Polarization in NK-Target Cell Interactions Roles of Chemokines

Roles of Chemokines and Receptor Polarization in NK-Target Cell Interactions1

Marta Nieto,* Francisco Navarro,* Juan Jose´ Perez-Villar,* Miguel Angel del Pozo,* Roberto Gonza´lez-Amaro,* Mario Mellado,† Jose´ M. R. Frade,† Carlos Martı´nez-A,† Miguel Lo´pez-Botet,* and Francisco Sa´nchez-Madrid2*

We report that the ability of NK cells to produce chemokines is increased in NK-target cell conjugates. The chemokines produced play a critical role in the polarization and recruitment of NK cells as well as in the NK effector-target cell conjugate formation. Chemokines induce the formation of two specialized regions in the NK cell: the advancing front or leading edge, where chemokine receptors CCR2 and CCR5 cluster, which might guide the cells toward the chemotactic source, and the uropod, where adhesion molecules ICAM-1 and -3 are redistributed. NK cell polarity was intrinsically involved in conjugate formation. The redistribution of both adhesion receptors and CCR was preserved during the formation of NK-target cell conjugates. Time-lapse videomicroscopy studies of the formation of effector-target conjugates showed that morphologic poles are also functionally distinct; while the binding to target cells was preferentially mediated through the leading edge, the uropod was found at the rear of migrating NK cells and recruited additional NK cells to the vicinity of K562 target cells. Inhibition of cell polarization and adhesion receptor redistribution blocked the formation of NK-K562 cell conjugates and the cytotoxic activity of NK cells. We discuss the implication of NK-cell polarization in the development of cytotoxic responses. The Journal of Immunology, 1998, 161: 3330–3339.

K cells belong to a distinct lineage of lymphocytes ca- nesis of NK cells in vitro. In addition, it has been reported that pable of killing a variety of target cells, including tumor chemokines stimulate Ca2ϩ mobilization and cytolytic granule re- N cells, cells infected by viruses or bacteria, and some nor- lease, promote cytotoxic activity, and regulate the adhesiveness of mal cells (1). NK cells resemble CTL in certain respects, including NK cells (11–14). NK cells, in turn, synthesize chemotactic fac- the cytokine production profile and the use of a perforin-dependent tors, including IL-8, MIP-1␣,3 and lymphotactin (15–17), which lytic mechanism, but they do not rearrange TCR or Ig genes. The may induce the migration of additional effector cells into the target migration and recruitment of NK cells from blood vessels to target tissue. Chemokines specifically interact with receptors that possess tissues are the first steps in the cascade of events in the NK cyto- seven transmembrane domains and are coupled to a G protein sig- toxic response. The diverse factors involved in this complex pro- naling pathway. Several of these receptors have been cloned and cess have recently been explored. NK cells exhibit rapid migration classified into two groups, CCR and CXCR (6, 18), but their pat- ␤ in vitro (2) and express a repertoire of adhesion molecules from 1 terns of expression on NK cells have not been fully elucidated. ␤ and 2 integrin families, enabling them to interact with endothelial Leukocyte migration requires cell polarization, a phenomenon cell counter-receptors (3, 4) and extracellular matrix proteins (5). that is involved in many other processes, such as cell differentia- It has been widely proposed that chemokines are the major sol- tion, vectorial transport of molecules across cell layers, induction uble mediators of leukocyte recruitment in inflammation and im- of immune response, cognate interactions between APC and T munity (6–8). These molecules are differentially secreted by plate- cells, and target cell recognition and killing (19–24). Cell polar- lets and a broad spectrum of nucleated cells, including endothelial ization is required for the release of NK cytolytic granules as well cells, T lymphocytes, and monocytes (7, 9, 10). Although little is as for the formation of conjugates between killer cells and their known about the effects of chemokines on human NK cells, it has target cells (25). We recently reported that chemokines and other been shown that these cytokines induce chemotaxis and chemoki- chemotactic cytokines induce in T cells the polarization of chemokine receptors (CCR) to the cell leading edge, probably directing the cells along the chemokine gradient (26). In addition, che- *Servicio de Inmunologıá, Hospital de la Princesa, Universidad Autońoma de Madrid, mokines induce the formation of a cell projection at the rear of the and †Department of Immunology and Oncology, Centro Nacional de Biotecnologıá, CSIC, Universidad Autońoma de Madrid, Campus de Cantoblanco, Madrid, Spain cell, termed the uropod, accompanied by redistribution of specific Received for publication February 23, 1998. Accepted for publication June 1, 1998. adhesion molecules such as ICAM-1, ICAM-3, CD43, and CD44 The costs of publication of this article were defrayed in part by the payment of page to this structure (27–29). charges. This article must therefore be hereby marked advertisement in accordance We have studied the polarization of NK cells induced by che- with 18 U.S.C. Section 1734 solely to indicate this fact. mokines or during NK-target cell interaction as well as the possi- 1 This work was supported by Grant SAF 96/0039 from the Ministerio de Educacioń ble role of the cellular uropod in the migration and recruitment of y Ciencia, Grant 07/44/96 from the Comunidad Autońoma de Madrid, a grant from Fundcioń Cientifica de la Asociacioń Espanõla centra el Cancer (to F.S.-M.), and NK cells. We have found that the leading edge of NK cells con- fellowships from the Fondo de Investigaciones Sanitarias BAE 97/5089 (to M.A.P.) centrates chemokine receptors and is implicated in target interac- and the Ministerio de Educacioń y Ciencia (to M.N.). The Department of Immunology tions, while the uropod mediates homotypic NK cell interaction, and Oncology was founded and is supported by the Consejo Superior de Investiga- ciones Cientificas and Pharmacia and Upjohn. which might be important for the recruitment of these cells into 2 Address correspondence and reprint requests to Dr. Francisco Sańchez-Madrid, Servicio de Inmunologıá, Hospital de la Princesa, Universidad Autońoma de Madrid, Diego de Leoń 62, E-28006, Madrid, Spain. E-mail address: fsmadrid/ 3 Abbreviations used in this paper: MIP-1, macrophage inflammatory protein-1; rh, [email protected] recombinant human; MCP, monocyte chemoattractant protein; FN, fibronectin.

Copyright © 1998 by The American Association of Immunologists 0022-1767/98/$02.00 The Journal of Immunology 3331 target tissues. We discuss the implication of these findings in re- culated by direct counting of total cells (n ϭ 400–500) and uropod-bearing lation to mechanisms relevant to the in vivo migration of NK cells. cells in 10 random fields (ϫ60 objective) for each condition. Preparations were photographed on Ektachrome 400 film (Eastman Kodak, Rochester, NY). For immunofluorescence studies of NK-K562 cell conjugates, 105 K562 Materials and Methods cells and 2 ϫ 105 NK cells were mixed and incubated in Ca2ϩ-free com- Abs, cytokines, and reagents plete medium (ICN, Costa Mesa, CA), to avoid cytolysis, at 37°C in a 5% CO atmosphere for 20 min in 96-well plates (Costar). Cells were then The mAbs anti-ICAM-3 HP2/19 (IgG2a), anti-CD11a TP1/40, anti-CD45 2 fixed with 3.7% formaldehyde, rinsed in Tris-buffered saline (TBS), D3/9 (IgG1), and anti-CD94 HP3B1 (IgG2a) have been previously de- stained with mAb, washed twice with TBS, and sequentially rinsed and scribed (30–32). The anti-ICAM-1 MEM 111 was a gift from Dr. V. incubated for 15 min with a biotinylated anti-mouse IgG mAb (1/300 di- Hore¨jsi (Institute of Molecular Genetics, Videnska, Czech Republic), and lution), a fluorescein-labeled avidin D (1/1000), a biotinylated anti-avidin the anti-CD16 B73.1 (IgG1) (33) was a gift from Dr. B. Perussia. The D (1/100), and a fluorescein-labeled avidin D (1/500; all from Vector, anti-CD56 mAb K218 (IgG1) was provided by Dr. A. Moretta (Instituto Burlingame, CA). CCR2 and CCR5 were stained with the appropriate mAb Nazionale per la Ricerca sul Cancro e Centro Biotecnologie Avanzate, and then incubated with a 1/50 dilution of an FITC-labeled rabbit F(abЈ) University of Genova, Genova, Italy), and the L16 mAb was donated by 2 anti-mouse IgG (Pierce). Cells were observed and photographed using a Dr. C. Figdor (Nijmegen, The Netherlands). Anti-CCR2 (CCR2-03) and Nikon Labophot-2 photomicroscope with ϫ40 and ϫ60 oil immersion anti-CCR5 (CCR5-01) mAb have previously been described. Recombinant objectives. human (rh) RANTES, MCP-1, MIP-1␣, MIP-1␤ chemokines, and rhIL-15 ␣ were purchased from PeproTech (London, U.K.), and rhTNF- (sp. act., Time-lapse videomicroscopy 5 ϫ 107 U/mg; purity, Ͼ95%) was provided by Genentech (San Francisco, CA). The rhIL-2 was provided by Hoffmann-La Roche (Nutley, NJ). Videomicroscopy analysis was performed as previously described (35) using a Nikon Diaphot 300 inverted microscope equipped with a black Protein substrates and white video camera (Sony SSC-M350CE) coupled to a time-lapse Recombinant chimeric ICAM-1-Fc, consisting of the total extracellular videocassette recorder (Sony SVT-5000P) and a video monitor (Sony domains of ICAM-1 fused to the IgG1 Fc fragment, was obtained as pre- PVM-1453 MD). NK cells were allowed to attach for 30 min at 37°C to viously described (34). Briefly, COS-7 cells were transiently transfected with ICAM-1-Fc (ICAM-1 cDNA cloned in pCD8IgG1). After 4 days, culture supernatants were precipitated with ammonium sulfate, and chimeric proteins were isolated using protein A coupled to Sepharose (Phar- macia, Uppsala, Sweden). The tryptic 38- and 80-kDa fibronectin frag- ments (FN40 and FN80) were gifts from Dr. A. Garcıá-Pardo (Centro de Investigaciones Biolo´gicas, Madrid, Spain). Collagen type I, laminin, poly- L-lysine, and fibrinogen were purchased from Sigma (St. Louis, MO). BSA was obtained from Boehringer Mannheim (Mannheim, Germany). Cells IL-2-cultured NK cells were obtained essentially as previously described (32). In brief, PBL were cultured with irradiated (5 Gy) RPMI 8866 lym- phoblastoid cells for 6 to 9 days in RPMI 1640 supplemented with 10% FCS (complete medium), followed by a negative selection step using an anti-CD3 mAb plus rabbit complement (Behring, Marburg, Germany). The CD3Ϫ cells (Ͻ5% CD3ϩ) were cultured with 50 IU/ml of rhIL2 until use. Fresh NK cell-enriched populations were obtained from PBL by removal of adherent cells on plastic petri dishes, followed by passage through a nylon wool column, and depletion of T cells with an anti-CD3 mAb plus rabbit complement (32). These cell populations are hereafter referred to as NK cells. After each purification process, the resulting population was characterized by flow cytometric analysis. We routinely obtained a cell population with percentages of CD56ϩ and CD16ϩ cells Ͼ95% and with Ͻ5% of CD3ϩ, CD19ϩ, or CD14ϩ contaminating cells. For polarization studies, NK cells were cultured in the absence of exogenous IL-2 for 12 h before the experiment. Erythroleukemic K562 target cells were cultured in complete medium. Flow cytometric analysis Cells were saturated with ␥-globulin, incubated with the appropriate mAb for 30 min at 4°C, then washed twice and incubated with a 1/50 dilution of Ј FITC-labeled rabbit F(ab )2 anti-mouse IgG (Pierce, Rockford, IL) for 20 min at 4°C. Double-staining experiments on unfractionated PBMC were performed using phycoerythrin-labeled anti-CD56 mAb after incubation with 10% mouse serum. Flow cytometric analysis was performed using a FACScan cytofluorometer (Becton Dickinson, Mountain View, CA). FIGURE 1. Chemokine production and chemoattraction of NK cells during NK-K562 cell interaction. A, Chemoattraction of NK cells during Immunofluorescence analysis NK-K562 cell interaction. IL-2-activated NK and K562 cells, either alone The induction and detection of the uropod on NK cells were performed or mixed, were cultured in the lower well of a transwell chamber for 4 h. essentially as previously described (28). Briefly, 1 ϫ 106 NK cells were Then, the chemotaxis of 51Cr-labeled NK cells, poured in the upper cham- incubated on coverslips coated with various protein substrates in flat-bot- ber, was calculated as described in Materials and Methods. The arithmetic tom 24-well plates (Costar, Cambridge, MA) in a final volume of 500 ml mean Ϯ SEM of three independent experiments with three different donors p Ͻ 0.05 compared with cell migration in response to the ,ء .of complete medium. Chemokines and other cytokines were added at dif- is shown p Ͻ 0.05 compared with cell migration in response to NK ,ءء ;ferent concentrations, and cells were allowed to settle for 30 min at 37°C medium ina5%CO atmosphere. Cells were then fixed with 3.7% formaldehyde in 2 cells (as determined by Student’s t test). B, Chemokine production induced PBS for 10 min at room temperature, rinsed in TBS (50 mM Tris-HCl (pH by NK-K562 conjugate formation. IL-2-activated NK and K562 cells, ei- 7.6), 150 mM NaCl, and 0.1% NaN3), and incubated with specific mAb. After washing, cells were stained with a 1/50 dilution of an FITC-labeled ther alone or mixed, were incubated in complete medium for 4 h. Then, cell Ј supernatants were assayed for different chemokines as described in Mate- rabbit F(ab )2 anti-mouse IgG (Pierce) and analyzed using a Nikon Labo- phot-2 photomicroscope (Nikon, Melville, NY) with ϫ40, ϫ60, and ϫ100 rials and Methods. The arithmetic mean Ϯ SEM of two independent ex- oil immersion objectives. The proportion of uropod-bearing cells was cal- periments with cells from two different donors are shown. 3332 CHEMOKINES AND NK CELL POLARIZATION

FIGURE 2. Chemokines induce the redistribution of chemokine receptors to the leading edge of NK cells. a, Cell membrane expression of CCR2 and CCR5 chemokine receptors in freshly isolated and IL-2-activated NK cells. Flow cytometric analysis of the binding of the anti-CCR2 (CCR2-03; solid histogram, A and B) and anti-CCR5 (CCR5-01; dotted histogram) mAb to freshly isolated peripheral blood NK cells (A) and IL-15-activated NK cells (C). Dashed line histograms correspond to the irrelevant myeloma P3X63 mAb. b and c, The roles of chemoattractants in the redistribution of CCR2 and CCR5 receptors on IL-2-activated NK cells. IL-2-activated NK cells stimulated with 10 ng/ml of different chemokines, TNF-␣, or PMA were allowed to adhere to coverslips coated with FN80 (20 ␮g/ml) or recombinant soluble ICAM-1-Fc (10 ␮g/ml). Cells were then ﬁxed and stained for CCR2 (b) and CCR5 (c), and the percentage of cells on which these receptors were redistributed was calculated as described. The arithmetic mean Ϯ SEM of three independent p Ͻ 0.01 (compared with untreated cells, as determined by ,ءء ;p Ͻ 0.05 ,ء .experiments performed with cells from three different donors are shown Student’s t test).

10-mm plastic petri dishes (Costar) previously coated with ICAM-1 (10 Chemokine quantification ␮g/ml), in the presence of different chemokines or the polarization- ϫ 5 ϫ 5 inducing anti-ICAM-3 mAb HP2/19. A second cohort of unstimulated K562 target cells (5 10 ), NK cells (5 10 ), or a mixture of both cells were incubated in a final volume of 1 ml of complete medium for4hat NK cells was then added, and cell-cell interactions were filmed for 1 h ϫ 37°C in 5% CO2. Cells were then centrifuged, and the chemokines present using a 20 phase contrast objective. When indicated, cells were pre- in the supernatant were quantified. Human chemokines MCP-1 and treated for 15 min at 37°C with different blocking mAb. Images were RANTES were measured using the Cytoscreen immunoassay kit (Bio- acquired every 30 s, and sequential frames were photographed. Under Source, Camarillo, CA), and human MIP-1␣ and MIP-1␤ were determined each experimental condition, the number of attached cells from the first using the Quantikine immunoassay kit (R&D Systems, Minneapolis, MN), cohort that were moving on the substrate (phase-dark cells) was counted following the manufacturer’s instructions. as well as the number of cells from the second cohort (phase-bright cells) that were interacting with the uropod of adhered cells. The re- Cytotoxicity assays cruitment index was expressed as the number of cells of the second NK cell cytotoxicity was tested against the K562 cell line in a 4-h 51Cr cohort being captured/number of the cells of the first layer that adhered release assay at different E:T cell ratios, and specific lysis was calculated to the substrate. as previously described (32). Data are expressed as the arithmetic mean of triplicate determinations. In each case, spontaneous release was always Ͻ10% of the maximum lysis. Chemotaxis transwell assays Statistical analysis Chemotaxis assays were performed in triplicate in transwell cell culture chambers (Costar). These chambers contain an upper and a lower well Significant values were determined using one-tailed Student’s t test. separated by a tissue culture-treated polycarbonate membrane (poly- vinylpyrrolidone free), 6.5 mm in diameter, 10 mm thick, and with a Results pore size of 5 ␮m. K562 target cells (5 ϫ 105), NK cells (5 ϫ 105), or NK-target cell cocultures release chemokines and attract a mixture of both cells were preincubated in a final volume of 0.6 ml of additional NK cells complete medium in the lower chamber for 4 h. Then, 10651Cr-labeled NK cells were added to the upper chamber. After 2-h incubation at 37°C Chemokines are produced by different cell types, including NK cells (15–17). We analyzed whether NK cells, alone or mixed with in 5% CO2, the number of cells that migrated to the lower chamber was calculated relative to the total input of 51Cr-labeled NK cells. Specific target cells, release soluble factors that attract additional NK cells. migration was calculated after subtracting spontaneous migration. Using a chemotaxis transwell assay, we found that cocultures of The Journal of Immunology 3333

FIGURE 3. Chemokines induce the redistribution of ICAM-3 to the uropod on NK cells. a, ICAM-3 redistribution to the uropod of NK cells. ICAM-3 distribution on NK cells adhered to rsICAM-1-coated coverslips, either untreated (A) or after stimulation with 10 ng/ml of RANTES (B). Bar ϭ 10 ␮m. b, Dose-response assay of chemokine-mediated uropod formation. NK cells adhered to recombinant soluble ICAM-1-coated coverslips were stimulated with different doses of chemokines or cytokine TNF-␣. After 30-min incubation, cells were fixed and stained for ICAM-3, and uropod formation was quantified as described in Materials and Methods. One representative experiment of three independent ones conducted with cells from different donors is shown. c, The effects of different chemoattractants on the redistribution of ICAM-3 to the uropod of NK cells. NK cells adhered to rsICAM-1-coated coverslips were stimulated with 10 ng/ml of chemokines, 10 ng/ml of IL-15, or 10 ng/ml of TNF-␣. After 30-min incubation, cells were fixed and stained for ICAM-3, and uropod formation was quantified as described in Materials and Methods. The arithmetic p Ͻ 0.01 compared with untreated cells, as ,ء .mean Ϯ SEM of three independent experiments performed with three different donors are shown determined by Student’s t test.

NK cells with K562 target cells induced a noticeable migration of key cytokine in the development of NK cells (38, 39), showed a additional NK cells, whereas K562 or NK cells alone had a weak comparable effect on the induction of expression of these two che- effect (Fig. 1A). Measurement of chemokines in these supernatants mokine receptors (Fig. 2aC). These results are consistent with studies also revealed that NK cells produced MIP-1␣, RANTES, and MIP- of the up-regulated CCR2 mRNA expression in NK cells (40). 1␤, but not MCP-1, whereas only trace amounts of these chemo- We have reported that chemokines induce the formation of two kines were found in K562 cell supernatants. Interestingly, the se- specialized cell compartments in T cells; while chemokine recep- cretion of these chemokines, probably by NK cells, was tors cluster at the leading edge of migrating T cells, adhesion mol- dramatically increased upon coculture of NK effectors with K562 ecules (ICAM-1, ICAM-3, CD43, and CD44) redistribute to the target cells (Fig. 1B). These results indicate that the interaction of cell uropod (26, 28). We therefore explored the roles of chemo- NK cells with target cells induces the release of physiologically kines in regulating NK cell morphology and receptor distribution. relevant amounts of chemokines. The chemokines RANTES, MCP-1, and, to a lesser extent, MIP-1␣ and MIP-1␤ induced polarization of both CCR2 and Chemokines induce the formation of two specialized domains in CCR5 to the leading edge of IL-2-activated NK cells. Neither the NK cells proinflammatory cytokine TNF-␣ nor the activating molecule Membrane expression of CCR2, a receptor for MCP-1, MCP-3, PMA induced CCR polarization. Two integrin ligands, FN80 and and MCP-4 (36), and CCR5, which specifically binds RANTES, ICAM-1, supported CCR redistribution, which is consistent with MIP-1␣, and MIP-1␤ (37), was barely detectable on freshly iso- our previous data indicating that receptor polarization requires ad- lated NK cells (Fig. 2aA). In contrast, CCR2 and CCR5 expres- hesion through integrins (Fig. 2, b and c). Chemokines also trig- sions were up-regulated in a significant proportion of IL-2-acti- gered the formation of the uropod and ICAM-3 redistribution on vated NK cells (Fig. 2aB). IL-15, which has been reported to be a IL-2-activated NK cells in a dose-dependent manner (Fig. 3, a–c). 3334 CHEMOKINES AND NK CELL POLARIZATION

FIGURE 4. Polarization of chemokine receptors during NK-target cell conjugate formation. a, The cell uropod is displayed at the rear of NK cells during conjugate formation. The migration of NK cells on FN80 and NK- K562 effector-target cell conjugate formation was recorded by time-lapse video. Sequential video images are shown. Arrowheads indicate the uropod in migrating NK cells. Small arrows indicate the advancing direction of the leading edge of the NK cell. T indicates target cell. A schematic representation of the cell movement is also shown. b, CCR are concentrated at the area of cell-cell contact in NK- b target cell conjugates. Immunofluo- rescence studies of NK-K562 conjugates were performed as described in Materials and Methods. Cells were photographed under epifluorescent for CCR2 (A) and CCR5 (D) staining and using brightfield conditions (B and E). A double exposure (brightfield and epifluorescence) is shown in C. Bar ϭ 10 ␮m.

Although MIP-1␣ and MIP-1␤ induced the uropod in a lower pro- through the uropod were rarely observed, indicating that the exis- portion of cells than did RANTES and MCP-1, these chemokines tence of two morphologic poles correlates with different functional showed a signiﬁcant effect at lower concentrations (0.01–1 ng/ml; domains. Fig. 3b). IL-15, known to be a chemotactic factor, also induced the To further assess the role of cell polarization during the NK redistribution of ICAM-3 to the uropod of NK cells and the che- cell cytotoxic response, we studied the redistribution of CCR mokine receptors to the leading edge (Fig. 3c and data not shown) and adhesion molecules during the formation of NK-target cell (26, 27, 29). Redistribution of CCR and adhesion receptors hence conjugates. The IL-2-cultured NK cells exhibited a polarized correlated with the acquisition of the migratory morphology and morphology when bound to K562 target cells (Fig. 4a). In these may guide cell migration and cell-cell interactions during NK cell immunoﬂuorescence studies, NK cells were clearly distin- effector function. guished from K562 cells based on their polarized cell shape, lower refringence, and smaller size. CCR2 and CCR5 were lo- Role of cell polarization during NK-target cell conjugate calized at the area of target-effector cell-cell contact (Fig. 4b, formation A–E). In contrast, ICAM-3 and ICAM-1 were concentrated in We next studied the role of the uropod and the leading edge in NK the uropod, at the opposite side of the cell-cell contact area (Fig. cell motility, and the formation of effector-target cell conjugates by 5a, A–D). Since K562 cells express ICAM-1, but not ICAM-2, time-lapse microscopy. IL-2-activated NK cells migrated onto ICAM-3, or VCAM-1, it is likely that NK cell polarization as FN80-coated surfaces; they show a polarized shape, with the uro- well as binding to K562 target cells are mediated through the pod at the rear of the cell and the leading edge at the opposite pole. LFA-1/ICAM-1 adhesion pathway (41). Although LFA-1 is Motile NK cells contacted the K562 target cells through the lead- evenly distributed, it shows a reinforced location at the cell-cell ing edge, maintaining their polarized shape (Fig. 4a). Contacts contact areas (Fig. 5a, insets in E and F). This was further The Journal of Immunology 3335

FIGURE 5. Redistribution of adhesion molecules in NK-target cell conjugates. a, Redistribution of ICAM-1 and -3 to the uropod. ICAM-3 (A and B) and ICAM-1 (C and D) are concentrated on the uropod of polarized NK cells bound to K562 targets, while LFA-1 is found uniformly throughout the cell (E and F). Insets in E and F show the staining of the activation-dependent epitope of the ␣-chain of LFA-1. b, Cellular distribution of the NK cell-associated receptors. Cellular distributions of CD94 (A and B), CD56 (C and D), and CD16 (E and F) on NK-target conjugates are shown. Bar ϭ 10 ␮m.

confirmed using the L16 mAb, which reacts specifically with an ac- Chemokines induce uropod-mediated recruitment of NK cells: tivation-dependent epitope on the LFA-1 ␣-chain. This Ab stained the involvement of ICAM-3 and relevance in effector-target cell cell-cell contact areas, but not the uropod region (Fig. 5a, E and F). In conjugate formation contrast, none of the NK cell-specific receptors, such as CD56, CD16 (Fc␥RIII), or CD94, accumulated at the uropod; but they were evenly The fact that the uropod concentrates several adhesion molecules in a distributed throughout the cell membrane (Fig. 5b, A–F). These re- highly exposed region of the cell prompted us to study the role of this sults show that upon target cell conjugate formation, NK cells appear structure in homotypic adhesion and NK cell recruitment. We induced to maintain a polarized shape, implying the formation of specialized cell polarization and adhesion receptor redistribution in a layer of NK functional domains on the cell membrane. cells adhered to ICAM-1-coated plates. Unstimulated NK cells were

FIGURE 6. Uropod-bearing NK cells recruit other NK cells. NK cells were allowed to bind to plastic petri dishes coated with ICAM-1-Fc and then were incubated with RANTES (10 ng/ml) for 30 min at 37°C. After addition of a second cohort of untreated NK cells, cells were ﬁlmed with a time- lapse videocassette recorder for 1.5 h. Time frames obtained from one representative experiment are shown. Arrowheads indicate cells from the second cohort (bright cells) bound to the uropod of migrating NK cells (dark cells). Small arrows indicate the direction of the leading edge of the cell. A schematic representation of the cell movement is also shown. 3336 CHEMOKINES AND NK CELL POLARIZATION

non was long lasting, persisting for at least several minutes. Quanti- tative analysis of the chemokine-induced NK cell recruitment showed that MCP-1 and RANTES clearly enhanced this phenomenon, whereas MIP-1␣ and MIP-1␤ had weaker activity, and TNF-␣ had no significant effect (Fig. 7A). The strongest response was induced by the polarization-inducing anti-ICAM-3 mAb HP2/19 (Fig. 7). The effects of MCP-1 and the anti-ICAM-3 HP2/19 mAb on cell recruitment were specifically inhibited with the anti-ICAM-3 blocking mAb 140.11, but not with the control anti-CD45 D3/9 (Fig. 7, B and C). These results show the involvement of chemokines in NK cell recruitment and the dependence of this phenomenon on ICAM-3. We next studied the role of NK cell recruitment on effector- target cell conjugate formation. NK cells adhered to FN80 were induced to polarize by MCP-1 (Fig. 8). A second cohort of non- stimulated NK cells was then added, and after 30 min, K562 cells were added, and NK cell migration was filmed. As described above, uropod-bearing NK cells tightly trapped NK cells from the second cohort, which were guided by the polarized migrating NK cells to contact the K562 cell (Fig. 8). This finding suggests that during the cytotoxic response, adhesion receptor redistribution to the uropod may exert an amplifying effect on cell conjugate formation by increasing the number of NK cells available to interact with target cells.

Prevention of cell polarization and receptor redistribution inhibits NK cell-mediated cytotoxicity We have reported that disruption of the myosin motor by specific drug inhibitors prevents cell polarization and redistribution of adhesion receptors to the cell uropod without affecting cell adhesion (42, 43). To further analyze the role of cell polarization and adhesion receptor redistribution in the cytotoxic activity of NK cells, we investigated the effects of the myosin-disrupting agent butanedione monoxime in effector-target conjugate formation as well as in NK cell-mediated lysis of K562 target cells. Butanedione monoxime, which inhibited NK cell polarization and ICAM-3 redistribution to the cell uropod (Fig. 9aB), prevented conjugate formation (Fig. 9a, F and G) and blocked NK cell-mediated cytotoxicity (Fig. 9b). In contrast, nocodazole, which disassembles FIGURE 7. Quantitative analysis of NK cell recruitment. A, Chemo- the microtubule network but did not inhibit uropod formation and kines induce the recruitment of NK cells. NK cells were allowed to bind to ICAM-1-Fc-coated petri dishes in the presence of 10 ng/ml of chemokines cell polarization (Fig. 9aC) (44), did not alter the formation of cell or TNF-␣ or 5 ␮g/ml of the uropod-triggering anti-ICAM-3 mAb HP2/19 conjugates (Fig. 9a, H and I). NK cell cytotoxicity mediated by for 30 min at 37°C. A second cohort of untreated NK cells was then added, cells with disassembled microtubules was only weakly influenced, and cells were recorded by time-lapse video for 1.5 h. The recruitment in accordance with a previous report (Fig. 9b) (45). These results index was estimated as described in Materials and Methods. The arithmetic suggest that chemokines may affect NK cell-mediated target cell mean Ϯ SEM of three independent experiments performed with cells from conjugate formation and cytotoxicity by regulating receptor redis- .p Ͻ 0.05 compared with untreated cells, as tribution and cell polarization ,ء .different donors are shown determined by Student’s t test. B and C, Involvement of ICAM-3 in NK recruitment through the uropod. NK cells were pretreated with either the blocking mAb anti-ICAM-3 140.11 or the control anti-CD45 D3/9, as in- Discussion dicated. NK cells were then allowed to bind to ICAM-1-Fc-coated petri NK cell motility and NK-target cell conjugate formation are lim- dishes in the absence or the presence of 10 ng/ml MCP-1 (B), the polar- iting steps in NK cell cytotoxic activity. Chemokines and some ization-inducing anti-ICAM-3 mAb HP2/19 (C), or the anti-MHC class I chemoattractant cytokines such as IL-15 have been described as mAb W6/32 (C) for 30 min at 37°C before addition of a second cohort of modulators of NK cell motility and cytotoxic activity (11–14, 46, NK cells. Cells were video recorded, and the recruitment index was calculated as described for Figure 5a. The arithmetic mean Ϯ SEM of three 47). When migrating, lymphocytes adopt a polarized phenotype, independent experiments performed with cells from three different donors displaying a cytoplasmic projection at the rear of the cell, termed are shown. uropod due to its similarity with ameboid morphology (20, 48). Here we show that chemokines and IL-15 also induce the polarization of NK cells bound to endothelial or extracellular matrix then added, and cell-cell interactions were videorecorded. Following proteins, a phenomenon that is accompanied by redistribution of treatment with RANTES (Fig. 6), the NK cells from the first layer adhesion receptors (ICAM-1 and ICAM-3) to the uropod. We have migrate onto the substrate (phase-dark cells), contact unstimulated also found that the chemokine receptors CCR2 and CCR5 are re- NK cells (phase-bright cells) through the uropod, and capture and distributed to the leading edge of NK cells, and that this chemokine transport them, thus moving together. These captured cells were receptor distribution correlates with the acquisition of a migratory tightly attached through the uropod, and the cell transport phenome- phenotype. Our results using time-lapse video studies confirm that The Journal of Immunology 3337

FIGURE 8. NK cells recruited by uropod-bearing NK cells are driven to contact target cells. NK cells pretreated with the chemokine MCP-1 capture other NK cells through the cellular uropod and subsequently transport them to the vicinity of target cells. NK cells were allowed to bind to plastic petri dishes coated with FN80 and then were incubated with MCP-1 (10 ng/ml) for 30 min at 37°C. After addition of a second cohort of untreated NK cells and the K562 target cells, cells were filmed with a time-lapse videocassette recorder for 1.5 h. Time frames obtained from one representative experiment are shown. Arrowheads indicate NK cells from the second cohort bound to the uropod of the migrating NK cells. Small arrows indicate the direction of the leading edge of the cell. T and E indicate target and effector cells, respectively. A schematic representation of the cell movement is also shown. uropod-bearing NK cells correspond to migrating cells, further in- leukocyte recruitment. In fact, uropod-mediated cell-cell contacts ap- dicating the association between uropod formation and leukocyte pear to recruit additional NK cells to the vicinity of the target cells, migration (20, 28, 35, 48). It therefore seems feasible that the CCR which may increase contact frequency between NK and target cells, cluster at the advancing front may function as a mechanism to favoring activation of the recruited cells. This concurs with our earlier guide the NK cell to mediate cell-cell interactions. studies of T lymphocytes, in which ICAM redistribution to the uropod Cell polarization plays an important role in the immune re- is a cooperative mechanism in lymphocyte recruitment, acting as an sponse as well as in other biologic processes (21–24, 27). For amplification system during both cell extravasation and migration to- instance, T cells recognize and bind APCs through their leading ward inflammatory foci (28, 35). On the other hand, the molecules edge (21, 23). Using time-lapse videomicroscopy, we observed a inducing NK cell polarity appear to be chemokines produced by ac- similar behavior for NK cells when they interact with the K562 tivated NK cells. A role for chemokines in the activation of NK cell- target cells; NK cells contacted the target cells through the ad- mediated cytotoxicity has also been previously reported (12, 13). In vancing front, a region in which CCR2 and CCR5 are concen- agreement with our findings, the authors found that the enhancement trated, whereas the effector-target cell interactions through the uro- of NK cytotoxicity mediated by chemokines was dependent on con- pod were rarely observed. We also found that during the formation jugate formation (12). We have found that the formation of NK-target of effector-target cell conjugates, ICAM-1 and ICAM-3 adhesion cell conjugates induces the release of chemokines, which are respon- molecules are concentrated in the uropod of NK cells, which is sible for NK cell migration, a finding that further substantiates our located at the most distal point from the area of NK-target cell hypothesis of a cooperative mechanism of NK cell migration to the contact. The redistribution of adhesion molecules and CCR to the target. Some of these events, including cell polarization and the in- opposite poles of NK further supports the idea that cell polariza- duction of cytotoxic activity, can be blocked by ADP ribosylation of tion arises in two specialized regions on the cell membrane, which the GTP binding protein RhoA, indicating the key role of this che- make feasible several key phenomena, such as cell migration and mokine receptor-associated signaling pathway (49). In summary, intercellular adhesion. The functional relevance of cell polarization these observations are consistent with the view of the NK cell as a is made evident by the fact that both CCR and adhesion receptor sensitive immune cell, responsive to external cues and able to alter its redistribution are maintained not only during NK cell migration, shape in response to diverse stimuli. but also when they are bound to their targets. Furthermore, pre- As occurs in vitro, it is also conceivable that the in vivo inter- venting cell polarization and adhesion receptor redistribution action of NK with target cells may induce the release of soluble blocked the formation of effector-target cell conjugates and NK factors, including chemokines probably produced by effector cells, cell-mediated cytotoxicity, thus demonstrating the functional im- which can trigger NK cell polarization, chemotaxis, and recruit- portance of these phenomena. ment of additional NK cells to the tissue. It is thus entirely feasible The mechanisms by which NK cells promote cytotoxic activity in that the in vivo recruitment of NK cells triggered by chemokines specific tissues remains poorly understood. We propose the existence and cytokines at sites of immune activation has an important am- of a cooperative mechanism involved in NK cell migration. On the plifying effect in the elimination of tumor cells and parasites. The one hand, by inducing cell polarization, chemokines and cytokines induction of NK cytotoxicity by IL-15 secreted during human her- play important roles in the development of the NK cell cytotoxic pes virus infection may be an example of such a mechanism (47). response. Cell polarization determines the formation of different func- IL-15 acts as a chemotactic factor that induces cell polarization tional domains in the NK cell. The leading edge, which concentrates (27) and adhesion receptor distribution (29) and regulates the mi- the CCR, is involved in target cell adhesion, granule release during gration of lymphocytes at inflammation sites in diseases such as cytotoxic phenomena (25), and probably directing the cell during mi- rheumatoid arthritis (50). Our data suggest that chemokines have a gration, whereas the uropod, which accumulates ICAM, is involved in similar in vivo role in NK cell-mediated phenomena. 3338 CHEMOKINES AND NK CELL POLARIZATION

Acknowledgments We thank Dr. J. P. Albar for help in peptide design for the production of anti-CCR5, C. Mark for editorial assistance, C. Caban˜as and M. Montoya for assistance with the time-lapse videomicroscopy studies, and R. Tejedor and M. Vito´n for technical support.

References 1. Trinchieri, G. 1989. Biology of natural killer cells. Adv. Immunol. 47:187. 2. Somersalo, K., J. Tarkkanen, M. Patarroyo, and E. Saksela. 1992. Involvement of ␤-2 integrins in the migration of natural killer cells. J. Immunol. 149:590. 3. Bender, J. R., R. Pardi, M. A. Karasek, and E. G. Engleman. 1987. Phenotypic and functional characterization of lymphocytes that bind human microvascular endothelial cells in vitro: evidence for preferential binding of natural killer cells. J. Clin. Invest. 79:1679. 4. Allavena, P., C. Paganin, I. Martin-Padura, G. Peri, M. Gaboli, E. Dejana, P. C. Marchisio, and A. Mantovani. 1991. Molecules and structures involved in the adhesion of natural killer cells to vascular endothelium. J. Exp. Med. 173:439. 5. Gismondi, A., S. Morrone, M. J. Humphries, M. Piccoli, L. Fragi, and A. Santoni. 1991. Human NK cells express VLA-4 and VLA-5 which mediate their adhesion to fibronectin. J. Immunol. 146:384. 6. Baggiolini, M., B. Dewald, and B. Moser. 1997. Human chemokines: an update. Annu. Rev. Immunol. 15:675. 7. Schall, T. J., and K. B. Bacon. 1994. Chemokines, leukocyte trafficking, and inflammation. Curr. Opin. Immunol. 6:865. 8. Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte emigration: the multistep paradigm. Cell 76:301. 9. Oppenheim, J. J., C. O. Zacharie, N. Mukaida, and K. Matsushima. 1991. Prop- erties of the novel proinflammatory supergene “intercrine” cytokine family. Annu. Rev. Immunol. 9:617. 10. Rollins, B. J. 1997. Chemokines. Blood 90:909. 11. Maghazachi, A. A., A. Al-Aoukaty, and T. J. Schall. 1994. C-C chemokines induce the chemotaxis of NK and IL-2-activated NK cells. J. Immunol. 153:4969. 12. Taub, D. D., T. J. Sayers, C. R. D. Carter, and J. R. Ortaldo. 1995. ␣ and ␤ chemokines induce NK cell migration and enhance NK-mediated cytolysis. J. Im- munol. 155:3877. 13. Loetscher, P., M. Seitz, I. Clark-Lewis, M. Baggiolini, and B. Moser. 1996. Activation of NK cells by CC chemokines. J. Immunol. 156:322. 14. Allavena, P., G. Bianchi, D. Zhou, J. van Damme, P. Jıˆlek, S. Sozzani, and A. Mantovani. 1994. Induction of natural killer cell migration by monocyte chemotactic protein-1, -2 and -3. Eur. J. Immunol. 24:3233. 15. Somersalo, K., O. Carpeń, and E. Saksela. 1994. Stimulated natural killer cells secrete factors with chemotactic activity, including NAP-1/IL-8, which supports VLA-4 and VLA-5-mediated migration of T lymphocytes. Eur. J. Immunol. 24: 2957. 16. Bluman, E. M., K. J. Bartynski, B. R. Avalos, and M. A. Caligiuri. 1996. Human natural killer cells produce abundant macrophage inflammatory protein-1␣ in response to monocyte-derived cytokines. J. Clin. Invest. 12:2722. 17. Hedrick, J. A., V. Saylor, D. Figueroa, L. Mizoue, Y. Xu, S. Menon, J. Abrams, T. Handel, and A. Zlotnik. 1997. Lymphotactin is produced by NK cells and attracts both NK cells and T cells in vivo. J. Immunol. 158:1533. 18. Mackay, C. R. 1996. Chemokine receptors and T cell chemotaxis. J. Exp. Med. 184:799. 19. Drubin, G. D., and W. J. Nelson. 1996. Origins of cell polarity. Cell 84:335. 20. Wilkinson, P. C. 1986. The locomotor capacity of human lymphocytes and its enhancement by cell growth. Immunology 57:281. 21. Negulescu, P. A., T. A. Krasieva, A. Khan, H. H. Kerschbaun, and FIGURE 9. Inhibition of conjugate formation and NK cell cytotoxicity by M. D. Cahalan. 1996. Polarity of T cell shape, motility, and sensitivity to antigen. a myosin-disrupting drug. a, Conjugate formation is inhibited by disruption of Immunity 4:421. 22. Donnadieu, E., G. Bismuth, and A. Trautmann. 1994. Antigen recognition by myosin-motor. A to C correspond to immunofluorescence studies showing helper T cells elicits a sequence of distinct changes of the shape and intracellular ICAM-3 distribution and cell shape morphology of RANTES-stimulated NK calcium. Curr. Biol. 4:584. cells, untreated (A) or pretreated for 30 min with either 20 mM butanedione 23. Kupfer, A., and S. J. Singer. 1989. Cell biology of cytotoxic and helper T cell monoxime (B)or10␮M nocodazole (C). The effects of these cytoskeleton- functions: immunofluorescence microscopic studies of single cells and cell cou- disrupting agents on NK-target cell conjugate formation are shown in D to I. ples. Annu. Rev. Immunol. 7:309. 24. Helander, T. S., O. Ca´rpen, O. Turunen, P. E. Kovanen, A. Vaheri, and IL-2-cultured NK cells were incubated for 30 min in the absence (D and E)or T. Timonen. 1996. ICAM-2 redistributed by ezrin as a target for killer cells. the presence of 20 mM butanedione monoxime (F and G)or10␮M nocoda- Nature 382:265. zole (H and I). Then, cells were allowed to adhere to FN for 30 min at 37°C, 25. Lang, P., L. Guizani, I. Vitte-Mony, R. Stancou, O. Dorseuil, G. Gacon, and K562 target cells were added, and the cytotoxic process was video recorded for J. Bertoglio. 1992. ADP-ribosylation of the ras-related, GTP binding protein Rho-A inhibits lymphocyte-mediated cytotoxicity. J. Biol. Chem. 267:11677. 1 h. Two sequential representative video frames of each experimental condi- 26. Nieto, M., J. M. R. Frade, D. Sancho, M. Mellado, C. Martinez-A, and F. tion are shown. To illustrate, some targets (T), polarized NK cells bound to the Sańchez-Madrid. 1997. Polarization of chemokine receptors to the leading edge targets (arrowheads), or effector cells (E) are indicated. Note the round shape during lymphocyte chemotaxis. J. Exp. Med. 18:153. of NK cells treated with butanedione monoxime (F and G). b, Effect of the 27. Wilkinson, P. C., and F. Y. Liew. 1995. Chemoattraction of human blood T lymphocytes by interleukin-15. J. Exp. Med. 181:1255. myosin disruption drug butanedione monoxime on NK cell cytotoxicity. NK 28. del Pozo, M. A., P. Sańchez-Mateos, M. Nieto, and F. Sańchez-Madrid. 1995. cells were incubated in the absence or the presence of 20 mM butanedione Chemokines regulate cellular polarization and adhesion receptor redistribution monoxime, 10 ␮M nocodazole, or control dissolvent for 30 min at 37°C. Then, during lymphocyte interaction with endothelium and extracellular matrix: in- NK cell killing activity was determined in a 4-h 51Cr release assay, as de- volvement of cAMP signaling pathway. J. Cell Biol. 131:495. Ϯ 29. Nieto, M., M. A. del Pozo, and F. Sańchez-Madrid. 1996. Interleukin-15 induces scribed in Materials and Methods. The arithmetic mean SD of triplicate adhesion receptor redistribution in T lymphocytes. Eur. J. Immunol. 26:1302. determinations is shown. A representative experiment of three performed with 30. Campanero, M. R., M. A. del Pozo, A. G. Arroyo, P. Sańchez-Mateos, T. Her- cells from different donors is shown. nańdez Caselles, A. Craig, R. Pulido, and F. Sańchez-Madrid. 1993. ICAM-3 The Journal of Immunology 3339

interacts with LFA-1 and regulates the LFA-1/ICAM-1 cell adhesion pathway. 40. Polentarutti, N., P. Allavena, G. Bianchi, G. Giardina, A. Basile, S. Sozzani, J. Cell Biol. 123:1007. A. Mantovani, and M. Introna. 1997. IL-2 regulated expression of the monocyte 31. Pulido, R., M. Cebriań, A. Acevedo, M. O. de Landa´zuri, and F. Sańchez-Madrid. derived chemotactic protein-1 receptors CCR-2 in human NK cells: character- 1988. Comparative biochemical and tissue distribution study of four distinct ization of a predominant 3.4-kilobase transcript containing CCR2B and CCR2A CD45 antigen specificities. J. Immunol. 140:3851. sequences. J. Immunol. 158:2689. 32. Aramburu, J., M. A. Balboa, A. Ramı´rez, A. Silva, A. Acevedo, F. Sańchez- 41. Schmidt, R. E., G. Bartley, H. Levine, S. F. Schlossman, and J. Ritz. 1985. Madrid, M. O. de Landa´zuri, and M. Lo´pez-Botet. 1990. A novel functional cell ϩ Functional characterization of LFA-1 antigens in the interaction of human NK surface dimer (Kp43) expressed by natural killer cells and T cell receptor-␥/␦ T clones and target cells. J. Immunol. 135:1020. lymphocytes: inhibition of IL-2 dependent proliferation by anti-Kp43 monoclonal 42. Campanero, M. R., P. Sańchez-Mateos, M. A. del Pozo, and F. Sańchez-Madrid. antibody. J. Immunol. 147:714. 1994. ICAM-3 regulates lymphocyte morphology and integrin-mediated T cell 33. Cassatella, M. A., M. C. Anegoń, M. C. Cuturi, P. Griskey, G. Trinchieri, and interaction with endothelial cell and extracellular matrix ligands. J. Cell Biol. ␥ 2ϩ B. Perussia. 1989. Fc R (CD16) interaction with ligand induces Ca mobiliza- 127:867. tion and phosphoinositide turnover in human natural killer cells: role of Ca2ϩ in Fc␥R (CD16)-induced transcription and expression of lymphokine genes. J. Exp. 43. Sańchez-Mateos, P., M. R. Campanero, M. A. del Pozo, and F. Sańchez-Madrid. Med. 169:549. 1995. Regulatory role of CD43 leukosialin on integrin-mediated T cell adhesion 34. Berendt, A. R., A. McDowall, A. G. Craig, P. A. Bates, M. J. E. Sternberg, to endothelial and extracellular matrix ligands and its polar redistribution to a K. Marsh, C. I. Newbold, and N. Hogg, 1992. The binding site on ICAM-1 for cellular uropod. Blood 86:2228. Plasmodium falciparum-infected erythrocytes overlaps, but is distinct from the 44. Ratner, S., W. S. Sherrod, and D. Lichlyter. 1997. Microtubule retraction into the LFA-1 binding site. Cell 14:9. uropod and its role in T cell polarization and motility. J. Immunol. 159:1063. 35. del Pozo, M. A., C. Cabanãs, M. C. Montoya, A. Ager, P. Sańchez-Mateos, and 45. Carpeń, O. 1987. The role of microtubules in human natural killer cell-mediated F. Sańchez-Madrid. 1997. ICAMs redistributed by chemokines to cellular uro- cytotoxicity. Cell. Immunol. 106:376. pods as a mechanism for recruitment of T lymphocytes. J. Cell Biol. 137:493. 46. Warren, H. S., B. F. Kinnear, R. L. Kastelein, and L. L. Lanier. 1996. Analysis 36. Charo, I. F., S. J. Myers, A. Herman, C. Fraci, A. J. Connolly, and S. R. Coughlin. of the costimulatory role of IL-2 and IL-15 in initiating proliferation of resting 1994. Molecular cloning and functional expression of two monocyte chemoat- (CD56dim) human NK cells. J. Immunol. 156:3254. tractant protein 1 receptors reveals alternative splicing of the carboxyl-terminal 47. Flamand, L., I. Stefanescu, and J. Menezes. 1996. Human herpes virus-6 en- tails. Proc. Natl. Acad. Sci. USA 91:2752. hances natural killer cell cytotoxicity via IL-15. J. Clin. Invest. 6:1373. 37. Raport, C. J., J. Gosling, V. L. Schweickart, P. W. Gray, and I. F. Charo. 1996. Molecular cloning and functional characterization of a novel human CC chemo- 48. Haston, W. S., J. M. Shields, and P. C. Wilkinson. 1982. Lymphocyte locomotion kine receptor (CCR5) for RANTES, MIP-1␣, and MIP-1␤. J. Biol. Chem. 271: and attachment on two dimensional surfaces and in three-dimensional matrices. 17161. J. Cell Biol. 92:747. 38. Suzuki, H., G. S. Duncan, H. Takimoto, and T. W. Mak. 1997. Abnormal de- 49. Lang, P., F. Gesbert, M. Delespine-Carmagnat, R. Stancou, M. Pouchelet, and velopment of intestinal intraepithelial lymphocytes and peripheral natural killer J. Bertoglio. 1996. Protein kinase A phosphorylation of RhoA mediates the mor- cells in mice lacking the IL-2 receptor ␤ chain. J. Exp. Med. 185:499. phological and functional effects of cyclic AMP in cytotoxic lymphocytes. EMBO 39. Ohteki, T. H. Yoshida, T. Matsuyama, G. S. Duncan, T. W. Mak, and J. 15:510. P. S. Ohashi. 1997. The transcription factor interferon regulatory factor 1 (IRF-1) 50. McInnes, I. B., B. P. Leung, R. D. Sturrock, M. Field, and F. Y. Liew. 1997. is important during the maturation of natural killer cells, and intestinal intraepi- Interleukin-15 mediates T cell-dependent regulation of tumor necrosis factor-␣ thelial T cells. J. Exp. Med. 6:9. production in rheumatoid arthritis. Nat. Med. 3:189.