Document q30x0vq5vJBJq6EGQBdVgRm3R

Journal o f Clink til Inimunuhixy. Vot. 2. Nn. -4. /VR? V x 201*1 Special Article The Biology of the Human Natural Killer Cell JOHN C. ROOER1and HUGH F. PROSS`-: A trt'fih fl: Ala v 12. /Wf2 Natural killer (NK) cells in [he human are a population of brge granular lymphocytes (LGL) with at least one unique surface antigen not expressed on cells of other iincagcs. NK-target-cell interaction appears to involve carbohydrate recognition and. following binding, the NK celts arc induced to generate ()*". transmethylate mem brane phospholipids, and activate phospholipase A;. Some or all of these activities trigger a cascade of events which ultimately leads to the secretion of a substance toxic to the target cell. A variety of genes controls various steps in this cytolytic pathway. There is a good deal of evidence in the mouse, and some in the human, that NK ceils play a role in host surveillance against tumor devel opment. resistance to viral infections, and, possibly, hematopoietic regulation. KEV WORDS: Natural killer cells: immune surveillance; viral resistance: hematopoietic regulation. introduction Natural killer (NK) cells are lymphocytes from normal nonimmune donors with the capacity to rapidly lyse (within hours) tumor cells, virally infecled cells, or certain undifferentiated norma! cell types both in vivo and in vitro. This " spontaneous" cytolysis was first recognized as the function of a discrete lymphoid subpopulation in 1975 (1-3) and several excellent reviews on early developments in the field have been written (4-6). In this article we attempt to review some recent advances in our understanding of endogenous human NK cells. 'jVrrinrucn.l of Microbiology Immunology. .*0<psriment of Radiation Oncology. Queen'* Unfvcrviiv. Rings* Ian. Ontario, Canada K7I. AN6. CHARACTERISTICS NK cells originate in the fetal liver of man as early as 9 weeks of gestation (7) and are present in significant levels at birth in the cord blood (7. X). I:eial liver also harhnrs N K progenitors in the mouse (9). Following birth there is a gradual in crease in NK levels to high levels in adult hood, with no decrease in activity late in life (10-16). Most NK activity in the adult is found in the peripheral Mood and spleen, with low levels in bone marrow, lymph node, and thoracic duct and no activity in thymus (14. 16. 17). . Human NK cells have been described morpho logically as large granular lymphocytes (L.GL) with a high cytoplasm;nueleus ratio, pale cytoplasm, and slightly eccentric reniform nucleus with 6 to 17 azurophilic granules n their cytoplasm (IS). It has not been possible to separate L.GL from cytolytic function by adherence to target cells (IS) or sedi mentation in Pcrcoll density gradients (19) and the number of LGL binding to K562 tumor cells corre lated with the distribution of cytotoxicity in normal individuals (18). Therefore most, if not all. NK cells arc LGL. Most LGL bear receptors for the Fc portion of IgG and express low-altinily receptors for sheep erythrocytes (20). and two new monoclonal anti bodies have been shown to bind selectively to LGL. Almost till (>929M HNK--I 4 cells separated on the lluorescencc-aciivated cell-sorter (FACS) exhibited the LGL morphology and cytolytic function, whereas HNK --I ' cells lacked hoth characteristics (21). other cells in the blood or a panel of tumor ceil lines bound To IIN K -1 with the exception of ILSB--2. the line used to immunize the mice for 249 ll'lbVN'41.' ribtl r ; 2.M hybridoma production. In double marker studies no H N K -I* cells expressed surface immunoglobulin fslg) or phagocylosed latex beads. However, some H N K -1 ' cells stained with OKT3 or leu*!, which arc pan T reagents, but all of the cytolytic activity was confined to a OKT3" .HLA-DR" .OKM 1* subpopulation comprising approximately 60% of ihe HNK-1 * pool (Abo and Balch, personal communi cation). Since NK cells originate in the bone mar row (y. 22) and since all of the H N K -1' cells in the bone marrow were OKT3Y. then one could specu late that the cyiolyiicalJy active H N K - 1\O K T 3 ". NK subpopulation differentiates from an HNK - I +.OKT3 ' precursor (pre-NK) which is not yet cytolytic (Abo and Hatch, personal communica tion). Another monoclonal antibody. NK-K. from mice immunized with human L.GL. bound to a subpopulation of LG L (33%) but no other cell types in peripheral blood or a panel of cell lines represent ing the T. B. myeloid, or erythroid lineage (23). Seventy percent of the LGL forming conjugates with K562 tumor cells were N K - 8 \ fret real ment of effectors with anti-N K -8 antibody in the ab sence of complement blocked cytotoxicity against isotope-labeled K562 cells and also blocked inter feron release by LGL. although turgei~eff*ceiorcon jugate formation was normal. The mechanism of blockings unknown but it is interesting to note that N K -8 did not block cytotoxicity against fetal fibro blast targets. In summary it would appear that HNK--1 marks all LGL. whereas NK-K is specific only for those LGL capable of binding to target cells. It should be kept in mind that the number of cytolytically active NK cells accounts for only a few percent of peripheral blood lymphocytes. Human NK cells share a plethora of additional surface antigens in common with other cell lin eages. KM1is found on NK cells, monocytes, and granulocytes (24-26). whereas MAC-1 is present on monocytes and NK cells t27). It is unlikely that NK cells belong to the monocyte lineage, however, since they lack the monocyte lineage-specific murk er Mo2 detected on all human monocytes but no other cell types (28). In our hands the majority of nonadhercnl, mononuclear cells enriched for NK cells on Pcrcoll gradients was H N K - l ` and exhib ited LGL morphology, whereas <1% stained with Mo2 or phagocylosed antibody-coated erythrocytes (29?. Furthermore. LGL are positive for -naphthyl acetate esterase (ANAL) in the presence of sodium fluoride, which is known to inhibit the monocyte ANAL reaction (30). Ht>1)1.KXNI) |*H0SJ ,j Ji .J Human LGL also share a number of antigens 3 with T cells. OKTIO is expressed on 62% of LGL j (31). most immature thymocytes, and some (107c) * myeloid cells in the bone marrow (32). whereas 3 markers restricted exclusively to mature f cells \ (OKT3, Icu4) are ahscnl from LGL. 3AI and 5Al2 are expressed on both T cells and LGL (31, 33). These findings do not allow the placement of NK cells on a known lineage, and until proven other wise we assume that NK cells diverge very eariy in lymphopoiesis on a lineage of their own. The exis tence of NK^pccific markers will greatly facilitate work in the field. MECHANISM The development of a single-cell assay for disso ciating NK-mediated binding and lysis in the mouse (34. 35) led to the development of a stimulus secre tion model for NK-mediated eytulysis (35, 36). The model has some similarities to. but also distinct differences from, earlier work in the cytolytic T-cell : (CTL) field (37. 38) as discussed previously (39), Under this model specific target-coll contact results 1 in a change in the cyclic nucleotide balance, or \ some other second messenger, which triggers secre- lion of the lethal substance (39). These results have ' been confirmed and extended in the human (4Qj 1 following the development of conjugate assays in | this species (41-43). ! One of the earliest events (seconds) that can be * detected in the human NK cell, subsequent to ; targct-cci! contact, is the rapid burst in the genera- j tion of the superoxide anion. O j- . as measured by : chemiluminescence or cytochrome reduction (44t j 45). Intact cells or plasma membrane vesicles from } NK-sensitivc, hut not NK-iuscnsitive. targets in- ! duced the response which was necessary for j cytolysis to proceed since superoxide dismulase : but not catalase, specifically blocked both O r ; generation and cytolysis. Although was ncces- sary for NK-mediated cytolysis. it was not sulH-'j cicnt since Chcdiak-Higashi patients produced nor- s mal amounts in the absence of cytolysis. We do not j believe that O; * is the lytic agent but postulate in ' role as a colaclor or activator lor some subsequent I step in the cascade of events leading tu lysis. I The triggering mechanism in NK cells subsequent ! to target-effector binding is unknown, but indirect evidence implicates the cyclic nucleotides. In huth i the mouse (46) and the human (47) the addition of* ni CUnirui !tninnnh<ay, tW. J. ,W. 4, 1 I MOI.OOY OF H U M A N N A T U R A L K II.L IIR l t.l.LS 253 cAMP. or inducers (hereof, to highly enriched NK cells caused an inhibition of cylolysis. whereas cGMP or inducers, including interferon (4X). caused augmentation. Although specific antagonists blocked these effects, specific measurements of intracellular cAMP and cGMP in the NK cell, before and after target-cell contact, have not yet been made. The initial finding that cGMP or induc es could reverse the defective NK response in Chediak-Higashi (CHJ patients (49) has not been ronfirmed in a study of four additional patients (50) or in beige mice (36). the animal model of the CH ..yndrome. A later event in the cytolytic pathway, measur able directly at I hr poslhinding. involves the meth/lation of phospholipids within the NK cell (51). Inhibitors of transmethylation (3-deazaadenosinc) blocked cytolysis, and it is conceivable that trans methylation may enhance signal transduction across the plasma membrane of NK cells as occurs in many other systems (52). A concurrent increase .n phospholipase activity was also measured directly and could be blocked by several specific inhibitors of this enzyme including DL-(2.3-dislearuyloxy propyIdimcthyl-2-hydroxyl)cihy!ammonium icetatc. These results have been confirmed in the nousc (53) but it remains to be shown if phospholi pase A; or its dctcrgcnt-IfKc product, lysophosphaidylcholinc. is the lethal substance. The evidence that secretion may be important in he cytolytic pathway of human NK cells can be ummurized as follows: (i) pretrealmcnl of LGL 'ilh S r* led to degranulalion. and following wash ing, there was a concomitant loss of cytolytic .iciiviiy (54); () cytotoxicity was inhibited by monjsin. a carboxylic ionophorc. which blocked cellu lar secretion in LGL (40); (iii) agents which inhibit cytolysis (cAMP. PGEl, theophylline, histamine) M6) also block lysosomc secretion in other cells (46. 55) at the same concentrations, and conversely, agents which increase lysosomal discharge (cGMP. curbamylchulmci augment NK cytolysis (46): (iv) clilorequine. a selective inhibitor of lysosoincs, -imultar.cously decreased NK cytolysis and N.Tcciyl-p-glucosaminidase levels in mouse spleen 136) and human peripheral blood lymphocytes :PBL) (56); tv) secretion is energy dependent, and ,3NP and CCCP. uncouptcrs of oxidative phosphor ylation. hlnck both NK cytolysis (35. 40) and '.ecrc;:on: and (vi) CHT patients and heige mice carry a fene mutation which selectively impairs both the !cihal-hit stage of NK-mcdialed cytolysis (39) and the secretion of lysosorruil enzymes tin oihe iadney and into the phagolysosome of ilciifcocyta; 157&. Since H N K - P LGL from CH paiiortts ihave a single giant granule ralhenlhun'mnrry ^rnalbynajiiilcs (5H_). it is likely that lack of tatorcium imay atlso explain this genetic NK iin.'mflidiijnt.'y.. NeOibef monocytes norT cells IrarottliCHCjpuiumKurKfofbacd giant granules, and both uifiecttiiiHiiitll trapes food normal cytolytic functionsi(^y..(60). lit'Hiioms'iibdy, therefore, that the giant igcunuL* im :>3K odNs oiay be linked to the functional iclcliiOlimiDHl puticrtls.. Two hypotheses have ihecn put dor-wml far ihe nature of the lethal-hit phic.df^fcHniDUi;itedc\'tc4- ysis. First. it was suggesiuil>{hui;nilytHinm;ii:pr*ti-c- nsc was the lethal substance 'ciCKcliion -cif lysosomal or granule contemn maty lie: iiQvo'f'vwl (above) and since inhibiuw tlf 'KiKiiu.* ictitenuses (PMSF, DIFP) blocked klllingihulmnilhmdnig'toaihc target (35. 39). Subsequent w/ir5k losing n Itarger panel of more specific inilfhikimihik;nxwmiwd rthal n chvmolrypsin-likc activity...-iHiwfrptiiitciito:a:try.p&in activity, may be involvcdiiu.notifynctiwilttern.c^-cnts rather than the lethal hh itstiff ;(5iSi The `second hypothesis suggests that KJ)luliL:i'(Lviicmuili?jivrleaire released by the NK cell mull iKinil .to sfuuiific.jtccep- tor sites on the targct-crill monlbuinniGW). Journals P13L stimulated by fectim:'Kiletwe `solifdle factors which are toxic to NK-^onsUTve.. ;hut 'not NK- rcsistant. targets (61). ill scmiims ft Ihe -iJhown if these toxins are releasad 'Ram .edlln unr tolher cells in the mixture. Thcu:;-;isiunai:oilu:F:ji/nieccp(or sites on the target-cell mimibrutuftha.sLhiteniitffcrred from data showing that 'NiK-nesi+itant imutunis of YAC bound normally to iihe tilK mell Ann iresisicd lysis (62). Cytolysis by nilldim numeOiL w aimti- body and complement waMminiiULtihoxahy-nfUiccsl- ing the existence of a spneifu: uagziilsodll idraermi- nant important for lysis .but mot invuliv.ufl in ttanget- c(feetor binding. It is ooniioiviihlc -fhai Hike Ixixin contains a hexose phosphjtii'Keiadue^diHrihfbmds to the acceptor sties since niunnose-i6HiTho>Hfthaic and stereochemically similar tffuoioK'HNphowp'ttiae in hibit N K-incdialed cyiolysis.a! the inriuit-c.'ioUJcvel. and gelonin conjugated a -.inannoHC-i'ipphosphate was highly toxic to NK-Rcixifirv.::.. -hut :iun NK- resistant, targets (56). In.a similar 'maurior various carbohydrates competed with uncoi*. -denrtod from PPL in another system tuili. The tiuuhiUiT o f NK cells to kill NK-insensiir.ve t;rgot^ (ianooont hy- siant!eis) alter triggering -with N'K-HonMiiivc targets does not exclude the exi+tionoe of -scildHU: toxins. The insensitive target may luck .jicmt.:pi.etr sites for `Hf:..:! t;f Clinn'ol /tiniunnlttgr. Vo/. . ;V.>, 4, I 'VX' ! 252 RODF.R AND PROSS j the toxin or there may be a specialized mode of delivery that prevents leakage. SPECIFICITY NK cells -bind to their targets using one of two alternative modes. The first mode involves a recep tor-ligand interaction between a target structure (TS) on the plasma membrane of the target cells and a recognition receptor (RR) on the surface of the NK cell. In the second mode the target cell is coaled with antibody specific for a largct-ccll anti gen and the NK cell then binds the I;e portion of the antibody to complete the necessary bridge. Both of these binding events appear to activate the same cytolytic pathway within (he NK effector cell. The nature of the RR is not known, but in the mouse (35J and man (unpublished observations) pretreatment of NK cells with proteolytic enzymes (trypsin, papain, pronn.se) eliminated both targetcell binding and cytotoxicity. These two functions regenerated in parallel over a 4-hr time course and could be blocked by cyclohcximidc. Treatment of nontrypxinized cells for IK hr with the minimum concentrations of puromycin or cyclohcximidc re quired to inhibit protein synthesis by 90% also decreased both target-cell binding and lysis. It is likely."therefore, that the NK RR is a dc novo synthesized protein with a turnover lime similar to that of other membrane proteins. The ability of simple sugars to inhibit target-effector binding in the mouse (63) suggests that lectin-like receptors are invojved. although the sugar specificities in the human are clearly different (56; Wcrkmeistcr, per sonal communication). In view of the lack of signifi cant segregation of target specificities in cloned NK lines in the mouse (64). it is unlikely that N K RR are clonally distributed. It is possible, however, that the RR is widely distributed among non-NK leuko cytes since nonlytic thymocytes and peritoneal macrophages preferentially bird NK-scnsitive tar get cells (65). The target structure (TS) has been more exten sively analyzed (66. 67). Preincubation of human PBL with solubilized glycoproteins from NK-sensitive. hut not N-resistant, target cells inhibited binding to the homologous intact target cells by approximately 50% 166). However, these molecules were no; related to any known surface proteins including viral and MHC products, and the quanti ties recovered were loo low to inhibit eytolysis. In retrospect it became clear that only a minority of the conjugates detected in this assay consisted of* lytic NK cells since extensive purifications were! not employed (6K), Other investigators have used* absorption techniques whereby NK cells were de*| plelcd on largct-ccll monolayers and then analyzed; for cytotoxicity against a panel of targets (6V). j Absorption of PBL from some donors with certain cell lines removed cytolytic activity against the; homologous cell line and some third-party lines which were presumed to share an antigen. Reactiv ity against some third-party lines was not removed,! and this analysis concluded that up to seven TSi were involved. Some were unique to a given group j of targets, whereas others were shared among all. Ihe cell lines. Similar evidence lor multiple specific* I ilics was obtained by others using "inicraciiun] analysis" (70), monolayer ahsorptions (71). or com*; pelilive inhibition assays with unlabeled targets ("?.j 73) or purified membrane glycoproteins (66). | In sonic target cells the TS may represent early; antigens that arc lost upon differentiation or matura tion. Hence immature thymocytes lrom neonates.: but not mature thymocytes or T cells, were sensi tive to NK-mediated eytolysis irt the mouse (74. 75i. Human thymocytes were also NK sensitive (76) andthere was an inverse correlation between the sensi-; tivity of fetal fibroblasts and the age of the donor fetus (41). Adult fibroblasts were usually resistant to NK cells. Undifferentiated embryonal carcinoma cells were highly NK susceptible in the mouse, whereas their more differentiated cndodermal pro*, eny were NK resistant (77). In the human there was a good correlation between controlled differentialion in K562. U-937, or CM-K6 tumor cell lines anJ increased resistance to NK-medialed eytolysis (78j, These results have been confirmed and extended in differentiated K562 and HL-60 lines, and Wcrk meistcr cl til. (79) have shown that the NK resist ance is due to a loss of TS and is selective since eytolysis mediated by monocytes or T cells was normal. One could hypothesize that NK cellsrecoynize "fclal-likc" antigens which arc lost upon dif ferentiation of undifferentiated tumor cells or nor mal cells. In the case of the more di[fereniiaiC(< tumurs, these antigens may be derepressed durini the malignant transformation process. Howe\er exceptions exist, and in tlieophj l!mc-induecd niela noma lines (79) NK .sensitivity was increased rathe; than decreased following differentiation. An intricu ing alternative hypothesis has been developed bi Karre which suggests that NK cells are triggered hi a lack of self-MHC antigens of the class I lypc.ani Jtittriutl n j (Vimt'if/ liitHtnrfrtiiijiy, 2. A'd, J 'j lOl.OGY OF H U M A N N A T U R A L K IL L E R C E L L S 253 :i is known that undifferentiated lines express less ' ;l-2 IW>). . ' The changes that occur in the TS during dillcren- lialion are not yel known. However, analysis of sever,j! TS loss variants in the mouse (81--83) has revealed a negative correlation between NK sensi- tivity and the amount of cell-surface sialic acid (S3), V ivhereas a positive correlation between asialo-CM2 and NK sensitivity was found in two independent i*. studies (85. 84). These results suggest that sialic . ucid may mask the TS but further work is needed to : implicate GM2or other gangliosidcs. It has recently been proposed that the mechanism of cffeclortarget interaction involves the binding of elfcclor.rorne transferrin to target-borne transferrin recep' tors (85). and therefore one would predict a correla tion between the rate of proliferation in a given ;argct cell and NK sensitivity since the quantity of transferrin receptor expression is related directly to , trie proliferation rale (86. 87). Alternatively, the transferrin receptor itself inay be the TS. REGULATION Human NK function-is markedly augmented by human interferon of either the alpha or the beta family (88. 89). As previously observed in the mouse (35), interferon does not increase the number ef conjugate-forming NK*cells in the human (43. 90) nut may recruit nonlytic pre-NK cells (F eR ') to oecutr.e cytolytic NK cells (TeRT) by a process requiring mRNA aqd protein synthesis (91). which is also required for the induction of the antiviral : slate (92). Similar findings have been made in (he . mouse (93. 94). It is of interest that the totally deficient NK activity in patients with defective immune interferon production (DIIP) (95) can be ` completely restored to normal levels following in ' vivo injection of interferon (96). During human trials it should be kept in mind that interferon can oppose : the augmented NK activity by simultaneously pro; jading the malignant cells from lysis as shown in the mouse both ;// vitro and in vivo (97). A great many other agents including BCG (98) h;ive also been shown to augment NK activity but most likely operate as inducers (reviewed in Ref 59). I lie human N K cell itself can be induced by tumor cells to synthesize large amounts of interfer on within a short time (18 hr) after stimulation, whereas antigen-stimulated T cells require several . days (99). A variety of agents including f-i-cstradiol ji nonphysiological concentrations tl(Kl). prosta glandins (101), and corticosteroids (102) has been shown to inhibit NK function (39). Some of these agents seem to act through suppressor macrophages in mice (39. 103). Tumor growth itself can often inhibit NK activity by unknown mechanisms (39). On the other hand interleukin-2, a T-ccll product, can also augment NK cytolytic function (104).' NK cells therefore arc enmeshed in a cellular network of regulatory influences within the immune system. GENETIC AND CLINICAL FACTORS Target-cel! recognition in the mouse is under the control of autosomal dominant. H-2-linkcd genes on chromosome 17 (34). whereas the poslhinding cyto lytic pathway is blocked by mutations in the bn gene on chromosome 13 (105) and possibly other mutations on other chromosomes (106: reviewed in Ref. 39). In (he human, males have slightly higher NK activity than females (10. 12. 107). and an associa tion between NK activity and HLA A3. Ii7 (low NK) (107. 108), and Rh negativity (high NK) (109) has been reported. Not all investigators have ob tained similar results (It). I III). It is well estab lished. however, that the level of endogenous NK activity in healthy donors is a stable characteristic of the donor, even when tested over many years (10). and it is possible to categorize healthy people as "high" or "low " NK responders. Liven cord biood lymphocytes show distinct patterns of high and low NK activity (14), as well as the slight male/ female difference seen in adulthood (10). Differences between high and low individuals arc due largely to differences in the triggering of active NK cells, as opposed to differences in recycling capacity (68). The frequency of HNK -1 * cells and target-binding cells does not differ between the two groups, suggesting that low NK responders have a "defect" in the postbinding cytolytic pathway. Chemiluminescence and O ;' production upon tar get-cell contact was deficient, which suggests an impaired triggering mechanism in low NK donors (I II). Studies on the in vitro BCG reactivity of low normal donors (112) indicate that these individuals are less reactive than those with high or normal NK activity. Others have found that the icsponsivcness of low normal adults (fVoss unpublished) or neo nates (14) to interferon is normal. It is presumed that NK activity is genetically determined and analogous to the strain differences in NK activity seen in inbred mice (39). Family J.'ttnm l o f C lin im t fm u tm ii'tfy y . V W . ,V<>. J , / VS.1 254 ODER ANT) I'Ross studies have suggested that similar patterns of reac tivity exist among family members (110) but the results arc not clear-cut. both because of difficulties in NK quantitation (113, 114) and because mar riages between donors at the upper and lower IOth percentile extremes of NK activity are infrequent (2/100). We have studied two pairs of monozygous twins, and the NK activity of each member of the pair was identical to that of the sibling. One of the pairs was studied 20 times over 4 years to establish that they both have similar, high NK activity. Since their NK activity was higher than that of all except I of 105 healthy females who were repeatedly tested, it is unlikely that this observation is due to chance. Identical NK activity was confirmed in these' twins in a single-cell cytotoxicity assay, by H N K -I staining, and using kinetic analysis meth ods (Rubin. Callowaerl. and Pross, unpublished). Several genetic diseases arc also associated with an NK deficiency. The Chediak-Higashi (CH) syn drome caused a profound (59) and selective impair ment in NK cytolytic activity (60. 115), whereas the frequency of H N K -1 + cells (24) and their ability to bind target cells (60) were normal. These results have been confirmed by two independent studies (96, 115). Patients with X-linked lymphoprolifra tive syndrome (XLP) also had impaired NK activity compared tophenolypically normal XLP carriers or normals f 117L Patients with either XLP or CH usually develop fatal lymphoprolifrative malignan cies. which arc clearly EI3V associated in the case of XLP.- Of the classical immunodeficiencies, only severe combined immundefictcncy (SCID) had markedly impaired NK function^ 118. 119). In gen eral, patients with X-linked agammaglobulinemia, common variable immunodeficiency, hypcr-IgM immunodeficiency, hypogammaglobulinemia, and Immunodeficiency thymoma were relatively normal in NK function (120). In other disorders such as IgA deficiency, ataxia telangiectasia, and Wiscott-AIdrich syndrome some patients were low and others were normal in NK function depending on the investigation and the methods employed (119. 121). THE EFFECT OF DISEASE Low NK activity has been described in associa tion with malignant disease (122-I2S), leukemia t22_ 129-132). aplastic anemia (22). chronic ulcerative herpes simplex infection (133). chronic renal failure (i.U. 135). fibrocystic breast disease (I2X). and diseases such as rheumatoid arthritis (136), system^ ic lupus erythematosus (136. 137),. Hashitnoio's thyroiditis (138), and multiple sclerosis (139. 140). In none ol these disorders is there an invariable correlation between low NK function and disease, and some investigators have been unable to find any abnormalities in NK function in patients with multi ple sclerosis (140) or rheumatoid arthritis (141). The existence of N K depression due to therapy has been reported in association with radiation therapy (142), extensive surgery (143). cytotoxic chemotherapy (131). prednisone ( 102), and dielhylstilbeslrol (144). Removal of these agents results in recovery of normal NK function in most patients, and it has also been reported by Guillou et til. (143) that NK function may gradually recover in spite oflong-term immunosuppression with steroids and azathioprine. The mechanisms by which various diseases brine about NK depression arc"extremely"complex and may involve one or more components of the NK system and modulating substances such as interfer on and prostaglandins (145-147). Decreased NK in MS patients was accompanied by a decrease in the ability to make or respond to interferon (139) (McGarry. personal communica tion). A similar defect was inferred by H,crsey et ai. in Fanconi's syndrome I 148) and by Kadish et al. in a study of low NK donors with breast cancer (127). Low NK activity in most cancer patients with solid tumors and leukemia can be accounted for by a low frequency of cytolytic NK cells as detected in a single-cell assay, an observation similar to that made in low NK responder normals (Pross. unpub lished observation). Immunosuppressive therapy may also result in reduced NK frequency as well as defective recycling capacity (6K). In other studies suppression of NK in PBL, in lymph nodes, and in situ in cancer patients has been attributed to sup pressor cells (126. 149--151). u conclusion based on the inhibitory effect of tumor-associated lympho cytes on NK cells hi.vttnt. NK docs not correlate with serum immune complexes in malignant disease (Dorval. Pross, and Baines, unpublished), and in most instances, the mechanisms underlying low S'K activity have not been elucidated. Elevated NK activity in vivo has usually been attributed to activation by interferon. A number of in11aminatory disorders such as atopic dermatitis (152). chronic hepatitis (153. 154). incipient renal transplant rejection (143), and graft vs host disease (155. 156) have been co:related with increased NK activity, although it is unknown whether interferon r . 1'.>/,J t i u n u t l i i f ( ' t i n 'll n t J i i i i i i i i i n i l n y A't . |^ i b o l o g y o f h u m a n n a t u r a l k il l f .r c u l l s 155 is ihe activating agent in all eases. As in animal models, viral (156, I57J and parasitic (158) infec tions lead to an elevation in NK activity which may precede or coexist with specific elfcctor cells. The NK response of normal donors to viral infection was recently determined in a double blind trial in which influent virus or placebo was administered intranasally to volunteers. NK-mediated lysis of K562 lumor cells and MRC-5 fibroblasts increased in parallel with interferon induction, peaking on the third day and correlating with clinical evidence of viral infection (157). antitumor SURVEILLANCE IN MICE As reviewed in detail elsewhere (39. 159. 160), there is a large body of evidence which suggests that NK cells play a role in host defense against transplantable tumors. More recently it has been shown that selectively NK-dclicient (105) homozy gous bg/bg mice are more suceptible to the growth and metastasis of transplantable, syngeneic tumors than their normal +/bg liltermate controls (161163). whereas no differences were found using NKresistani tumors or regressing MSV-induccd tumors (80). which are killed by a T cell-dependent mecha nism rather than NK c^IIs (164). In preliminary f studies by Haliotis ct at. beige mice have a higher incidence of spontaneous tumors (usually lympho mas) laic in life, compared to a lower incidence in a roup of age-matched littcrmate controls (165). As expected, treatment-with NK-supprcssivc doses of carcinogens or radiation did not show any differ ences in the incidence of primary tumors in these two groups of mice (165. 166). In chickens NK cells may prevent the growth of tumors induced by Marck's disease virus (167). These data arc compat ible with the suggestion that NK cells play a role in ,iost defense against primary oncogenesis, at least in the lymphoid system. If NK ceils arc important in antitumor surveil lance, then one would predict a positive correlation between NK sensitivity and tumorigcnic potential in transformed cells. In one study (16S) a cloned line .if normal letal fibroblasts from Galh/'c mice was found to resist NK lysis in vino and these cells did not grow in agarose and were noniumongenic in normal mice or NK-deficicnt mice (type N cells). Clones of chemically transformed variants grew in agarose and formed tumors in NK-deprived mice (NK") but not normal mice with high levels of NK cells (NK' ). These lypi II ooUs w/ow' [highly NK sensitive in vitro. Cell loir* 'established ifrom Tiipe 3 cells growing as turnon: irn NK -mux \wcre NK resistant and grew as lumoire.inmormiibmicciLNKM (type C cells). The N-l mutirtioiiileadingiioianahorage independence was lirikuti !u; iNKf ,*cnsiTni.iiy_ whereas a second mulut;Lii,.witti,xiU|ui(ofJiio*:scap surveillance and grow us .a iiunron iin .luirnui! INK"' mice. In an independent ^tudy t(lfi\9i ;h .was found that a permanent cell line nil IL ai/llbi ($prrKtimabiy type I) was highly sensitive tip ouJlk, -whereas scvcralmalignant sublines'tt-ypr-O-wo.N'.Katcsitflanl. There was no dillorciiucHiuLvuLeininiilignimtumd nonmalignant sublincs in Ueiims .U'^uacupiibitfr*' So lysis by alloimmune Cil!L. aniemirih nges. amtHhody and complement. `Them: dinilings ;in :mioe therefore explain experimentslindhellunrum \whcrfs by spontaneous lumors.HxnnllV.n.tiilb ipdody lysed by NK cells, if at all. ANTI TUMOR SURVr.lIILiNvt'*':.dw;f'i.lA\V*VN-S Animal studies have *:irnvjrTtiitlirxuiiuriCL,lls:icre NK sensitive in vitro. ;(tiun aniimtlks '.with .high INK activity arc more efficient aOiun !iow-?&!Ks umnruils vrt resisting the growth anil mcijnurne vipsend oTthese tumor cells in vivo (39. ii./U. lb IsulNo'Cnriain that this resistance is due :lu cidi Is dOM. 'Con versely. patients with ciiifiiiiniujiossmiih asC H or XLP have an increasedfindideivumLl 'sairKeaancy (117. 172). These obsur-vaiiiwx 'meiietfi :ihul NK cells play a fundamental :rolc in amirimnor surveil lance. Such a theory musi titml-several seemingly paradoxical ub tatuai tinivs: tfii) -the wcsl majority of cancer paliom* in ulte cuifty -Kta.ges of disease has normal levels if .NrK :auFvt.}'i(il12--328, 151 ,~173).*(ii) fresh tmorr.::iKe;tUbkvdkyinarmwaive to lysis by NK cells Pi ^4.. IfTSi). p ikymp'tiocylcs isolated from tumor hk\ptii>;limit NHS.aiCDrwtty,,i(l51. 176-178) (iv) interferon -k, .sehrti.vri^ ineffeouve at inducing NK activity -^griin-st aiiiicflcuicnis rbiiopsy- dcrived target cells ( I7^L amd f/wt) liu: iLneil *c NK activity in irradiated Icukunim putintit* .after suc cessful bone marrow irnutsriliiipuhciu .<!u<rs noil cor relate with the prevent on welaparLM').. All of these observations-nixiuouiiniitWie'wjnh the livpothcsis which stales blm1 euil1*. arre *.\ircum- vcnlcd by lumor v;u i;mi .(il T!! K2. ISO). As shown in Lhc mouse otti. .Sii. ;KT.. IS-IL early neoplastic cells may beihigHly NIK -sntisfLke;.. where as variants can be solerud which irestsa NK lysis Journet nt Clinic,ti Itut,tttni,h v. l 'ni. J . .Vo. J , /V.VJ : 5 6 m m i:K a n d [>roSS and grow up as an observable tumor. This mecha nism may also be operative in metastasis formation: it was shown hy Brooks ct 0!. (182) that mammary adenocarcinoma cells isolated from some metastat ic deposits in the rat were less NK sensitive than the primary tumor. Conversely. Gorelik ct ni. (IKI) selected an NK-resislant sublinc of Lewis lung carcinoma by in vivo passage and found that these cells had an increased potential for spontaneous metastasis formation. In the human, few studies have been underway long enough to relate individ ual NK activity in early-stage disease with ultimate survival. Hcrscy ct al. (125) found that the poslsurgical cell-mediated cytotoxicity of patients with melanoma correlated inversely with recurrence. Mukherji and Dayal (IN3) found no correlation between the development of mne primary melano mas and NK activity in six patients. This " model*' resembles one of surveillance more than studies on tumor recurrence. On the other hand. Morales (personal communication) has noted high NK activ ity in patients receiving intravesical BCG. a treat ment which reduces the frequency of /icn1 transi tional cell carcinomas in patients with unstable bladder epithelium (184. 185). Other investigators have also reported the stimulation of human NK activity in vivo with agents such as BCG (18b). interferon (T87. 188). poly(IC) (189). and Strepto coccal Preparation OK-432 (I9(J). A cause and effect relationship between NK function and appar ent therapeutic benefit can only be inferred from these studies, however. Escape from NK surveillance may also result from the local suppression of NK activity by tumor products (c.g.. prostaglandins (M7). tumor surface " antigens" (66. 191)] or even carcinogens and tu mor promoters, such as has been shown with ure thane (192) and phorhol esters (193). It is unlikely ~ that NK cells function against allogeneic targets only, in vitro, since early work established that fresh. Fc receptor-positive lymphocytes from nor mal donors can lyse syngeneic LBV-transformed cells (4j. Based on the data from animal models, the pe ripheral blood and splenic distribution of NK-ccIl activity, the NK resistance of established tumors, and the lymphoproliferalive nature of malignancies in NK-deficient patients, we support the hypothesis that NK cells arc respon>iblc tor early surveillance against malignancies of the hematopoietic system and. possibly, blood-borne mctastases Iruin solid tumors. OTHER FUNCTIONS A detailed discussion of other functions which have been attributed to human NK cells is beyond the scope of this brief review. As mentioned above. NK cells may also be important in host reactivity against vitally infected cells. .Human cell lines which are NK resistant (Hela) become markedly NK sensitive in vivo in nude mice upon infection with a variety of persistent viruses (194). Infected cell lines are also sensitive to human NK cells (195. 196) and NK-deficicnl beige mice have a much lower LDmi for murine cytomegalovirus than nor mal litlcrmate controls (197). an elfecl which is due to impaired NK resistance rather than impaired viral replication. Since human NK cells lysed virusproducing cells at an early stage of the LBV cycle before virus panicles were assembled, it was sug gested that they may restrict the spread of virus from latently infected cells (198). It is also possible that NK cells serve primarily as regulator cells of hematopoiesis. The evidence for this is derived mainly from studies on hybrid resist ance by Cudkowic/. and colleagues (103) which show a high degree of parallelism between the characteristics of the parental murine spleen cells which are the effectors in Ft bone marrow rejection and those of NK cells. In man. it has been suggest ed that hone marrow transplant-recipient NK cells (as defined by HSV-I-infected fibroblast lysis) interact with IILA A. B. C. and D matched donor hematopoietic cells. The result of this interaction was stimulation of the donor cells to produce graft vs host disease. In transplant-surviving donors with no NK (H S V -I) activity prior to transplant, no GVH disease was observed. Aside from the prog nostic value of these NK results, it s possible that NK ccll-hcmaiopoictic cell interaction is the result of the NK cells attempting to carry out a normal regulatory function. Other evidence suggesting that NK cells may have a regulatory function stems from studies on patients with NK proliferation disorders. Although several patients have been de scribed with T,- or K-cctl lymphocytosis (199-201: reviewed in Ref. 201)..only two have been de scribed in which NK cells are functionally active and increased in number (200. 201). In these pa tients. leukopenia and recurrent infections were the principal presenting problems, suggesting`that NK cells actively suppress leukopoiesis. Research into the effects of purified human NK cells on hemato poiesis in vitro will hopefully provide definitive answers as to the regulatory role of these cells. Jn im iitl t>( Ch'tiit til liinnnt)" l'>!L`v. 1'>!. 2. .Vii, J. /; BIOLOGY OF H UM AN N A TU RA L K IL L E R C E L L S 257 acknowledgments Wc lh;ink the Medical Research Council of Cana da. ihc National Cancer Institute of Canada, and .he Ontario Cancer Foundation for support. Mrs. Christine Jackson and Dianne Robertson provided excellent secretarial assistance. REFERENCES 1. Pross I IF. Jondal M: Cytotoxic lymphocytes from normal donors. A functional marker of human non-T lymphocytes. Clin Exp Immunol 21:226-235. 1975 2. Herberman RU. Nunn MK. Holden HT. l..i\rm DM: Natu ral cytotoxic reactivity of mouse lymphoid cells against syngeneic and allogeneic tumors. II. Characterization of elfecior cells. Ini J Cancer 16:230-239. 1975 3. K<cxs!ing K. Klein E. Wigzetl II: Natural killer cells in the mouse. I. Cytotoxic cells with specificity for mouse Mo loney ieukemia cells. Specificity and distribution according to genotype. Fur J Immunol 5:112-117. 1975 J. Pross III-'. Haines MG: Spontaneous human lymphocytemediated cytoloxicily against tumor targci cells. VI. A brief review. Cancer Immunol Immunother 3:75-85. Iy77 5. Herberman RB. Holden HT: Natural cell-mediated immu nity. Adv Cancer Res 27:305-377. 1978 6. Herberman RB. Djcu JY. Kay HD. Ortaldo JR. Kiccardi C. Bonnard GD. Holden HT. Kagnani R. Sanioni A. Puceetti I1: Natural kilter cells: Characteristics and regulation of activity. Immunol Rev *tJ;'i?-70. 1979 7. Toivancn I1. Uksila J. I.eino A. I.assila O. Itirvonen T, Kuuskanen O: Development of mitogen responding T cells and natural killer cells in the human fetus. Immunol Rev 57:89-105. 1981 8. Baines M G . Pross HK, Millar KG: Spimtancous human lymphocyte-mediated cytotoxicity against tumor target cells. IV. The suppressive elfecl of normal pregnancy. Am J Ohstet Gynecol 130:741-744. 1978 9. Haller 0 . Kicssling R. Orn A. Wigzcll H: Generation of natural killer cells: An autonomous function of the bone marrow. J Exp Med 145:1411-1416. 1977 10. Pross I IF. Haines MG: Sliidies"'df human natural killer cells. I. In im i parameters affecting normal cytotoxic function. Ini J Cancer 29:3S3-390. 1982 11. Fernandes G. Gupta S: Natural killing and antibody-depen dent cytotoxicity by lymphocyte subpopulalions in young and aging humans. J Clin Immunol 1:141-148. 1981 12. Pcnschow J. MacKay 1R: NK and K cell activity of human blood: dillercnees according to sex. age. and disease. Ann Rheum Dis 39:82-86. 1980 13. Bloom Iff: Lymphocytes from chi Mien and infants mediate natural cell mediated cytoloxicily. Cancer Immunol 1mnitnnithcr 10:261-263. 1981 |J. Anionclli l \ Stewart W. Dupont B: Distribution of natural killer cell activiit m peripheral blood, coal Mood, thymus, lymph nodes, and spleen amt the ellcct of/ rim treatment ith interferon preparation. Clin Immunol (mninnupathol 19.161-169. |';SI 15. Nagel JE, Collins GD. Adler WII: Spontaneous nr natural killer cytotoxicity of K562 crythrolcnkemic cells in normal patients. Cancer Res 41:2284-2288. 1981 16. Forbes JT. Niblack GD. Fuchs K. Richie RE. Johnson HK. Oldham RK: Human Natural Cell-mediated Cylotoxicily. I. Levels in peripheral blood. cord blood, and thoracic duct Ivmphocylcs. C a n c e r'Immunol Immunother 11:09-145. 1981 17. Turgan SR. Gale RP. Jomlni M: Bone marrinx modulation of spontaneous killer cellular cytotoxicity tSKCX'l induc tion. J Clin Lab Immunol 4:1-6. 19X0 18. Timonen T. Ranki A. Saksela E. Hayry I*: Fractionation, morphological and functional chat act cri/al ion of ctTcclor cells responsible for human natural killer activity against cell-line targets. Cell Immunol 48:121 I *2. 1979 19. Timonen T. Saksela E: Isolation o f human natural killer cells by density gradient centrifugation. J Immunol Meth ods 36:285-291. 19X11 20. Timonen T, Ortaldo JR. Ilcrhcniiun Kit: ( `haractci sties ol human large granular lyin p liiw les anti lelaiioiiship to natural killer and K cells. J Exp Med l53:N.9-sgJ. I`<si 21. Abo T. Halch CM: A ditlcrentiaiiou antigen of human NK and K colls identified by a monoclonal antibody H IN K -I). J Immunol 127:1(124-11129. 1981 22. l.ivnat S. Seigneuret M. Snub R. Picntice Kl.: Analysis of cytotoxic effector cell function in patients with leukemia or aplastic anemia before and after man no uansplaiiUdion. J Immunol 124:481-49(1. 1980 23. Nieminen P. Paasivuo R. Saksela li: KlVeet o f a monoclonal anti-large granular lymphocyte antibody on the human NK activity. J Immunol 128:1097-1101. 19X2 24. Ilreard J. Reinher/. EL. t i 'llrien Schlossman SI-: Delin eation of an elTcctor population responsible lor natural killing and antibody dependent cellular cytotoxicity in man. Clin Immunol linmtumpaihol I7:.|J5 -I5 I. 1980 25. Breard J. Reinherz EL. Kttng I'C. Goldstein ( i. Sehlossmau SF: A monoclonal antibody reactive with human peripheral blood monocytes. J Immunol 124: l943-ty4X. 198(1 26. Zarling JM. Kung PC: Monoclonal-antibodies which distin guish between human NK and cytotoxic T lymphocytes. Nature 288:394-396. 198(1 27. Ault KA. Springer TA: Cross reaction o f a nit anti-mouse phagocyte specific monoclonal antibody (ami-M ac-ll with human monocytes and natural killer cells. J Immunol 126:359-364. 1981 28. Todd RF. Nadler LM. Schlossman SF: Antigens on human monocytes identified by monoclonal antibodies. J Immunol 126:1435-1442. 1981 29. Rodcr JC. Todd RF. Rubin f*. Ilaliolis T. flelfand S. W crkmcister J. Pross IIF. Boxer I.. Sehlossmari SF. Fauei AS: The Chcdiak-lligashi gene in Iminaus. HI. Studies on the mechanism of NK impairment (submitted for publica tion) 31). Ranki A : N o iisp ccih c esterase a ctiv ity in liumuu lym pho cytes. Histochem ical ekaracieri/aii-m and distribution among major lymphocyte siihcl.is-.es. i lm limn...... inimunopathol 10:47-52. 1978 31. Ortaldo JK, Shaitou St). *1inixnen I. 1tcih ciiu.ui RB: Determination of surface antigens on highly purified human NK cells by (low cytometry with monoclonal .mtibodies. J Immunol 127:24111-24(19. |98l Jnunml i>{ CHi'iciit \ 3 , A n, 4 . 19X2