) infection. Yokoyama and colleagues25 have further shown that MCMV induces expansion of the Ly49H-expressing mouse NK-cell subset that specifically recognizes a



REVIEWS virus product expressed by infected cells. This increase in Ly49H-expressing cells might be a consequence of the influx from the spleen and blood into the liver. In a different model of tumour rejection, Smyth and colleagues27 have shown that NK cells can be recruited to the site of tumour challenge in a TNF-dependent manner. These observations show that NK cells can migrate to the site of infections and tumours, in response to chemokine and cytokine gradients. Making the killer work in the clinic

Data from animal models indicate that antibody-mediated reduction of tumours is, in part, dependent on the balance between inhibitory and activating Fc receptors (FcRs) that are expressed by innate immune effector cells114. NK cells can mediate ADCC of tumour cells through CD16. Furthermore, human NK-cell populations can be markedly expanded and activated in vivo with low and moderately high doses of IL-2, respectively115. Therefore, several clinical trials are being carried out at present in patients with lymphoma or breast cancer, combining IL-2 treatment with humanized monoclonal antibodies, rituxan or trastuzumab, respectively.

1.

Matsunaga, T. & Rahman, A. What brought the adaptive immune system to vertebrates? The jaw hypothesis and the seahorse. Immunol. Rev. 166, 177−186 (1998). 2. Medzhitov, R. & Janeway, C. Jr. Innate immune recognition: mechanisms and pathways. Immunol. Rev. 173, 89−97 (2000). 3. Colucci, F., Di Santo, J. P. & Leibson, P. J. Natural killer cell activation in mice and men: different triggers for similar weapons? Nature Immunol. 3, 807−813 (2002). A recent update on receptors, ligands and signalling molecules that regulate natural killer (NK)-cell activation. 4. Godin, I. & Cumano, A. The hare and the tortoise: an embryonic haematopoietic race. Nature Rev. Immunol. 2, 593−604 (2002). An excellent review of haematopoiesis with special emphasis on the comparison between formation of blood cells in the embryo and in the adult. 5. Seaman, W. E., Gindhart, T. D., Greenspan, J. S., Blackman, M. A. & Talal, N. Natural killer cells, bone, and the bone marrow: studies in estrogen-treated mice and in congenitally osteopetrotic (mi/mi) mice. J. Immunol. 122, 2541−2547 (1979). 6. Kumar, V., Ben-Ezra, J., Bennett, M. & Sonnenfeld, G. Natural killer cells in mice treated with 89strontium: normal target-binding cell numbers but inability to kill even after interferon administration. J. Immunol. 123, 1832−1838 (1979). 7. Herberman, R. B., Nunn, M. E., Holden, H. T. & Lavrin, D. H. Natural cytotoxic reactivity of mouse lymphoid cells against syngeneic and allogeneic tumors. II. Characterization of effector cells. Int. J. Cancer 16, 230−239 (1975). 8. Sirianni, M. C., Businco, L., Seminara, R. & Aiuti, F. Severe combined immunodeficiencies, primary T-cell defects and DiGeorge’s syndrome in humans: characterization by monoclonal antibodies and natural killer cell activity. Clin. Immunol. Immunopathol. 28, 361−370 (1983). 9. Sihvola, M. & Hurme, M. The development of NK cell activity in thymectomized bone marrow chimeras. Immunology 53, 17−22 (1984). 10. Ramos, S. B., Garcia, A. B., Viana, S. R., Voltarelli, J. C. & Falcao, R. P. Phenotypic and functional evaluation of natural killer cells in thymectomized children. Clin. Immunol. Immunopathol. 81, 277−281 (1996). 11. Schwarz, R. E. & Hiserodt, J. C. Effects of splenectomy on the development of tumor-specific immunity. J. Surg. Res. 48, 448−453 (1990). 12. Passlick, B. et al. Posttraumatic splenectomy does not influence human peripheral blood mononuclear cell subsets. J. Clin. Lab. Immunol. 34, 157−161 (1991).

The discovery that inhibitory KIRs recognize MHC class I molecules is also being explored in clinical trials of HLA-mismatched T-cell-depleted bone-marrow transplantation for patients with relapsed acute myeloid leukaemia. Patients who have a KIR–MHC class I mismatch in the direction of the donor NK cells to recipient leukaemia have less clinical disease relapses compared to patients who undergo the same transplant but lack the KIR−MHC class I mismatch in the same donor-recipient direction. These results can probably be explained on the basis of the conditioning regimen that is used and on the pattern of NK-cell recognition116. Prospective studies to confirm these exciting results are ongoing. Concluding remarks

In recent years, there have been several advances that have enriched our knowledge of NK-cell development, differentiation, trafficking and effector functions. An improved understanding of these processes will continue to drive the clinical applications for NK cells over the next decade in the fight against infection and tumours, and in the suppression of autoimmune manifestations.

13. Phillips, J. H. et al. Ontogeny of human natural killer cells: fetal NK cells mediate cytolytic functions and express cytoplasmic CD3ε, δ proteins. J. Exp. Med. 175, 1055−1063 (1992). 14. Jaleco, A. C. et al. Fetal liver contains committed NK progentiors, but is not a site for the development of CD34+ cells into T cells. J. Immunol. 159, 694−702 (1997). 15. Mingari, M. C. et al. In vitro proliferation and cloning of CD3− CD16+ cells from human thymus precursors. J. Exp. Med. 174, 21−30 (1991). 16. Rodewald, H.-R. et al. A population of early fetal thymocytes expressing FcγR II/III contains precursors of T lymphocytes and natural killer cells. Cell 69, 139−146 (1992). 17. Ikawa, T., Kawamoto, H., Fujimoto, S. & Katsura, Y. Commitment of common T/natural killer (NK) progenitors to unipotent T and NK progenitors in the murine fetal thymus revealed by a single progenitor assay. J. Exp. Med. 190, 1617−1626 (1999). 18. Rosmaraki, E. E. et al. Identification of committed NK cell progenitors in adult murine bone marrow. Eur. J. Immunol. 31, 1900−1909 (2001). The first description of a committed NK-cell precursor in the bone marrow of adult mice. Its human counterpart has not been identified. 19. Spits, H. et al. Early stages in the development of human T, natural killer and thymic dendritic cells. Immunol. Rev. 165, 75−86 (1998). 20. Bennett, I. M. et al. Definition of a natural killer NKRP1A+/CD56−/CD16− functionally immature human NK cell subset that differentiates in vitro in the presence of interleukin 12. J. Exp. Med. 184, 1845−1856 (1996). 21. Miller, J. S., Alley, K. A. & McGlave, P. Differentiation of NK cells from human primitive marrow progenitors in a stromabased long-term culture system: identification of a CD34+7+ NK progenitor. Blood 83, 2594−2601 (1994). 22. Shibuya, A., Nagayoshi, K., Nakamura, K. & Nakauchi, H. Lymphokine requirement for the generation of natural killer cells from CD34+ hematopoietic progenitor cells. Blood 85, 3538−3546 (1995). 23. Mrozek, E., Anderson, P. & Caligiuri, M. A. Role of interleukin-15 in the development of human CD56+ natural killer cells from CD34+ hematopoietic progenitor cells. Blood 87, 2632−2640 (1996). 24. Williams, N. S. et al. Generation of lytic natural killer 1.1+, Ly-49− cells from multipotential murine bone marrow progenitors in a stroma-free culture: definition of cytokine requirements and developmental intermediates. J. Exp. Med. 186, 1609−1614 (1997). Reference 23 and 24 describe the minimal cytokine requirements (stromal-cell free) for generating NK cells


25.

26.

27. 28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

in vitro. This process is remarkably similar in humans and mice. Dokun, A. O. et al. Specific and nonspecific NK cell activation during virus infection. Nature Immunol. 2, 951−956 (2001). Salazar-Mather, T. P., Orange, J. S. & Biron, C. A. Early murine cytomegalovirus (MCMV) infection induces liver natural killer (NK) cell inflammation and protection through macrophage inflammatory protein 1α (MIP-1α)-dependent pathways. J. Exp. Med. 187, 1−14 (1998). Smyth, M. J. et al. Differential tumor surveillance by natural killer (NK) and NKT cells. J. Exp. Med. 191, 661−668 (2000). Akashi, K., Traver, D., Miyamoto, T. & Weissman, I. L. A clonogenic common myeloid progenitor that gives rise to all myeloid lineages. Nature 404, 193−197 (2000). Kondo, M., Weissman, I. L. & Akashi, K. Identification of clonogenic common lymphoid progenitors in mouse bone marrow. Cell 91, 661−672 (1997). Igarashi, H., Gregory, S. C., Yokota, T., Sakaguchi, N. & Kincade, P. W. Transcription from the RAG1 locus marks the earliest lymphocyte progenitors in bone marrow. Immunity 17, 117−130 (2002). Sivori, S. et al. Early expression of triggering receptors and regulatory role of 2B4 in human natural killer cell precursors undergoing in vitro differentiation. Proc. Natl Acad. Sci. USA 99, 4526−4531 (2002). Yu, H. et al. Flt3 ligand promotes the generation of a distinct CD34+ human natural killer cell progenitor that responds to interleukin-15. Blood 92, 3647−3657 (1998). Lyman, S. D. & Jacobsen, S. E. c-kit ligand and Flt3 ligand: stem/progenitor cell factors with overlapping yet distinct activities. Blood 91, 1101−1134 (1998). Colucci, F. & Di Santo, J. P. The receptor tyrosine kinase c-kit provides a critical signal for survival, expansion, and maturation of mouse natural killer cells. Blood 95, 984−991 (2000). Waskow, C., Paul, S., Haller, C., Gassmann, M. & Rodewald, H. Viable c-Kit(W/W) mutants reveal pivotal role for c-kit in the maintenance of lymphopoiesis. Immunity 17, 277−288 (2002). Sitnicka, E. et al. Key role of flt3 ligand in regulation of the common lymphoid progenitor but not in maintenance of the hematopoietic stem cell pool. Immunity 17, 463−472 (2002). Mackarehtschian, K. et al. Targeted disruption of the flk2/flt3 gene leads to deficiencies in primitive hematopoietic progenitors. Immunity 3, 147−161 (1995). McKenna, H. J. et al. Mice lacking flt3 ligand have deficient hematopoiesis affecting hematopoietic progenitor cells, dendritic cells, and natural killer cells. Blood 95, 3489−3497 (2000).

VOLUME 3 | MAY 2003 | 4 2 3


REVIEWS 39. Spritz, R. A. Molecular basis of human piebaldism. J. Invest. Dermatol. 103, S137−S140 (1994). 40. Billia, F., Barbara, M., McEwen, J., Trevisan, M. & Iscove, N. N. Resolution of pluripotential intermediates in murine hematopoietic differentiation by global complementary DNA amplification from single cells: confirmation of assignments by expression profiling of cytokine receptor transcripts. Blood 97, 2257−2268 (2001). 41. Warren, L. A. & Rothenberg, E. V. Regulatory coding of lymphoid lineage choice by hematopoietic transcription factors. Curr. Opin. Immunol. 15, 166−175 (2003). 42. Wang, J. H. et al. Selective defects in the development of the fetal and adult lymphoid system in mice with an Ikaros null mutation. Immunity 5, 537−549 (1996). 43. Tucker, S. N., Jessup, H. K., Fujii, H. & Wilson, C. B. Enforced expression of the Ikaros isoform IK5 decreases the numbers of extrathymic intraepithelial lymphocytes and natural killer 1.1+ T cells. Blood 99, 513−519 (2002). 44. Nichogiannopoulou, A., Trevisan, M., Neben, S., Friedrich, C. & Georgopoulos, K. Defects in hemopoietic stem cell activity in Ikaros mutant mice. J. Exp. Med. 190, 1201−1214 (1999). 45. Bories, J. C. et al. Increased T-cell apoptosis and terminal B-cell differentiation induced by inactivation of the Ets-1 proto-oncogene. Nature 377, 635−638 (1995). 46. Barton, K. et al. The Ets-1 transcription factor is required for the development of natural killer cells in mice. Immunity 9, 555−563 (1998). 47. Scott, E. W., Simon, M. C., Anastasi, J. & Singh, H. Requirement of transcription factor PU.1 in the development of multiple hematopoietic lineages. Science 265, 1573− 1577 (1994). 48. Colucci, F. et al. Differential requirement for the transcription factor PU.1 in the generation of natural killer cells versus B and T cells. Blood 97, 2625−2632 (2001). 49. DeKoter, R. P., Lee, H. J. & Singh, H. PU.1 regulates expression of the interleukin-7 receptor in lymphoid progenitors. Immunity 16, 297−309 (2002). 50. Radtke, F. et al. Notch1 deficiency dissociates the intrathymic development of dendritic cells and T cells. J. Exp. Med. 191, 1085−1094 (2000). 51. Schaniel, C., Bruno, L., Melchers, F. & Rolink, A. G. Multiple hematopoietic cell lineages develop in vivo from transplanted Pax5-deficient pre-B I-cell clones. Blood 99, 472−478 (2002). 52. Engel, I. & Murre, C. The function of E- and Id proteins in lymphocyte development. Nature Rev. Immunol. 1, 193− 199 (2001). 53. Heemskerk, M. H. et al. Inhibition of T cell and promotion of natural killer cell development by the dominant negative helix loop helix factor Id3. J. Exp. Med. 186, 1597−1602 (1997). 54. Spits, H., Couwenberg, F., Bakker, A. Q., Weijer, K. & Uittenbogaart, C. H. Id2 and Id3 inhibit development of CD34+ stem cells into predendritic cell (pre-DC)2 but not into pre-DC1. Evidence for a lymphoid origin of pre-DC2. J. Exp. Med. 192, 1775−1784 (2000). 55. Yokota, Y. et al. Development of peripheral lymphoid organs and natural killer cells depends on the helix-loop-helix inhibitor Id2. Nature 397, 702−726 (1999). The first example of a transcription factor that selectively affects NK-cell development. 56. Ikawa, T., Fujimoto, S., Kawamoto, H., Katsura, Y. & Yokota, Y. Commitment to natural killer cells requires the helix-loophelix inhibitor Id2. Proc. Natl Acad. Sci. USA 98, 5164−5169 (2001). 57. Zamai, L. et al. Natural killer (NK) cell-mediated cytotoxicity: differential use of TRAIL and Fas ligand by immature and mature primary human NK cells. J. Exp. Med. 188, 2375− 2380 (1998). 58. Williams, N. S. et al. Differentiation of NK1.1+, Ly49+ NK cells from flt3+ multipotent marrow progenitor cells. J. Immunol. 163, 2648−2656 (1999). 59. Roth, C., Carlyle, J. R., Takizawa, H. & Raulet, D. H. Clonal acquisition of inhibitory Ly49 receptors on developing NK cells is successively restricted and regulated by stromal class I MHC. Immunity 13, 143−153 (2000). 60. Miller, J. S. & McCullar, V. Human natural killer cells with polyclonal lectin and immunoglobulin-like receptors develop from single hematopoietic stem cells with preferential expression of NKG2A and KIR2DL2/L3/S2. Blood 98, 705−713 (2001). 61. Kim, S. et al. In vivo developmental stages in murine natural killer cell maturation. Nature Immunol. 3, 523−528 (2002). 62. Iizuka, K. et al. Requirement for membrane lymphotoxin in natural killer cell development. Proc. Natl Acad. Sci. USA 96, 6336−6340 (1999). 63. Wu, Q. et al. Signal via lymphotoxin-β R on bone marrow stromal cells is required for an early checkpoint of NK cell development. J. Immunol. 166, 1684−1689 (2001).

424

64. Ogasawara, K. et al. Requirement for IRF-1 in the microenvironment supporting development of natural killer cells. Nature 391, 700−703 (1998). 65. Ohteki, T. et al. The transcription factor interferon regulatory factor 1 (IRF-1) is important during the maturation of natural killer 1.1+ T cell receptor-αβ+ (NK1+ T) cells, natural killer cells, and intestinal intraepithelial T cells. J. Exp. Med. 187, 967−972 (1998). 66. Briard, D., Brouty-Boye, D., Azzarone, B. & Jasmin, C. Fibroblasts from human spleen regulate NK cell differentiation from blood CD34+ progenitors via cell surface IL-15. J. Immunol. 168, 4326−4332 (2002). 67. Puel, A., Ziegler, S. F., Buckley, R. H. & Leonard, W. J. Defective IL-7R expression in T−B+NK+ severe combined immunodeficiency. Nature Genet. 20, 394−397 (1998). 68. DiSanto, J. P. et al. Absence of interleukin 2 production in a severe combined immunodeficiency disease syndrome with T cells. J. Exp. Med. 171, 1697−1704 (1991). 69. Kundig, T. M. et al. Immune responses in interleukin-2deficient mice. Science 262, 1059−1061 (1993). 70. Sharfe, N., Dadi, H. K., Shahar, M. & Roifman, C. M. Human immune disorder arising from mutation of the α chain of the interleukin-2 receptor. Proc. Natl Acad. Sci. USA 94, 3168− 3171 (1997). 71. Kennedy, M. K. et al. Reversible defects in natural killer and memory CD8 T cell lineages in interleukin 15-deficient mice. J. Exp. Med. 191, 771−780 (2000). 72. Lodolce, J. P. et al. IL-15 receptor maintains lymphoid homeostasis by supporting lymphocyte homing and proliferation. Immunity 9, 669−676 (1998). References 71 and 72 illustrate the essential and nonredundant role of IL-15 and its receptor in mouse NKcell development. A defect in the human IL-15 receptor results in NK-cell deficiency , as described in reference 46. 73. Suzuki, H., Duncan, G. S., Takimoto, H. & Mak, T. W. Abnormal development of intestinal intraepithelial lymphocytes and peripheral natural killer cells in mice lacking the IL-2 receptor β chain. J. Exp. Med. 185, 499−505 (1997). 74. Di Santo, J. P., Müller, W., Guy-Grand, D., Fischer, A. & Rajewsky, K. Lymphoid development in mice with a targeted deletion of the interleukin-2 receptor γ chain. Proc. Natl Acad. Sci. USA 92, 377−381 (1995). 75. Noguchi, M. et al. Interleukin-2 receptor γ chain mutation results in X-linked severe combined immunodeficiency in humans. Cell 73, 147−157 (1993). 76. Gilmour, K. C. et al. Defective expression of the interleukin2/interleukin-15 receptor β subunit leads to a natural killer cell-deficient form of severe combined immunodeficiency. Blood 98, 877−879 (2001). 77. Lodolce, J. P., Burkett, P. R., Boone, D. L., Chien, M. & Ma, A. T cell-independent interleukin 15Rα signals are required for bystander proliferation. J. Exp. Med. 194, 1187−1194 (2001). 78. Pereno, R. et al. IL-15/IL-15Rα intracellular trafficking in human cells and protection from apoptosis. Ann. NY Acad. Sci. 876, 236−245 (1999). 79. Dubois, S., Mariner, J., Waldmann, T. A. & Tagaya, Y. IL-15Rα recycles and presents IL-15 in trans to neighboring cells. Immunity 17, 537−547 (2002). This paper proposes a new model for the function of IL-15. 80. Giri, J. G. et al. Identification and cloning of a novel IL-15 binding protein that is structurally related to the α chain of the IL-2 receptor. EMBO J. 14, 3654−3663 (1995). 81. Cerwenka, A. & Lanier, L. L. Ligands for natural killer cell receptors: redundancy or specificity. Immunol. Rev. 181, 158−169 (2001). 82. Raulet, D. H., Vance, R. E. & McMahon, C. W. Regulation of the natural killer cell receptor repertoire. Annu. Rev. Immunol. 19, 291−330 (2001). 83. Braud, V. M. et al. HLA-E binds to natural killer cell receptors CD94/NKG2A, B and C. Nature 391, 795−799 (1998). 84. Vance, R. E., Kraft, J. R., Altman, J. D., Jensen, P. E. & Raulet, D. H. Mouse CD94/NKG2A is a natural killer cell receptor for the nonclassical major histocompatibility complex (MHC) class I molecule Qa-1(b). J. Exp. Med. 188, 1841−1848 (1998). 85. Smith, H. R. et al. Nonstochastic coexpression of activation receptors on murine natural killer cells. J. Exp. Med. 191, 1341−1354 (2000). 86. Ljunggren, H. G. & Karre, K. In search of the ‘missing self’: MHC molecules and NK cell recognition. Immunol. Today 11, 237−244 (1990). 87. Lanier, L. L., Le, A. M., Civin, C. I., Loken, M. R. & Phillips, J. H. The relationship of CD16 (Leu-11) and Leu-19 (NKH-1) antigen expression on human peripheral blood NK cells and cytotoxic T lymphocytes. J. Immunol. 136, 4480−4486 (1986).


88. Smyth, M. J., Hayakawa, Y., Takeda, K. & Yagita, H. New aspects of natural-killer-cell surveillance and therapy of cancer. Nature Rev. Cancer 2, 850−861 (2002). An excellent review on recent advances in our understanding of the role of NK cells in cancer and the possibility of using them in therapy. 89. Parrish-Novak, J. et al. Interleukin 21 and its receptor are involved in NK cell expansion and regulation of lymphocyte function. Nature 408, 57−63 (2000). 90. Kasaian, M. T. et al. IL-21 limits NK cell responses and promotes antigen-specific T cell activation: a mediator of the transition from innate to adaptive immunity. Immunity 16, 559−569 (2002). 91. Orange, J. S. et al. Deficient natural killer cell cytotoxicity in patients with IKK-γ/NEMO mutations. J. Clin. Invest. 109, 1501−1509 (2002). 92. Zhou, J., Zhang, J., Lichtenheld, M. G. & Meadows, G. G. A role for NF-κB activation in perforin expression of NK cells upon IL-2 receptor signaling. J. Immunol. 169, 1319−1325 (2002). 93. Kaisho, T. et al. Impairment of natural killer cytotoxic activity and interferon-γ production in CCAAT/enhancer binding protein γ-deficient mice. J. Exp. Med. 190, 1573−1582 (1999). 94. Lohoff, M. et al. Deficiency in the transcription factor interferon regulatory factor (IRF)-2 leads to severely compromised development of natural killer and T helper type 1 cells. J. Exp. Med. 192, 325−336 (2000). 95. Ito, A. et al. Inhibitory effect on natural killer activity of microphthalmia transcription factor encoded by the mutant mi allele of mice. Blood 97, 2075−2083 (2001). 96. Lacorazza, H. D. et al. The ETS protein MEF plays a critical role in perforin gene expression and the development of natural killer and NK-T cells. Immunity 17, 437−449 (2002). 97. Cooper, M. A. et al. Human natural killer cells: a unique innate immunoregulatory role for the CD56bright subset. Blood 97, 3146−3151 (2001). This paper provides the first clear evidence that human NK cells can be divided in two distinct subsets on the basis of expression of cell-surface markers and division of labour. Whether this applies to mouse NK cells is not known. 98. Benlagha, K., Kyin, T., Beavis, A., Teyton, L. & Bendelac, A. A thymic precursor into the NK T cell lineage. Science 296, 553−555 (2002). 99. Loza, M. J. & Perussia, B. Final steps of natural killer cell maturation: a model for type 1–type 2 differentiation? Nature Immunol. 2, 917−924 (2001). 100. Loza, M. J., Peters, S. P., Zangrilli, J. G. & Perussia, B. Distinction between IL-13 and IFN-γ natural killer cells and regulation of their pool size by IL-4. Eur. J. Immunol. 32, 413−423 (2002). 101. Peritt, D. et al. Differentiation of human NK cells into NK1 and NK2 subsets. J. Immunol. 161, 5821−5824 (1998). 102. Hoshino, T., Wiltrout, R. H. & Young, H. A. IL-18 is a potent coinducer of IL-13 in NK and T cells: a new potential role for IL-18 in modulating the immune response. J. Immunol. 162, 5070−5077 (1999). 103. Lauwerys, B. R., Garot, N., Renauld, J. C. & Houssiau, F. A. Cytokine production and killer activity of NK/T-NK cells derived with IL-2, IL-15, or the combination of IL-12 and IL-18. J. Immunol. 165, 1847−1853 (2000). 104. Ranson, T. et al. IL-15 is an essential mediator of peripheral NK cell homeostasis. Blood (in the press) 105. Cooper, M. A. et al. In vivo evidence for a dependence on interleukin 15 for survival of natural killer cells. Blood 100, 3633−3638 (2002). 106. Frey, M. et al. Differential expression and function of L-selectin on CD56bright and CD56dim natural killer cell subsets. J. Immunol. 161, 400−408 (1998). 107. Fehniger, T. A. et al. CD56bright natural killer cells are present in human lymph nodes and are activated by T cell derived IL-2: a potential new link between adaptive and innate immunity. Blood 101, 3052−3057 (2003). 108. Moffet-King, A. Natural killer cells and pregnancy. Nature Rev. Immunol. 2, 656−663 (2002). 109. Croy, B. A., Chantakru, S., Esadeg, S., Ashkar, A. A. & Wei, Q. Decidual natural killer cells: key regulators of placental development. J. Reprod. Immunol. 57, 151−168 (2002). 110. Cooper, M. A., Fehniger, T. A. & Caligiuri, M. A. The biology of human natural killer-cell subsets. Trends Immunol. 22, 633−640 (2001). 111. Farag, S. S., Fehniger, T. A., Ruggeri, L., Velardi, A. & Caligiuri, M. A. Natural killer cell receptors: new biology and insights into the graft-versus-leukemia effect. Blood 100, 1935−1947 (2002). 112. Caligiuri, M. A. et al. Selective modulation of human natural killer cells in vivo after prolonged infusion of low dose recombinant interleukin 2. J. Clin. Invest. 91, 123−132 (1993).



REVIEWS 113. Fehniger, T. A. et al. Potential mechanisms of human natural killer cell expansion in vivo during low-dose IL-2 therapy. J. Clin. Invest. 106, 117−124 (2000). 114. Clynes, R. A., Towers, T. L., Presta, L. G. & Ravetch, J. V. Inhibitory Fc receptors modulate in vivo cytoxicity against tumor targets. Nature Med. 6, 443−446 (2000). 115. Meropol, N. J. et al. Evaluation of natural killer cell expansion and activation in vivo with daily subcutaneous low-dose interleukin-2 plus periodic intermediate-dose pulsing. Cancer Immunol. Immunother. 46, 318−326 (1998). 116. Ruggeri, L. et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science 295, 2097−2100 (2002). In a mismatched transplantation setting, NK cells prevented relapse of acute myeloid leukaemia. Importantly, these alloreactive NK cells did not cause graft-versus-host disease, mediated graft-versusleukaemia effects and allowed better engraftment of donor cells. 117. Minagawa, M. et al. Enforced expression of Bcl-2 restores the number of NK cells, but does not rescue the impaired development of NKT cells or intraepithelial lymphocytes, in IL-2/IL-15 receptor β-chain-deficient mice. J. Immunol. 169, 4153−4160 (2002).

118. Piguet, P. F., Grau, G., Irle, C. & Vassalli, P. Administration of recombinant interleukin 2 to mice enhances production of hemopoietic and natural killer cells. Eur. J. Immunol. 16, 1257−1261 (1986). 119. Nakajima, H. et al. The BCR/ABL transgene causes abnormal NK cell differentiation and can be found in circulating NK cells of advanced phase chronic myelogenous leukemia patients. J. Immunol. 168, 643−650 (2002). 120. Gilliland, D. G. Molecular genetics of human leukemias: new insights into therapy. Semin. Haematol. 39, 6−11 (2002). 121. Forbes, L. V. et al. An activating mutation in the transmembrane domain of the granulocyte colonystimulating factor receptor in patients with acute myeloid leukemia. Oncogene 21, 5981−5989 (2002). 122. Hongyo, T. et al. Specific c-kit mutations in sinonasal natural killer/T-cell lymphoma in China and Japan. Cancer Res. 60, 2345−2347 (2000). 123. Fehniger, T. A. et al. Fatal leukemia in interleukin 15 transgenic mice follows early expansions in natural killer and memory phenotype CD8+ T cells. J. Exp. Med. 193, 219− 231 (2001). 124. Marks-Konczalik, J. et al. IL-2-induced activation-induced cell death is inhibited in IL-15 transgenic mice. Proc. Natl Acad. Sci. USA 97, 11445−11450 (2000).


Acknowledgements F.C. and J.P.D. are supported by grants from the Pasteur Institute, Inserm, Association pour la Recherche sur le Cancer, Ligue National Contre le Cancer, and the Fondation pour la Recherche Medicale. M.A.C. is supported in part by grant from the United States National Cancer Institute. We thank C. A. Vosshenrich, B. Becknell and T. Ranson for critical reading of the review.

Online links DATABASES The following terms in this article are linked online to: LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink/ 2B4 | CCR7 | CD2 | CD7 | CD16 | CD34 | CD56 | CD94 | CD161 | c-KIT | c-KIT ligand | ETS1 | FLT3 | FLT3L | GATA3 | ID2 | ID3 | IFN-γ | IL-2 | IL-7 | IL-8 | IL-15 | IL-21 | IL-2Rβ | IRF1 | KIR | L-selectin | LTα | LTβR | Ly49 | NEMO | NKG2A | NKR-P1 | PU.1 | SAP | TRAIL FURTHER INFORMATION Michael A. Caligiuri’s lab: http://www.caligiurilab.com James P. Di Santo’s lab: http://www.pasteur.fr/recherche/rar/rar2000/cytotbnk-en.html Access to this interactive links box is free online.

VOLUME 3 | MAY 2003 | 4 2 5
