Mono-allelic Ly49 NK cell receptor expression

seminars in I M M U N OL OG Y, Vol 11, 1999: pp. 349]355 Article No. smim.1999.0191, available online at http:rrwww.idealibrary.com on

Mono-allelic Ly49 NK cell receptor expression Werner Held U , Beatrice Kunz, Vasilios Ioannidis and Bente Lowin-Kropf ´

Each cell is equipped with two copies (alleles) of each autosomal gene. While the vast majority use both alleles, occasional genes are expressed from a single allele. The reason for mono-allelic expression is not always evident and can serve distinct purposes. First, it may facilitate the tight control over the dosage of certain gene products such as some growth factors and their receptors or X-linked genes. Second, the differential usage of the two parental alleles may reflect the mechanisms that ensure mono-specificity, e.g. olfactory receptors, T and B cell receptors. The context of allele-specific expression of the murine Ly49 natural killer (NK) cell receptor genes suggests that their allele-specific expression reflects a process that generates clonal variability.

ficient bone marrow grafts in vivo. In addition, NK cells mediate the resistance to certain viral infections and play a role during the immune response to intracellular bacteria. Recent results also indicate a regulatory role in an experimental autoimmune disease.1 Since NK cells are able to respond quickly and without the need for prior antigen exposure, they are thought to provide a first line of immunological protection.

Regulation of NK cell function The rejection of bone marrow stem cells and the lysis of certain tumour cells indicated that MHC class I is a key regulator of NK cell function. Since then it has become clear that NK cells can kill target cells that either lack some or all class I molecules on their surface or that express non-self MHC class I molecules. This mode of reactivity has been termed ‘missing self’ recognition.2 Similar to T cells, however, NK cells are also able to mediate positive recognition of allogenic MHC class I molecules.3 ] 5. Considerable progress has been made in understanding the molecular basis for missing self recognition. It is now well accepted that NK cells can be activated to kill target cells via multiple different receptor ligand interactions. However, NK cell activation is prevented when inhibitory MHC class I specific receptors encounter their ligands on target cells. Therefore, NK cell function is regulated by a balance between activating and inhibitory signals and NK cell mediated lysis of target cells is based on the failure to receive an inhibitory signal via MHC receptors.6 Whether this type of regulation is also critical for NK cell mediated innate immunity against pathogens is currently not known.

Key words: NK cell r Ly49 r allele r MHC class I r stochastic Q1999 Academic Press

Introduction THE ROLE OF MONO-ALLELIC expression of Ly49 NK cell receptor genes will be addressed in the context of the function and specificity of these receptors. We will thus first review certain aspects of NK cell biology and the role of Ly49 receptors therein. Despite their relatively low abundance compared to other lymphoid cells, NK cells play important roles in a number of situations. Perhaps most importantly, they are able to lyse some syngeneic, allogenic or xenogenic tumour cells in vitro and reject small doses of grafted tumour cells in vivo, compatible with a role of these cells in tumour surveillance. Furthermore, NK cells mediate the rejection of allogenic or major histocompatibility complex ŽMHC. class I-deU

From the Ludwig Institute for Cancer Research, Lausanne Branch, University of Lausanne, 155, Ch. des Boveresses, 1066 Epalinges, Switzerland. E-mail: [email protected] Q1999 Academic Press 1044-5323r 99 r 050349q 07 $30.00r 0

NK cell receptors Insights into the molecular basis for missing-self recognition resulted from the identification of MHC 349

W. Held et al

class I specific inhibitory NK cell receptors, such as mouse Ly49A.7 It turned out that Ly49a defines a gene family that currently comprises fourteen highly related members Ž Ly49a-n..8 ] 10 The Ly49 genes are tightly linked in a ; 600-kb subregion of the so-called natural killer gene complex ŽNKC., a region on the distal part of mouse chromosome 6 that encodes a significant number of NK cell surface receptors8,11,12 ŽFigure 1.. While Ly49s are homodimeric, type II, lectin-like cell surface receptors, human NK cells utilise a set of structurally distinct receptors for the recognition of classical MHC class I molecules.13 Human receptors, which are termed killer inhibitory receptors ŽKIR., are predominantly monomeric and belong to the immunoglobulin superfamily. No functional human Ly49 receptors have yet been described, even though a NKC has been localised to chromosome 12.12 NK cell receptors recognise specific alleles of class I heavy chains that form a tri-molecular complex with b 2m and peptide.14 However, there is no stringent requirement for specific class I complexes as the majority or all of the bound peptides permit recognition by KIR or Ly49 family receptors, respectively.14,15 The Ly49A receptor binds to D d or D k Žbut, e.g. not to D b or K d . class I molecules. Consistent with missing self recognition, NK cells expressing the Ly49A receptor are inhibited when encountering the rele-

vant ligands on target cells. Other Ly49 receptors can have distinct yet often overlapping MHC specificities ŽFigure 1. Žfor review see ref 16.. Irrespective of the structural differences, transduction of an inhibitory signal by either the Ly49 or the KIR family receptors, is based on the so-called immunoreceptor tyrosine-based motif ŽITIM . ŽVrIxYxxL. present in the cytoplasmic tails. Phosphorylation of the tyrosine residue in the ITIM results in the recruitment of the tyrosine phosphatases SHP-1 and SHP-2. The former is clearly involved in transducing the inhibitory signal that aborts NK cell cytotoxicity.17,18 Intriguingly, some MHC class I-specific receptors lack a consensus ITIM. They are associated non-covalently with a small homodimeric molecule, DAP-12, that mediates an activating signal upon receptor cross-linking.19 Indeed, such receptors enable direct non-self MHC reactivity by NK cells 20 and thus extend the modes of NK cell reactivity beyond the one predicted by the missing-self hypothesis. The physiological role of activating MHC receptors and their regulation vis-a-vis the inhibitory counterparts ` are currently not well understood.

NK cell receptor repertoire Expression of Ly49 receptors is observed on NK cells,

Figure 1. Schematic representation of the Ly49 gene cluster within the natural killer gene complex ŽNKC. of C57BLr6 ŽB6. mice. Indicated are all described Ly49 genes and their putative relative location. The genes can be grouped into Ly49a-like and Ly49c like genes based on sequence similarity. Ly49b is more distantly related. Both major groups include members that do or do not contain an immunoreceptor tyrosine based inhibitory motif ŽITIM., that discriminates inhibitory and activating Ly49 receptors, respectively. The percentage of NK cells expressing a particular receptor indicated where known. Some receptor specificities are indicated.

350

Allele-specific Ly49 gene expression

NK1.1-positive and -negative T cells and a few T cell lymphomas. In sharp contrast to antigen receptors expressed by T or B cells, a single NK cell can express multiple MHC receptors. In fact, each NK cell displays a selection of the available Ly49 receptors. Self-MHC-reactive and non-reactive receptors are often co-expressed, suggesting that receptor acquisition is not stringently directed by the hosts MHC class I Žrandom expression.. Detailed expression analyses revealed that co-expression of Ly49 receptors in single NK cells is observed at frequencies that are expected when receptors are expressed independently and co-expression coincides Žindependent expression..21 The product of the frequencies of individual receptors predicts in fact fairly accurately the frequency of co-expressors Žproducts rule..16 Significantly, even though mouse and human NK cell receptors for classical class I molecules are structurally distinct their expression seems to be governed by similar rules.22 Assuming that the known Ly49 expression patterns, i.e. independent and random expression, extends to all 14 known Ly49 genes, the NK cell compartment may contain up to 2 14 Žor ) 15,000. different NK cells solely based on Ly49 expression. It is likely, however, that the number of available NK cell clones is considerably smaller, as MHC class I-dependent cellular events are thought to adapt the pool of immature NK cells to their MHC environment. With regard to Ly49, significant MHC-dependent effects on single receptor usage as well as positive and negative associations between distinct Ly49 receptors that deviate from the products rule are observed in normal mice.16,21 These effects may in part reflect processes which ensure that NK cells that fail to express a self-MHC specific receptor do not develop andror functionally mature. Such a process seems necessary to avoid the emergence of autoaggressive NK cell clones. Yet, the MHC-dependent Ly49 repertoire alterations in normal mice are subtle, perhaps suggesting additional mechanisms that ensure self tolerance.23 It is also possible, however, that adaptive effects on the Ly49 repertoire are somewhat obscured by the large number of possible clones. The independent and random expression together with the ligand selectivity of MHC receptors provides the NK cell compartment with considerable clonal heterogeneity. This may enable NK cells to also react to relatively subtle perturbations in a target cell’s MHC make-up, such as the absence of a single class I allele. Such alterations may result from mutation,

transformation or infection and thus indicate the presence of diseased cells.

Ly49 receptor acquisition NK cells are bone marrow derived and the marrow microenvironment plays an important role in their generation.24,25 The development of NK cells can be initiated in vitro from multipotent bone marrow precursor cells using cytokines Žinterleukin-6, -7, -15 and flt3 ligand.. However, the expression of Ly49 receptors can not be induced.26 It is thus possible that specialised factors produced by andror direct interactions with the bone marrow microenvironment are obligatory to initiate Ly49 receptor expression.24 Fetal NK cells do not express significant levels of Ly49 receptors.27 Ly49 expression is observed on occasional splenic NK cells in neonatal mice. Thereafter the proportion of Ly49q NK cells steadily increases to reach a plateau at approximately 6 weeks of age.28 The appearance of Ly49 defined NK cell subsets in an adoptive transfer model suggested that receptor acquisition occurs in a non-random order.28 Ly49 receptor acquisition is thus most likely a developmentally regulated process that may be operative during a particular stage of NK cell development and require specialised signals from the marrow environment.

Mono-allelic Ly49 gene expression Curiously, NK cells expressing the Ly49a gene do so predominantly from a single allele, whereby individual NK cells can use either parental allele. Two models for mono-allelic Ly49a expression were initially considered:29 A regulated process in which expression of a single Ly49a allele would prevent via a feedback mechanism the activation of the second Ly49a allele Žregulated model.. Alternatively, individual alleles of the Ly49a gene may be expressed independently and randomly Žstochastic model.. Two sets of experiments were used to discriminate between these possibilities. The forced expression of a Ly49a transgene in all NK cells did not prevent the expression of the endogenous Ly49a gene. Even the presence of the strong Ly49A ligand H-2D d did not abrogate endogenous Ly49a expression.30 Mono-allelic Ly49a expression is thus not imposed by a stringent feedback mechanism. Indeed, the analysis 351

W. Held et al

patterns of Ly49A alleles were evident during a 7 day culture period. Therefore, mono-allelic Ly49A expression is maintained during at least four to five cell divisions.31 Similarly, once acquired, several Ly49 receptors were stably expressed in vivo.28 These results suggest that once the putative acquisition process is terminated, Ly49 expression in mature NK cells may be regulated by an epigenetic mechanismŽs.. For instance methylation of the cytosine residue in CpG dinucleotides is known to negatively influence the expression of various tissue specific genes. Preliminary results indicate in fact an inverse correlation between the presence of methylated CpGs in the Ly49a promoter and Ly49A expression Žunpublished observation.. Therefore, the stability of mono-allelic Ly49 expression may be maintained via the inheritable patterns of CpG methylation.

Figure 2. Parental origin of Ly49 mRNA in short-term NK cell clones that co-express three Ly49 receptors. Schematic representation of the Ly49 gene cluster in ŽB6 = BALB.B.F1 cells ŽB6 chromosome: upper line, BALB: lower line.. The order of Ly49 genes represents the actual relative location in the NKC, additional Ly49 genes are interspersed between Ly49g and Ly49c. Of 58 ŽB6 = BALB.B.F1-derived short term NK cell clones that were selected based on Ly49A expression 15 clones Ž26%. co-expressed Ly49a, Ly49c and Ly49g mRNA.32 The parental origin of the different Ly49 mRNAs is shown in the individual clones.

Molecular models for clonally diverse Ly49 receptor expression Obviously, the elucidation of the molecular mechanism that allows the differential expression of Ly49 receptors is of general interest as it may provide insights into how clonally variable gene expression patterns are established. Clearly, in the case of Ly49 receptors, the expression patterns are generated independent from DNA recombination events in RAG-1 deficient mice ŽFigure 3.. This suggests that Ly49 receptor acquisition is regulated by transcriptional control mechanism. The expression of Ly49A by some but not all NK cells may be due to the NK cell subset-specific expression of a trans-acting factor that is required for recep-

of short-term NK cell clones derived from normal mice identified occasional Ly49Aq NK cells express simultaneously both alleles. The data therefore strongly suggest that Ly49A receptor expression is based on the mutually independent expression of Ly49a alleles Žstochastic model..31 These findings were subsequently extended to additional Ly49 family members, revealing that Ly49c and Ly49g genes are expressed in a fashion similar to Ly49a.32 Mono-allelic expression of individual Ly49 genes together with their tight linkage and frequent co-expression in single NK cells thus raised the question whether co-expression of Ly49 genes occurs from the same or the opposite chromosome. In other words whether co-expression of Ly49 receptors in single NK cells is subject to some sort of co-ordination Žor restriction. on the chromosomal level. The parental origin of Ly49 gene products, however, did not reveal evidence for co-ordinate or restricted expression. Ly49 receptors can be co-expressed from the same or the opposite chromosome Žillustrated in Figure 2..32 These data, therefore support and extend the above considerations: Expression of Ly49 receptors is based on the independent and random activation of receptor alleles. Importantly, Ly49 expression is relatively stable in vitro, as no significant alterations in the expression

Figure 3. Ly49 expression in RAG-1 deficient NK cells. Patterns of Ly49CrI and Ly49G2 expression and co-expression are normal in NK cells of RAG-1 deficient mice.

352

Allele-specific Ly49 gene expression

tor acquisition. In order to generate the complete Ly49 receptor repertoire, each Ly49 receptor would be controlled by the subset-specific expression of a different trans-acting factor. This scenario will not be discussed in more detail as the relevant issue is merely shifted to a higher level of control. Alternatively, Ly49A acquisition during NK cell development may require the simultaneous assembly of multiple trans-acting factors at a local cis-acting element. This process may be inefficient andror binding of one of the relevant factors may be rate limiting due to its limited availability. The inefficient assembly of the putative complex would determine receptor expression in a binary, all or none mode in single NK cells and mediate an acquisition probability on the population level. This scenario can account for the observation that NK cells usually express a single Ly49a allele: The concentration of formed complex may suffice to activate a single allele. It is obviously more difficult to incorporate mono-allelic Ly49a expression into the above model of subset-specific control over Ly49A expression. In extension of these considerations, two scenarios for Ly49 receptor repertoire generation based on limiting complex andror factor availability will be outlined below. For simplicity we will use the term limiting factor for both cases ŽFigure 4..

trans-acting factor, each of which is limiting with respect to receptor acquisition as described above ŽFigure 4A.. This model satisfies the observation that Ly49 receptors are apparently acquired independently. Moreover, sequential factor expression may account for the discrete appearance of distinct Ly49 receptors during NK cell development. However, completely independent control mechanisms for distinct Ly49 receptors may seem somewhat difficult to envisage as their sequences are highly related and these genes have most likely evolved by duplication.

Model II: competitive control of Ly49 genes Alternatively, all Ly49 genes may be controlled by a single set of collaborating trans-acting factors. One of the factors may be rate limiting for the formation of the putative complex. The relevant target sequences in the distinct Ly49 genes would thus compete for the binding of the limiting factor. This results in stochastic and allele-specific Ly49 receptor acquisition ŽFigure 4B.. The probability andror the timing of inducing a given Ly49 gene may be influenced by polymorphisms in the relevant cis-acting element. This scenario delineates an elegant way to differentially distribute the entire set Ly49 receptors. It requires a single trans-acting factor at limiting concentration. Obviously one could envisage a hybrid model in which subsets of Ly49 genes are controlled by distinct limiting factors. Consistent with this possibility

Model I: independent control of Ly49 genes Each Ly49 gene may be controlled by a different

Figure 4. Models for differential Ly49 gene expression. Restricted Ly49 gene expression is based on limited availability of a trans-acting factor Žor a complex thereof. required for Ly49 receptor acquisition. Distinct Ly49 genes in the cluster are regulated by distinct factors ŽA.. All Ly49 genes compete for the binding of a single trans-acting factor available at limiting concentration ŽB..

353

W. Held et al

we have recently identified a trans-acting factor that is required for the acquisition of the Ly49A receptor Žunpublished observation.. As some other Žbut not all. Ly49-defined NK cell subsets are in part dependent on the same factor, the data suggest that the generation of the Ly49 receptor repertoire is controlled by more than one trans-acting factor.

serve a similar purpose. It seems likely that additional examples will be found as the possibility that genes are regulated in an allele-specific manner has only recently received more attention.

Acknowledgements We thank D. Raulet, H.R MacDonald and J.-C. Cerottini for discussion and support. W.H. is the recipient of a START fellowship and supported in part by a grant from the Swiss National Science Foundation.

Role of mono-allelic Ly49 gene expression What is the importance of mono-allelic Ly49 receptor expression for NK cells? First, mono-allelic Ly49a gene expression could in principal act to control the Ly49A cell surface levels, perhaps to facilitate the NK cell’s adaptation to the class I MHC environment.33 Inconsistent with this hypothesis, however, mono-allelic Ly49A expression is observed in the presence and the absence of the Ly49A ligand H-2D d .31 On the other hand, Ly49 genes are polymorphic such that alleles of three Ly49 receptors contained between three and eight amino acid differences in their extracellular portions. As the respective products are distributed to distinct NK cells, mono-allelic expression of Ly49 loci directly increases the number of distinct NK cell clones. However, it is currently not known whether alleles of Ly49 receptors can display distinct ligand specificity. More globally, the expression of the Ly49a gene is restricted to a subset of NK cells. Among this subset it is often restricted to a single allele. Ly49A acquisition can thus be regarded as an inefficient Žand therefore allele-specific. event that restricts Ly49A usage. Since each Ly49 receptor is Žindependently. subject to the same process, NK cell clones ultimately express sets of randomly assorted Ly49 receptors. We therefore propose that mono-allelic Ly49 gene expression reflects the mechanisms that generates the repertoire of combinatorially distributed MHC class I specificities. Mono-allelic gene expression may thus reflect mechanisms that establish clonal variability from identical precursor cells. This variability is generated stochastically, perhaps due to the limiting availability of critical trans-acting factors. This scheme appears particularly suited to differentially express members of multi gene families ŽFigure 4.. Cellular events may subsequently be required to select for clones with useful patterns of gene expression. It appears unlikely that this scenario is used only to satisfy the specific needs of NK cells. Indeed, cytokine expression patterns may be established in a similar way and

References 1. Zhang B, Yamamura T, Kondo T, Fujiwara M, Tabira T Ž1997. Regulation of experimental autoimmune en cephalomyelitis by natural killer ŽNK. cells. J Exp Med 186:1677]1687 2. Ljunggren HG, Karre K Ž1990. In search of the ‘missing self’: MHC molecules and NK cell recognition. Immunol Today 11:237]244 3. Ohlen C, Kling G, Hoglund P, Hansson M, Scangos G, ¨ Bieberich C, Jay G, Karre K Ž1989. Prevention of allogenic bone marrow graft rejection by H-2 transgene in donor mice. Science 246:666]668 4. Vaage JT, Naper C, Lovik G, Lambracht D, Rehm A, Hedrich HJ, Wonigeit K, Rolstad B Ž1994. Control of rat natural killer cell-mediated allorecognition by a major histocompatibility complex region encoding nonclassical class I antigens. J Exp Med 180:641]651 5. Moretta A, Sivori S, Vitale M, Pende D, Morelli L, Augugliaro R, Bottino C, Moretta L Ž1995. Existence of both inhibitory Žp58. and activatory Žp50. receptors for HLA-C molecules in human natural killer cells. J Exp Med 182:875]884 6. Lanier LL, Corliss B, Phillips JH Ž1997. Arousal and inhibition of human NK cells. Immunol Rev 155:145]154 7. Karlhofer FM, Ribaudo RK, Yokoyama WM Ž1992. MHC class I alloantigen specificity of Ly-49q IL-2 activated natural killer cells. Nature 358:66]70 8. McQueen KL, Freeman JD, Takei F, Mager DL Ž1998. Localization of five new Ly49 genes, including three closely related to Ly49c. Immunogenetics 48:174]183. 9. Smith HRC, Karlhofer FM, Yokoyama WM Ž1994. Ly-49 multigene family expressed by IL-2-activated NK cells. J Immunol 153:1068]1079 10. Wong S, Freeman JD, Kelleher C, Mager D, Takei F Ž1991. Ly-49 multigene family. New members of a superfamily of type II membrane proteins with lectin-like domains. J Immunol 147:1417]1423 11. Brown MG, Fulmek S, Matsumoto K, Cho R, Lyons PA, Levy ER, Scalzo AA, Yokoyama WM Ž1997. A 2-Mb YAC contig and physical map of the natural killer gene complex on mouse chromomsome 6. Genomics 42:16]25 12. Brown MG, Scalzo AA, Matsumoto K, Yokoyama WM Ž1997. The natural killer gene complex: a genetic basis for understanding natural killer cell function and innate immunity. Immunol Rev 155:53]66 13. Colonna M, Samaridis J Ž1995. Cloning of immunoglobulinsuperfamily members associated with HLA-C and HLA-B recognition by human natural killer cells. Science 268: 405]408 14. Correa I, Raulet DH Ž1995. Binding of diverse peptides to

354

Allele-specific Ly49 gene expression

15.

16.

17.

18.

19. 20.

21.

22.

23.

24.

MHC class I molecules inhibits target cell lysis by activated natural killer cells. Immunity 2:61]71 Malnati MS, Peruzzi M, Parker KC, Biddison WE, Ciccone E, Moretta A, Long EO Ž1995. Peptide specificity in the recognition of MHC class I by natural killer cell clones. Science 267:1016]1018 Raulet DH, Held W, Correa I, Dorfman J, Wu M-F, Corral L Ž1997. Specificity, tolerance and developmental regulation of natural killer cells defined by expression of class I specific Ly49 receptors. Immunol Rev 155:41]52 Burshtyn D, Scharenberg A, Wagtmann N, Rajagopalan S, Peruzzi M, Kinet J-P, Long EO Ž1996. Recruitment of tyrosine phosphatase HCP by the NK cell inhibitory receptor. Immunity 4:77]85 Nakamura MC, Niemi EC, Fisher MJ, Shultz LD, Seaman WE, Ryan JC Ž1997. Mouse Ly-49A interrupts early signaling events in NK cell cytotoxicity and functionally associates with the SHP-1 tyrosine phosphatase. J Exp Med 147: 3244]3250 Lanier LL, Corliss BC, Wu J, Leong C, Phillips J Ž1998. Immunoreceptor DAP12 bearing a tyrosine-based activation motif is involved in activating NK cells. Nature 391:703]707 Nakamura MC, Linnemeyer PA, Niemi EC, Mason LH, Ortaldo JR, Ryan JC, Seaman WE Ž1999. Mouse Ly49D recognizes H-2Dd and activates natural killer cell cytotoxicity. J Exp Med 189:493]500 Held W, Dorfman JR, Wu M-F, Raulet DH Ž1996. Major histocompatibility complex class I dependent skewing of the natural killer cell Ly49 receptor repertoire. Eur J Immunol 26:2286]2292 Valiante NM, Uhrberg M, Shilling HG, Lienert-Weidenbach K, Arnett KL, D’Andrea A, Phillips JH, Lanier LL, Parham P Ž1998. Functionally and structurally distinct NK cell receptor repertoires in the peripheral blood of two human donors. Immunity 7:739]751 Johansson MH, Bieberich C, Jay G, Karre K, Hoglund P ¨ ¨ Ž1997. Natural killer cell tolerance in mice with mosaic expression of major histocompatibility complex class I molecules. J Exp Med 186:353]364 Williams NS, Klem J, Puzanov IJ, Sivakumar PV, Schatzle JD,

25.

26.

27.

28.

29.

30.

31.

32.

33.

355

Bennett M, Kumar V Ž1998. Natural killer cell differentiation: insights from knockout and transgenic mouse models and in vitro systems. Immunol Rev 165:47]61 Ogasawara K, Hida S, Azimi N, Tagaya Y, Sato T, YokochiFukuda T, Waldmann TA, Taniguchi T, Taki S Ž1998. Requirement for IRF-1 in the microenvironment supporting development of natural killer cells. Nature 391:700]703 Williams NS, Moore TA, Schatzle JD, Puzanov IJ, Sivakumar PA, Zlotnik A, Bennett M, Kumar V Ž1997. Generation of lytic natural killer 1.1 q , Ly49-cells from mutlipotent murine bone marrow progenitors in a stroma-free culture: definition of cytokine requirements and developmental intermediates. J Exp Med 186:1609]1614. Toomey JA, Shrestha S, de la Rue SA, Gayes F, Robinson JH, Chrzanowska-Lightowlers ZMA, Brooks C Ž1998. MHC class I expression protects target cells from lysis by Ly49-deficient fetal NK cells. Eur J Immunol 28:47]56. Dorfman JR, Raulet DH Ž1998. Acquisition of Ly49 receptor expression by developing natural killer cells. J Exp Med 187:609]618 Held W, Roland J, Raulet DH Ž1995. Allelic exclusion of Ly49 family genes encoding class I-MHC-specific receptors on NK cells. Nature 376:355]358 Held W, Raulet DH Ž1997. Ly49A transgenic mice provide evidence for a major histocompatibility complex-dependent education process in natural killer cell development. J Exp Med 185:2079]2088 Held W, Raulet DH Ž1997. Expression of the Ly49A gene in murine natural killer cell clones is predominantly but not exclusively mono-allelic. Eur J Immunol 27:2876]2884 Held W, Kunz B Ž1998. An allele-specific, stochastic gene expression process controls the expression of multiple Ly49 family genes and generates a diverse, MHC-specific NK cell receptor repertoire. Eur J Immunol 28:2407]2416 Sentman CL, Olsson MY, Karre ¨ K Ž1995. Missing self recognition by natural killer cells in MHC class I transgenic mice: a ‘receptor calibration model’ for how effector cells adapt to self. Semin Immunol 7:109]119