Natural Killer Cells and Their Relationship to T-Cells: Hypothesis

(CANCER RESEARCH 46, 2651-2658, June 1986) Perspectives in Cancer Research

Natural Killer Cells and Their Relationship to T-Cells: Hypothesis on the Role of T-Cell Receptor Gene Rearrangement on the Course of Adaptive Differentiation Zvi Grossman and Ronald B. Herberman' TelAviv University, TelAviv, JsraeIfZ. G.J,and Pittsburgh Cancerinstitute andDeparrments ofMedicinefR. B. HjandMathematics[Z. G.J, University of Pittsburgh, Pittsburgh, Pennsylvania 15213

Natural killer cells were discovered about 13 years ago during has concerned the lineage of NK cells and their possible rela studies of cell-mediated cytotoxicity against tumor cells (see tionship to the T-cell or macrophage lineages. Three major Ref. 1 for review). Although investigators in this area of re alternatives havebeenextensively considered: (a) NK cells may search expected to find specific cytotoxic activity of tumor be derived from, and closely associated with, the myelomono bearing individuals against autologous tumor cells or against cytic lineage, and most closely related to macrophages; (b) NK allogeneic tumors of similar or the same histological type, cells may be related to the T-cell lineage; (c) NK cells may appreciable cytotoxic activity was also observed with lympho represent a separate lineage derived from bone marrow stem cytes of normal individuals. It soon became clear that such cells and they simply may share some cell surface and other cytotoxic activity was mediated by a particular subpopulation characteristics with cells from other bone marrow-derived lin of effector cells, which have come to be known as NK cells.2 eages. NK cells may be defined as effector cells with spontaneous Most of the experimental evidence for an association of NK cytotoxic reactivity against a variety of target cells, particularly cells with macrophages has been relatively weak and could be tumor cells, tumor cell lines, and a limited variety of normal otherwise accounted for. The expression of 0KM! or Mac I cells (e.g., virus-infected fibroblasts and subpopulations of bone on NK cells was initially taken as indicative of an association marrow cells and thymus cells). The characteristics of NK cells with macrophages (1 1, 12), but recent evidence has indicated and their possible relationship to T-cells or macrophages have that these antigens are on the cell surface receptor for C3bi and been studied extensively (see Ref. 2 for review). NK cells have are not restricted to macrophages(13), being expressedalso on been clearly distinguished from typical macrophages, being a subsetof T-cells which appear unrelated to NK cells. A study nonphagocytic and mostly nonadherent and lacking some cell with mouse bone marrow cells indicated that NK activity de surface markers which are characteristic of most if not all veloped from a population enriched in macrophage precursors monocytes and macrophages. Similarly, NK cells have been (14). However, the population containing the NK precursors distinguished from CTL since their cytotoxic reactivity has not was not a pure population of macrophagesand could have also been found to be restricted by the MHC, with some of the most contained nonmacrophage precursors for NK cells. Along the sensitive targets for NK activity lacking detectable MHC anti samelines, Roder et a!. (15) reported that a population of cells gens, and NK cells have been shown to lack cell surface markers highly enriched in NK activity could undergo an oxidative burst which are characteristic of CTL (e.g., Lyt-2 in the mouse). One and this evidence was taken as definitive evidence for an asso ofthe most consistent features of NK cells, which has permitted ciation of NK cells with macrophages (16). However, subse their purification and detailed comparison with other types of quent studies indicated that the generation of the oxidative effector cells, is their close association with a morphological burst was dependent on the interaction of a small number of subpopulation of cells, the LGL (3). contaminating macrophages with the NK cell population (17). Over the past several years, much of the attention devoted to Another indication for the lack of an association of NK cells NK cells has concerned their possible in vivo relevance, partic with macrophages has come from studies of the expression of ularly with regard to their possible roles in host defense against receptors for colony-stimulating factor 1which havebeenfound to be expressed on most cells in the macrophage lineage. These tumor growth (4). Some evidence has been accumulated to receptors could not be detected on a purified population of rat indicate a possible role of NK cells in immunosurveillance NK cells.3 against certain types of tumors. More extensive and convincing The possible relationship of NK cells with the T-cell lineage evidence has been accumulated to indicate an important role of has been considerably more difficult to evaluate. On the one NK cells in resistance to progressive growth and metastasis of hand, it has been clear that NK cells are different from mature tumors (see Refs. 5 and 6 for recent reviews ofthis information). or classical T-cells. High levels of NK activity have been found There has also been much recent interest in elucidating the in nude athymic or neonatally thymectomized mice and rats mechanism of cytotoxicity by NK cells (see Refs. 7 and 8 for (18, 19). Characteristicmarkers of T-cells, suchas Lyt-2 for reviews). A potent cytolytic protein, termed cytolysin, has been mousecells(20) or T3 or TiOl for humancells(21), havebeen isolated from the granules of LGL (9) and this granule cytolysin undetectable on NK cells. In addition, recent studies have appears to be intimately involved in the mechanism of cytotox indicated that there is a lack of a productive rearrangement of icity by NK cells. In addition, NK cells have been shown to T-cell receptor genes in rat LGL leukemias (22) or in normal release a soluble NK cytotoxic factor which might be an alter LGLof variousspecies(23).@ native but less efficient mechanism of lysis (10). Despite such negative evidence, there have been numerous One of the most controversial areas of research on NK cells

3 T. Barlozzari, C. W. Reynolds, R. Stanley, and R. B. Herberman, manuscript Received 1/21/86; accepted 2/20/86. in preparation.

I To whom requests for reprints should be addressed. 4 H. A. Young, J. R. Ortaldo, R. B. Herberman, and C. W. Reynolds. Analysis 2 The abbreviations used are: NK cells, natural killer cells; CTL, cytotoxic T of T cell receptors in highly purified human and rat large granular lymphocytes lymphocyte(s); MHC, major histocompatibility complex; LGL, large granular (LGL): lack of functional 1.3 kb fl-chain mRNA, J. Immunol., 136: 2701â€”2704, lymphocyte(s); lL-2, interleukin 2. !986. 2651

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 1986 American Association for Cancer Research. ADAPTIVE DIFFERENTIATION OF NK CELLS AND T-CELLS suggestionsfor some relationship of NK cells to T-cells: have been generated with regard to the possible relationship of I . NK cells have been found to express some T-cell-associated NK cells and T-cells, which often seem rather contradictory or markers. The majority of human NK cells express low-affinity divergent, it seems quite important to reassessthe data and to receptors for sheeperythrocytes (24) and react with monoclonal provide a theoretical framework to accommodate most if not antibodies against these receptors (21). A subset of human NK all of the available information. As an early hypothesis on the cells also expressesthe T8 antigen, a characteristic marker for association between NK cells and T-cells, it was suggested, the cytotoxic-suppressor subsetof human T-cells (2 1) and most based on the data showing high levels of NK activity in nude rat NK cells express the analogous OX8 marker (25). or neonatally thymectomized mice, that NK cells might repre 2. NK cells as well as T-cells respond to IL-2, with stimulation sent a stage of prethymic T-cells, which accumulate when the of cytotoxic activity and also proliferation (26â€”28).The IL-2- ability to further differentiate into mature T-cells is blocked by dependent growth of NK cells has been found to depend on the absence of the thymus (39). Another hypothesis, along the expression of IL-2 receptors which are antigenically the same same line, suggested that NK cells represented an early stage as on activated T-cells (28). However, stimulation of cytotox in the development of T-cells and that both effector cell types icity of human NK cells appears to be independent of the Tac shared similar receptors (40). Subsequently, Klein (41) noted IL-2 receptor (28). that NK-like activity develops in alloantigen-stimulated cultures 3. The majority of human T@yleukemias have been found to prior to the development of CTL activity and suggested that have most of the characteristics of LGL, express characteristic NK activity represents a polyclonal response of T-cells during T-cell markers, often have NK activity (29), and have rearrange early stagesofactivation and prior to the development of MHC ments of the gene for the /3-chain of the T-cell receptor (23). restricted reactivity to specific antigens. Several years ago, we 4. NK cells and CTL appear to share similar if not identical also proposed a hypothesis on the origin and specificity of NK lytic mechanisms. The characteristics of the cytolysin isolated cells (42). This hypothesis offered an interpretation of the from the granules of LGL (9) are virtually identical to those similarities and differences between NK cells and CTL and was described for the polyperforin of CTL (30). In contrast, cyto basedon the assumption ofa common T-cell lineage but distinct toxic macrophages do not appear to have an analogous granule specificity repertoires. CTL either recognize MHC antigens or cytolysin.5 recognize other antigens in association with MHC products, 5. There have been recent indications for a sharing of the whereas NK cells interact with a different set of cell surface LGL morphology by T-cells in some situations. Biron et a!. antigens. We argued that since precursor lymphocytes undergo (71) have recently described virus-induced in vivo generation of dynamic selection (43, 44) which is driven by self-antigens that alloreactive CTL with LGL morphology. they recognize, the pattern and consequencesof this selection 6. Highly purified populations of human LGL, devoid of must depend on the nature of the antigens recognized and on detectable T-cells, have been found to produce several T-cell the nature of the target cells presenting these antigens. Several associatedcytokines, including -y-interferon and IL-2 (26, 31). major properties are thus affected by this dynamic selection: (a) 7. Highly purified populations of human LGL, initially de the identity of the targets for effector reactivity in relation to pleted of any detectable T-cells, have been found to develop the reference self-antigens (â€œrestrictionâ€•);(b)the degree of typical T-cell markers upon in vitro culture in the presence of cross-reactivity among effector clones; and (c) the predominant IL-2 (32). At least some of the cultured cells with such T-cell states of maturation of the selected cells and the associated markers have high levels of NK activity. Recently, it has been phenotypic and functional properties of the effector cells. shown that in parallel with the acquisition ofT3 by human T8@ We now propose to reevaluate this hypothesis, based on the LGL in culture, the cells also begin to show rearrangements of considerable accumulation of relevant data over the last 3 years, the fl-chain of the T-cell receptor (23). and to then extend or modify the hypothesis basedon the recent 8. In addition, many situations have been described in which data. In addition to the information discussedabove,our revised cultures of typical T-cells develop NK-like activity (33, 34). hypothesis has been influenced by more detailed information This has also been shown to occur with cloned CTL, which on the characteristics of NK cells and LGL, their cell surface develop NK-like activity simultaneous with continued expres markers, and their responsiveness to various inducers; studies sion of typical MHC-restricted CTL activity (35, 36), or loss of of cultured clones of LGL and T-cells; and in vivo kinetics of CTL activity (37). Some of theseT-cell clones also may express activation and proliferation under different conditions. markers which have been closely associated with NK cells, e.g., NKH1 (35). General Considerations One group has suggestedthat the receptor for NK-like activ ity may be the T-cell receptor, since antibodies to idiotypic The definition ofNK cells is largely related to their functional specificity on the cloned cells reacted with the typical hetero activities, i.e.: (a) their spontaneous or natural expression of dimer associated with the T-cell receptor, and this antibody effector activity and the lack of an apparent need for prior blocked the NK-like cytotoxicity (35, 38). In contrast, however, immunization or proliferation; and (b) their specificity reper another group, working with clones of rat CTL, showed that toire or their pattern of cytotoxicity against certain target cells the NK-like activity appeared to be due to receptors distinct but not others. In addition, as discussedabove, NK cells have from the T-cell receptors for antigens (36). To reconcile this beenclosely associatedwith characteristic morphological mark divergent finding, one might postulate that in the former studies ers. Although such a definition has been very helpful for clas (35, 38), blocking of NK-like recognition by anti-clonotypic sifying these effector cells and distinguishing them from various antibodies was due to steric hindrance of NK receptors in close other effector cells, there are several major problems with proximity to T-cell receptors. precisely defining NK cells, including the observations that Because of the large amount of experimental results which some markers associated with NK cells are not expressed on all of theseeffector cells and further, most of the NK-associated

S j@ Leonhart, C. W. Reynolds, R. H. Wiltrout, and R. B. Herberman, manu markers are not unique to NK cells but also may be expressed script in preparation. on other cell types. For example, as noted above, some LGL 2652

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 1986 American Association for Cancer Research. ADAPTIVE DIFFERENTIATION OF NK CELLS AND T-CELLS are T-cells, including CTL. In addition, a definition linked late that any self-antigen that actively participates in the gen directly to NK activity doesnot include the NK precursors, i.e., eration of tolerance is automatically restricting; the clones cells which do not have effector activity but which may require which it positively selectswill be able to respond only to similar stimulation or differentiation to develop activity. These precur antigens. The ability of a self-molecule to be a major restricting sors of NK cells also may lack some of the characteristic element depends on a number of factors: (a) its abundance; (b) markers of functional NK cells. It would seemmost helpful for its potential to associate with, or be initiated by, foreign anti both theoretical and experimental purposes to include precur gens; and (c) the capacity of the antigen-positive target cell to sors of NK cells within the same considerations as functionally induce significant proliferation in clones with low-affinity re active NK cells. Thus, a somewhat more general and satisfactory ceptors. The MHC-coded molecules certainly have all these definition might beâ€œcellswiththe ability to selectively recognize three characteristics and can account for the restriction associ a given range of(NK-susceptible) target cells, with the potential ated with T-cells. Other cell surface molecules might servethis to lyse these cells, either immediately or upon further matura same function for NK cells. tion or activation.â€•Oncesensitive and specific meansto identify We previously proposed (42) that NK cell precursors arise cells with NK recognition receptors (either by appropriate an from the T-cell lineage and initially differ from T-cell precursors tibodies or by molecular biological approaches) are available, a only in the expression of different sets of receptors. We postu more precise categorization of NK cells and their precursors lated the existenceof a classof molecules that serveasreference might be possible. In the meanwhile, we can define NK cells antigens in the development of NK clones and direct their and some of their precursors only by conjugate assays,which effector activity. We further postulated that the interaction of have the limitation that different effector cell types may bind relatively mature NK precursors with target cells expressing the to the same target cell by different types of receptors. In any appropriate molecules is a weak one in the sense that the relative event, our premise is that the NK recognition structure is the probability of maturation versusself-renewal is inherently large. central issueregarding the definition of NK cells and that other Therefore, except under special conditions (prevailing presum characteristics, e.g., the markers or other aspectsof phenotype, ably in the environment of the bone marrow; see below), self will be associated with such a definition but not completely reference antigens, and especially modified self-antigens, tend coincident. We predict that these other attributes would be to drive NK cell precursors towards terminal differentiation, strongly influenced by the dynamic mechanisms which regulate with little amplification. In contrast, recognition of MHC an the state of differentiation and maturation (42). We postulate tigens mediates a series of events, which includes the involve that the interaction of the recognized structures on target cells, ment of growth factors, which often leads to extensive ampli i.e., antigens, with the receptors on T-cells or NK cells plays a fication in MHC-specific T-cells. The cytolytically active NK central role in regulating their growth and maturation. cell is postulated to be predominantly a terminally differen According to our hypothesis, clones of lymphocytes and NK tiated, nondividing cell. A recent experimental study has pro cells resemble competing species in a Darwinian sense (43). vided support for this hypothesis. Zagury et a!. (45), using a External antigens and self-antigens may stimulate several single-cell micromanipulation technique, isolated cytolytic ef clones which differ in specificity, affinity, or class and, conse fector cells from clones of LGL or of CTL and placed them in quently, in their growth capacity. Each clone can in turn trigger culture with IL-2. Less than 10% of the NK effector cells were negative feedback control, which may or may not affect other able to proliferate. The active CTL appearedto haveevenlower clones. Cross-reactive stimulation and control define a compet proliferative potential, with none of more than 150 cells show itive relationship among these clones and can lead to suppres ing evidence of proliferation. sion of some clones, depending on the particular pattern of Our hypothesis departed from other hypotheses that either cross-reactivity and on the duration and strength of stimulation. regarded the NK lineage as completely unrelated to the T-cell The outcome of this dynamic selection depends on the growth lineage or identified the NK effector cells with an early, im capacities of the competing clones and these are assessedby the mature T-cell (e.g., Ref. 40) or with a form or stage of activation â€œbalanceofgrowthâ€•hypothesis (44). We propose that precur of a classical CTL (41). As noted above, there is experimental sors of T-lymphocytes and NK cells possess a capacity for self support for the concept that active NK cells are well differen renewal and that antigen-activated, but not resting, cells can be tiated with only limited proliferative potential. Along the same further induced to mature or regenerate, and their relative lines, Allavena et a!. (46), using a limiting dilution assay, probability of maturation increases with antigen dose or with showedthat the frequency ofcytotoxic progenitors derived from affinity. LGL wasonlyabout 1 in 100and that withinthis population The further aspect of this hypothesis is that the in vivo the effector cells expressing the Fc receptor, as detected by the induction of terminal differentiation of T-cells and NK cells is B73.1 monoclonal antibody, had a 5-fold lower frequency than antagonistic to clonal expansion. The hypothesis leads, in par the B73.l-negative LGL. Thus, it appears that most of the cell ticular, to the following interpretations of self-tolerance and lines and clones displaying NK-like reactivity do not arise from restriction: (a) self-tolerance would be attributed to the impair mature NK cells but rather from precursors in the same popu ment of growth of clones which recognize abundant self-anti lation. Also, the demonstration that the development of NK gens with high affinities. These clones are instead induced to like activity in someCTL clones may beattributable to different differentiation and consequently they are eliminated through sets of receptors, one for the MHC-restricted recognition of dynamic selection; (b) restriction would result from the prolif CTL-associated antigens and another for the NK cell targets eration of cells recognizing these self-antigens with a certain (36), must be incorporated into any hypothesis about the rela range of low affinities, and these thereby gain prominence in tionship between NK cells and T-cells. Clearly such findings the dynamic selection. Recognition of nonself-antigens by the support our basic assumption of different receptor repertoires. positively selected repertoire of clones must be associated with, On the basis of the recent advances in our understanding of or related to, recognition of self. recognition by T-cells, we now propose that the point of depar One may then ask which self-molecules can serve as restrict ture between T-cells and NK cells during the pathway of differ ing elements in the recognition of foreign antigens. We postu entiation in the T-cell lineage is when the T-cell receptor gene 2653

is rearranged and expressed. We suggest that the expression of observedshifts in functional phenotypes betweenCTL and NK receptors for NK cell targets is probably an earlier event in like activities (33, 34) may not reflect usual in vivo lines of ontogeny. The nonrearranged lymphocytes continue to interact differentiation. Overall, the evidence indicates that a cell is not with their cellular environment, in part through thesereceptors. an independent entity in terms of phenotype or function but This leads to expansion and maturation, associated also with rather reflects both environmental stimuli and intrinsic prop increased receptor expression. The capacity to express NK cell erties. receptors declines in maturing T-cells along with the maturation We believe that this observed plasticity reflects the normal of the MHC-restricted recognition mechanism but can be re potential for phenotypic adaptation, namely the capacity of generated even late in maturation by appropriate and persistent cells to change the pattern of gene expression in response to external stimuli as may be frequently or even regularly observed changes in the environment. Tissue interactions determine the in CTL clones under some environmental circumstances, par developmental fate of cells in the embryo and are required even ticularly the presence of high concentrations of IL-2 (35). A in adult life to maintain the identity of cells (47). A role for schematic representation ofour proposed model ofNK cell and DNA methylation in stabilizing epigenetically induced changes T-cell differentiation is depicted in Fig. 1. in gene expresion has been proposed (48, 49). In this regard, it The usual concept of lineages is based on the premise that, will be of considerable interest to carefully examine and com following some discrete genetic events of determination or pare the patterns of methylation of the DNA in NK cells and commitment, immature precursor cells continue along an in in T-cells. trinsically preprogrammed sequence of divisions and matura The orderly, predictable pattern ofdifferentiation in vivo, like tion steps toward a terminal state. In the lymphocytic lineages, the sequence of developmental events in embryogenesis, owes it has been realized that the number of divisions in some its regularity to a network of causal relations and not just to intermediate states of differentiation is subject to external reg the unfolding of a genetic program (50). Following receptor ulation, but it is still generally believed that the differentiation gene rearrangement and expression, T-cells are subject to dif pathway of normal cells is fixed in advance, under all condi ferent regulatory or inductive signals as compared to nonrear tions. This provides the rationale for many in vitro experiments ranged cells, and the differentiation pathways of these two sets which aim to take a particular subpopulation of cells out of the of cells necessarily diverge from each other. Thus, the devel complex physiological environment and place them into the opment of each set of cells will be associated with those organs simpler culture environment where they could be studied in and tissues which favor proliferation of a particular cell type. detail. The assumption is that the events observed in vitro will This in turn leads to an association of each population with accurately reflect, and provide insights into, the analogous additional microenvironmental factors which can affect the sequenceof events which occur during in vivo differentiation. growth characteristics and the phenotype. This approach ignores the excessive phenotypic plasticity which In our previous communication (42), we emphasized the characterizes cultured cells. For example, Zagury et a!. (45) consequencesof the â€œweakinteractionâ€•conjecture with regard showed that culture of individual cells selected for a particular to the morphological and functional characteristics of the NK marker could result in the acquisition of a different set of subpopulation. The essenceof this conjecture is that NK cell markers. Such shifts have not as yet been observed to occur in precursors are driven more readily to maturation as a result of vivo and thus appear to be rare or nonexistent. Similarly, the interactions with the self-environment and particularly when

NK receptor expression

Fig. 1. Schematic model of NK cell and T-cell differentiation. This depiction of cell development as lineage pathways is meant only as a convenient simplification of our hypothesis since commitment is not absolute but is only a progressive acquisition of a bias for development in a particular direction (see text). The event(s), or process, that leads to the expression of the NK cell receptor is not known. Ti gene (a- and $-chains) rearrangement marks the point of separation of the classical T cell developmental pathway from that of the NK cell. We postulate that the relative levels of expression of restricted Ti receptor versus the NK cell receptor are inversely related: the ratio increases as Ti gene-rearranged precursors mature but can be reversed, or modified, in cultured T-cells. As indicated, LGL leukemias appear to be derived from cells close to the point of bifurcation between NK cells and T-cells, with rat leukemias more closely associated with the former and human leukemias with the latter. The proliferative capacity generally declines with maturation. Peripheral NK cells reach a higher degree of maturation as compared to peripheral T-cells, even prior to encounter with their respective target cells (hence, uflatu@ reactivitf), and thus exhibit low proliferative responses. The LGL morphology may be a common trait of terminally differentiated T-cells (at least of CTL) as well as of NK cells, with the cytoplasmic granules reflecting a well developedlyricapparatus in effectorcells. 2654

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 1986 American Association for Cancer Research. ADAPTIVE DIFFERENTIATION OF NK CELLS AND T-CELLS stimulated by a relevant external challenge. Consequently, we effector function of which is usually not considered to be identified the LGL phenotype and morphology with a close to immunologically specific. The very processofadaptation to the terminal stage of differentiation in the NK-T-cell lineage. We self microenvironment may endow these effector cell popula therefore predicted that phenotypic switches from CTL to the tions with the capacity of surveillance against structural aber LGLphenotypemightfrequentlyoccur,in vitroand in vivo.At rations. that time, such switches had not beenobserved.However, since The specificity of MHC-restricted T-cell recognition is gen then, such shifts in phenotype have been observed in culture erally believed to stem from clonal distribution of receptors on (34, 51) and recently the LGL phenotype has been shown to be the T-cells and the polymorphism of the targets. NK cells and a characteristic which may be common to both NK cells and their targets may differ in both respects, in that NK cells may activated functional CTL in vivo (71). The mature character of bear multiple receptors and target cells may share a small functionally active NK cells would suggest that they would be number of recognizable structures which are broadly distrib short-lived. Indeed, recent data have indicated that LGL have uted. This may account, at least in part, for the observation half-lives of less than 5 days (52). Along with this, one might that NK cells from different sources, including clones, show expect to havean extensively proliferating precursor population similar specificity patterns on panels of target cells (57). These at some earlier stage of differentiation, and indeed a high level patterns, however, are not identical and can be shown to vary of such proliferative activity has been observed (53). In our among subpopulations and clones of NK cells. The concept of original hypothesis, these kinetic characteristics were postu multiple receptors on NK cells, each for recognition of some lated to explain the natural reactivity of NK cells, ascontrasted what different specificities, is supported both by cold target to the requirement for priming and memory in CTL-mediated inhibition experiments (58) and by the loss of reactivity of immune responses. cloned populations against some NK-susceptible targets but not In summary, our hypothesis predicts distinct morphological, against others (59). However, in addition to such observedshifts functional, and distribution characteristics, and also more sub in specificity, there appear to be some consistent differences in tie differences in the pattern of gene expression, for NK cells specificity among clones, supporting the possibility that some and T-cells. Such observed differences (e.g., Ref. 54) may be recognition structures on NK cells are clonally distributed, causally related to the T-cell receptor rearrangement and despite considerable overlap among receptor sets, or cross expression in a portion of the common precursors but not in reactivity of target antigens, as indicated by the lack of fine others. We postulate that the differences develop dynamically specificity. and are subject to mutable environmental factors. In a sense, We have previously argued (42) that the thymus is not an the rearrangement of the T-cell receptor gene can be seenas a appropriate site for selection for NK cells, due to the strong commitment event, in that it endows the cells with a character competition for T-cell-directed differentiation in this organ. istic that substantially affects their future growth and matura Rather, the bone marrow and perhaps the spleen are likely sites tion. However, assignment of the NK cell or LGL to some for differentiation along the NK cell pathway. Recent studies elusive alternative lineage is unnecessary.Finally, we postulate have confirmed a rapid renewal of NK cells in the bone marrow that in culture T-cells and NK cell precursors express variable (e.g.,Ref. 60). As demonstrated earlier for T-cells, it is predicted patterns of gene expression. This makes them an interesting that NK cell precursors may be reeducated, if the microenviron subject for studies ofadaptive differentiation but this also limits ment in which the NK cell repertoire is selected is antigenically our ability to make inferences from such studies about cell modified, with resultant changesin the pattern of specificity of characteristics and in vivo regulatory pathways. the NK cells which develop. The ability of some NK cells to react with hematopoietic stem cells in the bone marrow appears to offer a good system for analysis of this aspect of the hypoth Speculations on Selective Recognition and on Regulation esis. Recent evidence indicates that NK cells may have the A. Discrimination between Self and Nonself by NK Cells. ability to recognize hematopoietic histocompatibility antigens According to the above hypothesis, the existing repertoire of and to discriminate parental bone marrow cells from those of NK cells results from a potentially larger repertoire by a process hybrids (61). In vivo experiments previously demonstated that of competition and selection. Cells are selectedwhich respond neonatal injection of parental bone marrow cells into hybrid by extensive proliferation to stimulation by certain types of mice induced tolerance to subsequenttransplantation of paren autologous cells, during some phase of their development. The tal cells (62). Such observations are quite compatible with our strength of interaction (affinity) should be sufficiently small to present hypothesis. avoid premature terminal differentiation and this constraint B. Recognition of Self and Growth Regulation. Not only ensures self-tolerance. Foreign antigens are recognized as foreign structures but also autologous cell surface molecules â€œmodifiedselfâ€•with different, often higher, affinities to the may appear as modified self. The modification may be mani extent that they resemble the original self-antigens or are as fested as changes in the structure, pattern, or density of mole sociated with them to form modified structures. This may cules on the cell membranes, associated with transformation or explain the observation that NK cells are generally more reac with normal maturational processes.For example, one might tive against transformed cells. This would also provide an envision that NK cells are adapted to tolerate bone marrow explanation for the observation that NK-resistant normal cells hematopoietic cells over some range of maturational states but become susceptible to NK activity after being infected with retain the ability to interact with and inhibit the growth of various viruses or other microorganisms (e.g., Refs. 55 and 56). autologous hematopoietic cells which lie outside this range. Further kinetic and molecular analyses are required to prove This might explain the observations of reactivity of NK cells that such changes are due to more efficient binding and to against some autologous or syngeneic bone marrow cells (63, identify the nature of the modified molecules. Self-nonself 64) and also the NK cell-associated rapid in vivo elimination of discrimination is the most basic specificity characteristic of NK some autologous bone marrow cells (65). It will be of consid cells, an attribute shared with classical T-cells and B-cells and erable interest to further explore the ability to selectively edu probably also by other types of cells such as macrophages, the cate NK cells and to modify their pattern ofspecificity, utilizing 2655

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 1986 American Association for Cancer Research. ADAPTIVE DIFFERENTIATION OF NK CELLS AND T-CELLS more highly selected target cell populations and highy selected Summary and Conclusions populations of NK cells for interaction with these targets. The NK resistance of some tumor cells might be interpreted We propose that the differentiation of NK cells and the as a form of tolerance, resulting from coadaptation of NK cell differentiation of T-cells are intimately interrelated, although precursors and tumor cells. This is particularly conceivable for mature effector cells of each type usually can be distinguished leukemias of bone marrow origin. In contrast, NK cells might from each other. The divergence in their characteristics may be interact more effectively with phenotypically modified tumors initiated upon rearrangement of the genesfor the T-cell recep originating in organs and tissues that are not associated with tor, with a subsequentinverse relationship between the expres NK cell development. sion of T-cell receptors and NK cell receptors. However, an Another potentially important mechanism for regulation of essential element of our hypothesis is that the differentiation differentiation of NK cells may be feedback growth control by of these cells is partially adaptive rather than rigidly prepro the progeny of proliferating cells. For example, interferon, grammed. This concept is considerably more compatible with which can be secretedby mature NK cells as well as by T-cells the phenotypic plasticity which has been exhibited by cultured (66), may regulate both the expression ofeffector reactivity and cells in general and by T-cells and LGL in particular. We the growth of NK cell precursors. Evidence gathered in the last suggest that the nature of the self environment has a major few years indicates that in vivo administration of interferon or influence on the direction of development of precursor cells, a variety of other immunomodulators results in a substantial both by controlling the ratio between the rates of proliferation increase in the number of morphologically identifiable LGL in and differentiation at each stage of maturation and by inducing the spleen, liver, and other organs, and this is accompanied by quantitative or qualitative changesin the pattern ofgene expres an increasein the proportion ofcells undergoing DNA synthesis sion. As maturation proceeds,the degreeof plasticity probably (67â€”69).Theseobservations could beattributed to the induction decreases,possibly due to inheritable epigenetic changesin the of terminal differentiation of mature NK cells from their pre genome. Our hypothesis accommodates most if not all of the cursors, which in turn accelerates the turnover in the tissues available experimental data on the phenotypic, genetic, and and triggers, through feedback interactions, a proliferative re functional interrelationships between NK cells and T-cells. In sponsein the bone marrow or other precursor compartments. particular, it accounts for the extensive and controversial data Other factors and cells are very likely to be involved in the on cultured cell lines with varying degrees of similarity to T regulation of differentiation of NK cells. In analogy to helper cells and to NK cells. In addition, our model emphasizes the T (T,j-cells and suppressor T (T,j-cells as populations of T inherent limitations in utilizing such data from cell lines as the cells specializing in regulation of T-cell functions, one might basis for drawing conclusions on the properties of cells devel predict the existence of natural helper (Nh) cells and natural oping under physiological conditions. Most importantly, our suppressor (N,) cells within the NK cell population. These cells hypothesis leads to a series of experimentally testable predic would share the same set of receptors with NK active cells and tions, which should provide considerably greater insight into might have the associatedphenotypic and functional character the ontogeny of NK cells and their relationship to the T-cell istics but also the capability to produce lymphokines. Indeed, lineage. LGL with no apparent cytotoxicity but secreting growth and! or differentiation factors (e.g., IL-2, interferon) have beeniden References tified (70). Our proposed model for NK cell and T-cell differentiation 1. Herberman, R. B. (ed.). Natural Cell-mediated Immunity against Tumors. (Fig. I) should generate a series of testable predictions. For New York: Academic Press, 1980. 2. Ortaldo, J. R., and Herberman, R. B. Heterogeneity of natural killer cells. example, an event that leads to expression of the NK cell Annu. Rev. Immunol., 2: 359â€”394,1984. receptor during an early phase of differentiation of NK cells 3. Timonen, T., Reynolds, C. W., and Herberman, R. B. Isolation of human and T-cells might be identified in the precursors of both. Our and rat natural killer cells. J. Immunol. Methods, 51: 269â€”280,1982. 4. Herberman, R. B. (ed.). NK Cells and Other Natural Effector Cells. New hypothesis predicts that NK cell precursors should have the York: Academic Press, 1982. capability under some circumstances to be induced to develop 5. Hanna, N. Review of evidence for the role of NK cells in immunoprophylaxis of metastasis. in: R. B. Herberman, R. H. Wiltrout, and E. Gorelik (eds.), along the classical T-cell pathway. Certainly, hints ofthis switch Immune Responses to Metastases. Boca Raton, FL: CRC Press, in press, have come from studies of cultured LGL, in which some cells 1986. appear to undergo rearrangements of the gene for the f3-chain 6. Wiltrout, R. H., Talmadge, J. E., and Herberman, R. B. Role of NK cells in treatment of metastases with biological response modifiers. in. R. B. Herber of the T-cell receptor and develop T-cell-associated markers man, R. H. Wiltrout, and E. Gorelik (eds.), Immune Responses to Metas (23, 32, 45). Such changes in phenotype need to be defined tases. Boca Raton, FL: CRC Press, in press, 1986. 7. Herberman, R. B., and Callewaert, D. M. (eds.) Mechanisms of Cytotoxicity more precisely and, in particular, it will be important to deter by NK Cells. Orlando, FL: Academic Press, 1985. mine whether cells with NK receptors but without T-cell mark 8. Herberman, R. B., Reynolds, C. W., and Ortaldo, J. Mechanism of cytotox ers can be induced to develop the phenotype and, of most icity by natural killer (NK) cells. Annu. Rev. Immunol., 4: 651â€”680,1986. 9. Henkart, P. A. Mechanism of NK-cell mediated cytotoxicity. in: R. B. interest, the characteristic functions of T-cells. Herberman (ed.), Basic and Clinical Tumor Immunology, Vol. 2, pp. 123â€” Another prediction from our model is that cells undergoing 130. The Hague: Martinus Nijhoff, in press, 1986. rearrangement of the gene for the T-cell receptor would be 10. Wright, S. C., Wilbur, S. M., and Bonavida, B. Functional and biochemical characterization of natural killer cytotoxic factors and their role in the NK found to show a progressive decrease in expression of NK lytic mechanism. in: R. B. Herberman and D. M. Callewaert (eds.), Mocha receptors. Clones ofcultured LGL might provide a useful model nisms of Cytotoxicity by NK Cells, pp. 323â€”334,Orlando, FL: Academic Press, 1985. to examine this prediction. Clones could be selected on the I 1. Breard, J., Reinherz, E. L., O'Brien, C., and Schlossman, S. F. Delineation basis of whether they had undergone rearrangement of the T of an elTector population responsible for natural killing and antibody-dc cell receptor genes and then followed to determine the time pendent cellular cytotoxicity in man. Clin. Immunol. Immunopathol., 18: 145â€”150,1981. course of retention of NK recognition structures. Our hypoth 12. Ault,K.A.,and Springer,T. A.Cross-reactionofa rat-anti-mousephagocyte esis predicts that the clones with rearrangement would show a specific monoclonal antibody (anti-Mac-!) with human monocytes and nat ural killer cells. J. Immunol., 126: 359â€”364,1981. greater decrease in NK receptors than the clones without rear 13. Sanchez-Madrid, F., Nagy, J. A., Robbins, E., Simon, P., and Springer, T. ranged T-cell receptor genes. A. A human leukocyte differentiation antigen family with distinct a-subunits 2656

and a common $-subunit. J. Exp. Med., 158: 1785â€”1803,1983. 40. Kaplan, J., and Callewaert, D. M. Are natural killer cells germ line V-germ 14. Lohmann-Matthes, M.-L., and Domzig, W. Natural cytotoxicity of macro encoded prothymocytes specific for self and nonself histocompatibility anti phage precursor cells and of mature macrophages. in: R. B. Herberman (ed.), gens? In: R. B. Herberman (ed.), Natural Cell-mediated Immunity against Natural cell-mediated Immunity against Tumors, pp. 117â€”129.New York: Tumors, pp. 893â€”907.New York: Academic Press, 1980. Academic Press, 1980. 41. Klein, E. Comment on the nomenclature of lymphocyte mediated cytotoxic 15. Roder, J. C., Helfand, S. L., Werkmeister, J., McGarry, R., Beaumont, T. effects. in: R. B. Herberman (ed.), NK Cells and Other Natural Effector J., and Duwe, A. Oxygen intermediates are triggered early in the cytolytic Cells, pp. 239â€”242.NewYork: Academic Press, 1982. pathway of human NK cells. Nature (Lond.), 298: 569â€”572,1982. 42. Grossman, Z., and Herberman, R. B. Hypothesis on the development of 16. Babior, B. M., and Parkinson, D. W. The NK cells: a phagocyte in lympho natural killer cells and their relationship to T cells. in: R. B. Herberman cyte's clothing? Nature (Lond.), 298: 51 1, 1982. (ed.), NK Cells and Other Natural Effector Cells, pp. 229â€”238,New York: 17. Pohajdak, B., Gomez, J. L., Wilkins, J. A., and Greenberg, A. H. Tumor Academic Press, 1982. associated NK cells trigger monocyte oxidative metabolism. J. Immunol., 43. Grossman, Z. Recognition ofselfand regulation of specificity at the level of 133: 2430â€”2436,1984. cell populations. Immunol. Rev., 79: 119â€”138,1984. 18. Herberman, R. B., Nunn, M. E., and Lavrin, D. H. Natural cytotoxic 44. Grossman, Z. Recognition of self, balance of growth and competition: hori reactivity of mouse lymphoid cells against syngeneic and allogeneic tumors. zontal networks regulate immune responsiveness. Eur. J. Immunol., 12: 747â€” I. Distribution ofreactivity and specificity. mt. J. Cancer, 16: 216â€”229,1975. 756, 1982. 19. Reynolds, C. W., Timonen, T. T., Holden, H. T., Hansen, C. T., and 45. Zagury, D., Morgan, D. A., Fouchard, M., Bernard, J., Feldman, M., and Herberman, R. B. Natural killer (NK) cell activity in the rat. Analysis of Herberman, R. B. Heterogeneity of human natural killer (NK) cells. J. Natl. effector cell morphology and effects of interferon in NK cell function in the Cancer Inst., in press, 1986. athymic (nude) rat. Eur. J. Immunol., 12: 577â€”582,1982. 46. Allavena, P., Klein, R., and Ortaldo, J. R. Characterization of human large 20. Herberman, R. B., Nunn, M. E., Holden, H. T., and Lavrin, D. H. Natural granular lymphocyte subpopulations: comparison of the phenotype of NK cytotoxic reactivity of mouse lymphoid cells against syngeneic and allogeneic cells and of interleukin 2-dependent progenitors of cytolytic effector cells. tumors. II. Characterization of effector cells. mt. i. Cancer, 16: 230â€”239, Nat. Immun. Cell Growth Regul., 4: 7â€”20,1985. 1975. 47. Rubin, H. Cancer as a dynamic developmental disorder. Cancer Res., 45: 21. Ortaldo, J. R., Sharrow, S. 0., Timonen, T., and Herberman, R. B. Analysis 2935â€”2942,1985. of surface antigens on highly purified human NK cells by flow cytometry 48. Razin, A., and Cedar, H. DNA methylation in eukaryotic cells. Int. Rev. with monoclonal antibodies. J. Immunol., 127: 2401â€”2406,1982. Cytol., 92: 159â€”185,1984. 22. Reynolds, C. W., Bonyhadi, M., Herberman, R. B., Young, H. A., and 49. Cedar, H. DNA methylation and gene expression. in: A. Razin, H. Cedar, Hendrick, S. M. Lack of gene rearrangement and mRNA expression of the and D. Riggs (eds.), DNA Methylation: Biochemistry and Biological Signif beta chain of the T cell receptor in spontaneous rat large granular lymphocyte icance, pp. 147â€”163.Berlin:Springer-Verlag, 1985. leukemia lines. J. Exp. Med., 161: 1249â€”1254,1985. 50. Stent, G. S. Thinking in one dimension: the impact of molecular biology of 23. Rambaldi, A., Pelicci, P.-G., Allavena, P., Knowles, D. M., Rossini, S., development. Cell, 42: 1â€”2,1985. Bassan, R., Barbvi, T., Dalla-Favera, R., and Mantovani, A. T-cell receptor 51. Shortman, K., Wilson, A., Scollay, R., and Chen, W.-F. Development of @ chain gene rearrangements in lymphoproliferative disorders of large gran large granular lymphocytes with anomalous, nonspecific cytotoxicity in ular lymphocytes/natural killer cells. J. Exp. Med., in press, 1985. clones derived from Ly-2@T cells. Proc. Nat!. Acad. Sci. USA, 80: 2728â€” 24. West, W. H., Cannon, G. B., Kay, H. D., Bonnard, G. D., and Herberman, 2732, 1983. R. B. Natural cytotoxic reactivity of human lymphocytes against a myeloid 52. Reynolds, C. W., Ward, J. M., Wiltrout, R. H., and Herberman, R. B. Tissue cell line: characterization ofeffector cells. J. Immunol., 118: 355â€”361,1977. distribution and in vivo localization ofrat LGL. in: T. Hoshino, H. S. Koren, 25. Reynolds, C. W., Timonen, T., and Herberman, R. B. Natural killer (NK) and A. Uchida (eds.), Natural Killer Activity and Its Regulation, pp. 9â€”15. cell activity in the rat. I. Isolation and characterization of the effector cells. Tokyo: Excerpts Medica, 1984. J. Immunol., 127: 282â€”287,1981. 53. Reynolds, C. W. Are natural killer cells just T cells in disguise? J. Leuk. 26. Domzig, W., Stadler, B. M., and Herberman, R. B. Interleukin-2 dependence Biol., 38: 161â€”162,1985. of human natural killer (NK) cell activity. J. Immunol., 130: 2970â€”2973, 54. Reynolds, C. W., and Ward, J. M. Tissue and organ distribution ofNK cells. 1983. in:E.LotzovaandR.B.Herberman(eds.),ImmunobiologyofNaturalKiller 27. Kuribayashi, K., Gillis, S., Kern, D. E., and Henney, C. S. Murine NK cell Cells. Boca Raton, FL: CRC Press, in press, 1986. cultures: effects of interleukin-2 and interferon on cell growth and cytotoxic 55. Rager-Zisman, B., Grose, C., and Bloom, B. R. Mechanisms of selective reactivity. J. Immunol., 126: 2321â€”2327,1981. nonspecific cell-mediated cytotoxicity of virus-infected cells. Nature (Lond.), 28. Ortaldo, J. R., Mason, A. 1., Gerard, J. P., Henderson, L. E., Farrar, W., 260: 369â€”370,1976. Hopkins, R. F., III, Herberman, R. B., and Rabin, H. Effects of natural and 56. Fitzgerald, P. A., Lopez, C., and Siegal, F. P. Role of interferon in natural recombinant IL-2 on regulation of IFN gamma production and natural killer kill of herpesvirus infected fibroblasts. in: R. B. Herberman (ed.), NK Cells activity: lack of involvement of the Tac antigen for these immunoregulatory and Other Natural Effector Cells, pp. 387â€”393.New York: Academic Press, effects. J. lmmunol., 133: 779â€”783,1984. 1982. 29. Reynolds, C. W., and Foon, K. A. T-y-lymphoproliferative disorders in man 57. Ortaldo, J. R., and Herberman, R. B. Specificity of natural killer cells. in: B. and experimental animals: a review of the clinical, cellular, and functional Serrou, C. Rosenfeld, and R. B. Herberman (eds.), Human Cancer Immu characteristics. Blood, 64: 1146â€”1158,1984. nology, Vol. 4, pp. 17â€”36.Amsterdam:Elsevier/North-Holland Biomedical 30. Podack, E. R., and Konigsberg, P. J. Cytolytic I cell granules. Isolation, Press, 1982. structural, biochemical, and functional characterization. J. Exp. Med., 160: 58. Kedar, E., Ikejiri, B. L, Sredni, B., Bonavida, B., and Herberman, R. B. 695â€”702,1984. Propagation of mouse cytotoxic clones with characteristics of natural killer 31. Kasahara, T., Djeu, J. Y., Dougherty, S. F., and Oppenheim, J. J. Capacity (NK) cells. Cell. Immunol., 69: 305â€”329,1982. of human large granular lymphocytes (LGL) to produce multiple lympho 59. Allavena, P., and Ortaldo, J. R. Characteristics of human NK clones: target kines: interleukin 2, interferon and colony stimulating factor. J. Immunol., specificity and phenotype. J. Immunol., 132: 2363â€”2369,1984. 131: 2379â€”2385,1983. 60. Miller, S. C. Production and renewal of murine natural killer cells in the 32. Allavena, P., and Ortaldo, J. R. Characteristics of human NK clones: target spleen and bone marrow. J. Immunol., 129: 2282â€”2286,1982. specificity and phenotype. In: T. Hoshino, H. S. Koren, and A. Uchida(eds.), 61. Bordignon, C., Daley, J. P., and Nakamura, I. Hematopoietic histoincom Natural Killer Activity and Its Regulation, pp. 64â€”69.Tokyo: Excerpta patibility reactions by NK cells in vitro: model for genetic resistance to Medica, 1984. marrow grafts. Science (Wash. DC), 230: 1398â€”1401,1985. 33. Brooks, C. G. NK cells and T cell differentiation. In. R. B. Herberman and 62. Cudkowicz, G., and Hochman, P. S. Do natural killer cells engage in D. M. Callewaert (eds.), Mechanisms of Cytotoxicity by NK Cells, pp. 645â€” regulated reactions against self to ensure homeostasis? Immunol. Rev., 44: 654. Orlando, FL: Academic Press, 1985. 13â€”28,1973. 34. Brooks, C. G. Reversible induction of natural killer cell activity in clonal 63. Nunn, M. E., Herberman, R. B., and Holden, H. T. Natural cell-mediated murine cytotoxic T lymphocytes. Nature (Lond.), 305: 155â€”157,1983. cytotoxicity in mice against non-lymphoid cells and some normal cells. Int. 35. Hercend, T., Meuer, S., Brennan, A., Edson, M. A., Acuto, 0., Reinherz, E., J. Cancer, 20: 381â€”387,1977. Schlossman, S. F., and Ritz, J. Identification of a clonally restricted 90 KD 64. Hansson, M., KÃ¤rre,K., Kiessling, R., Roder, J., Andersson, B., and HÃ£yry, heterodimer on two cloned natural killer cell lines: its role in cytotoxic P. Natural NK-cell targets in the mouse thymus: characteristics of the effector function. J. Exp. Med., 158: 1547â€”1560,1983. sensitive cell population. J. Immunol., 123: 765â€”771,1979. 36. Binz, H., Fenner, M., Frei, D., and Wigzell, H. Two independent receptors 65. Riccardi, C., Santoni, A., Barlozzari, T., and Herberman, R. B. in vivo allow selective target lysis by T cell clones. J. Exp. Med., 157: 1252â€”1260, reactivity of mouse natural killer (NK) cells against normal bone marrow 1983. cells. Cell. Immunol., 60: 136â€”143,1981. 37. Hengartner, H., Acha-Orbea, H., Lang, R., Stitz, L., Rosenthal, K. L., 66. Djeu, J. Y., Timonen, T., and Herberman, R. B. Production of interferon by Groscurth, P., and Keller, R. Permanently growing murine cell clones with human natural killer cells in response to mitogens, viruses and bacteria. in..@ NK like activities. In: R. B. Herberman (ed.), NK Cells and Other Natural R. B. Herberman (ed.), NK Cells and Other Natural Effector Cells, pp. 669â€” Effector Cells, pp. 893â€”901.New York: Academic Press, 1982. 674. New York: Academic Press, 1982. 38. Hercend, T., Meuer, S., Brennan, A., Edson, M. A., Acuto, 0., Reinherz, E. 67. Biron, C. A., and Welsh, R. M. Blastogenesis of natural killer cells during L., Schlossman, S. F., and Ritz, J. Natural killer-like function ofactivated T viral infection in vivo. J. Immunol., 129: 2788â€”2795,1982. lymphocytes: differential blocking effects of monoclonal antibodies specific 68. Santoni, A., Piccoli, M., Ortaldo, J. R., Mason, L, Wiltrout, R. H., and for a 90-kD clonotypic structure. Cell. Immunol., 86: 381â€”392,1984. Herberman, R. B. Changes in number and density oflarge granular lympho 39. Herberman, R. B., and Holden, H. T. Natural cell-mediated immunity. Adv. cytes upon in vivoaugmentation ofmouse natural killer activity. J. Immunol., Cancer Res., 27: 305â€”377,1978. 134:2799â€”2810,1985. 2657

69. Wiltrout, R. H., Mathieson, B. J., Talmadge, J. E., Reynolds, C. W., Zhang, Herberman, R. B., and Ortaldo, J. R. Production of multiple cytokines by S. R., Herberman, R. B., and Ortaldo, J. R. Augmentation of organ-associ- clones ofhuman large granular lymphocytes. Cancer Immunol. Immunother., ated natural killer activity by biological response modifiers: isolation and 19: 121â€”126,1985. characterizationoflarge granular lymphocytesfrom the liver.J. Exp. Med., 71. Biron,C. A., Natuk, R. J., and Welsh, R. M. Generation oflarge granular T 160: 1431â€”1449,1984. lymphocytes in vivo during viral infection. J. Immunol., 136: 2280â€”2286, 70. Allavena, P., Scala, G., Djeu, J. Y., Procopio, A. D., Oppenheim, J. J., 1986.

2658

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 1986 American Association for Cancer Research. Natural Killer Cells and Their Relationship to T-Cells: Hypothesis on the Role of T-Cell Receptor Gene Rearrangement on the Course of Adaptive Differentiation

Zvi Grossman and Ronald B. Herberman

Cancer Res 1986;46:2651-2658.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/46/6/2651

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/46/6/2651. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.