The migration of human natural killer cells
Eur. J. Immunol. 1991. 21: 35-42
Kristina Somersalo and Eero Saksela DeDartment of Patholoev, Uiversity of Helsinki, Helsinki
35
Fibronectin facilitates the migration of human natural killer cells* The interaction of lymphocytes with the extracellular matrix plays an important role in the immune defence-against tumor cells and virus-infected cells. We have examined the effect of matrix proteins on the migration of large granular lymphocytes (LGL) through 3-pm pores in Nuclepore filters in a Boyden invasion chamber. Fibronectin bound on the filter surface significantly increased (p < 0.001) the capacity of LGL to migrate, whereas soluble fibronectin did not. In addition, a significantly higher (p < 0.001) percentage of LGL was capable of migration through fibronectin-coated filters than through untreated filters. With fibronectin-coated filters, a strong enrichment of CD16+ and CD56+CD3- cells with LGL morphology and reduction of CD3+ cells was found among migrating cells when the incubation time was 4 h or less. Later agranular lymphocytes, mainly CD3+ T lymphocytes, also started to migrate. Laminin coating of filters also facilitated migration, and when filters were coated with both fibronectin and laminin the increase in migration was equal to the sum of the increases induced by each protein alone. Interactions between cell surface and the Arg-Gly-Asp (RGD) peptide sequence of many matrix proteins had no role in the LGL migration through untreated flters. However,when filters were coated with either fibronectin or laminin, or with both, peptide containing the RGD sequence reduced migration to the level of untreated filters, whereas an Arg-Gly-Glu control peptide had no effect. Our results show that unstimulated LGLhatural killer cells are capable of rapid migration through matrix-coated porous membranes, and that interactions between cell surface receptors and the RGD sequence of fibronectin and probably laminin are utilized in this process.
1 Introduction
fibronectin and laminin [6]. When migrating towards a tumor or infected target cells, lymphocytes have to crawl NK cells have large granular lymphocyte (LGL) morpho- along these matrix proteins. For this process lymphocytes logy and they comprise 2%-12% of blood lymphocytes [ 11. need adhesion receptors, which do not anchor them. but They are non-T, non-B lymphocytes, and do not exhibit allow reversible binding. A migrating cell is supposed to phagocytic activity. NK cells are considered to function in attach to the matrix protein with receptors at its leading the first line of defence against cancer and many virus- edge, and to dissolve the contracts at the trailing end of the infected cells [2,3], and are able to recognize the target cell [7]. The cell surface receptors which mediate matrix cells without prior sensitization in a non-MHC-restricted binding are primarily and p 3 integrins; the f i ~subunit is manner. Almost all NK cells have FcR (CD16), which also associated with one of at least six different a subunits. while enable them to kill via a specific antibody-dependent p 3 subunit associates with one of the two a subunits. Thus, cell-mediated cytotoxicity (ADCC). Nearly all CD16+ cells at least eight integrin subtypes exist, which have different also express CD56+ antigen [4]. The spontaneous NK selectivities for distinct matrix proteins. At least two activity is, thus, mediated by CD16+ and CD56+CD3- membersof the family [8,9] and both p 3 integrins [8, 101 recognize the cell-binding sequence Arg-Gly-Asp (RGD) cells. present in fibronectin [ l l , 121 and many other matrix To reach the tissue sites of defence, the circulating NK cells proteins [13-161. In addition to RGD, the presence of other have to penetrate through the endothelium, the basement cell-adhesion sites within the fibronectin molecule have membrane around the capillary and the matrix in the been reported [ 17-20]. Several cells produce slightly differperivascular tissue space. The major components of base- ent types of fibronectin [21], which have distinct adhesion ment membranes are collagen IVand laminin [5], while the sites and are recognized by different cells [ 191. Laminin also matrix is composed of collagen (I-IV), proteoglycans, has at least two cell-adhesion sites [22].Thus, extracellular matrixes of different composition and cells with different fiI and p3 integrins provide the basis for the selective sorting of lymphocytes to their various tissue destinations. It is. [I 85811 therefore, important to know which proteins in the matrix and which cell surface receptors are preferred when N K * This work was supported by grants from the Finnish Cultural cells perform the early phase of host defence. Foundation, the Finnish Cancer Society and the Paulo Foundation. Previous unpublished findings from this laboratory (J. Tarkkanen) have shown that unstimulated LGL separate Correspondence: Kristina Somersalo, Department of Pathology, University of Helsinki, Haartmaninkatu 3, SF-0290 Helsinki, Fin- from the suspension of blood mononuclear cells by migrating through untreated porous polycarbonate filters. In this land study we show that this process is greatly facilitated by fibronectin and to a lesser extent by laminin coating of the Abbreviation: LGL: Large granular lymphocytes 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991
36
K. Somersalo and E. Saksela
filters. LGL use their cell surface fibronectin receptor and the RGD-binding site in the migration.
Eur. J. lmrnunol. 1991. 21: 35-42 hyde and dehydrated in a series of ethanol followed by critical-point drying.The filters were covered with gold and examined in a Jeol JSM-820 SEM.
2 Materials and methods 2.1 Lymphocyte purification
2.4 Immunostaining
Buffy coats from healthy blood donors were obtained from the Finnish Red Cross Transfusion Service. Mononuclear cells were isolated by Ficoll-Isopaque (Pharmacia Fine Chemicals AB, Uppsala, Sweden) gradient centrifugation and subsequent filtration through nylon wool columns. LGL were further enriched by discontinuous four-step density gradient centrifugation on Percoll (Pharmacia) as previously described [23] with minor modifications.The cell composition was defined from Giemsa-stained cytocentrifuge smears.
Polycarbonate filters bearing the spread cells were treated with anti-CD16 mAb (Janssen Chimica. Beerse, Belgium) and as control normal mouse IgG including all subclasses (Coulter Immunology, Hialeah, FL) for 45 min at 4"C.The filtes were washed and exposed to FITC-labeled F(ab'1 fragments of goat anti-mouse IgG (Cappel Laboratories, Melvren, PA) for 45 min at 4°C. After washing, the cells were fixed with 1% paraformaldehyde and analyzed by fluorescence microscopy.
2.2 Migration assay
2.5 FCM
A Boyden chamber similar to that described by Axelsson et al. [24] was used. Polycarbonate filters with 3-pm pores (Nuclepore Corporation, Pleasanton, CA) were placed between the upper and the lower compartment of the chamber. The filters were either left untreated or coated overnight usually with 20 pg/ml, but where indicated with 5. 10, 20, 40 or 80pg/ml, human plasma fibronectin (courtesy of Dr. T. Vartio of this institute) purified by affinity chromatography on gelatin-Sepharose [25], or with 20 pg/ml laminin (Collaborative Research, Bedford, MA), 20 pg/ml collagen I , 20 pg/ml collagen IV (from Prof. I. Virtanen, Department of Anatomy, University of Helsinki) or both fibronectin and laminin.
For cell surface marker analysis, 2 x lo5 lymphocytes were suspended in PBS. mAb antLCD16 (Janssen), anti-CD3. anti-CD8, antLCD56 (Becton Dickinson, Mountain View. CA), anti-y/6 (from Dr. J.Tarkkanen, Dr. L. Lanier. Becton Dickinson Monoclonal Center) and as control normal mouse IgG including all subclasses (Coulter Immunology) were added (from 1 : 10-1 : 40 vol : vol) and incubated for 45 min at 4°C. The cells were washed and resuspended in FITC-labeled F(ab')2 fragments of goat anti-mouse IgG (Cappel) for 45 min at 4°C. After washing the cells were analyzed for fluorescence with a FACScan (Becton Dickinson).
The microcentrifuge tubes in the lower compartment of the chamber were filled with 400 pl RPMI 1640 (Gibco, Grand Island, NY) supplemented with 5% humanAB serum (Finnish Red Cross Transfusion Service) from which fibronectin had been depleted by absorption on a gelatinSepharose (Pharmacia) column at room temperature in PBS. Two hundred microliters of lymphocyte suspension (10 x loh cells/ml) in the same buffer was added to the upper compartment. In chemotaxis experiments 0, 5 , 10, 20, 40, 80 and 160 pg/ml of fibronectin was added to the lower compartment. Where indicated, 0.5 mg/ml of pentapeptide Gly-Arg-Gly-Asp-Ser (GRGDS) or control pentapeptide Gly-Arg-Gly-Glu-Ser (GRGES; produced by the peptide synthesizer at the Department of Biochemistry, University of Helsinki) was added to the lymphocyte suspension 30 min before experiments. After incubation in humidified air with 5% of CO2 at 37°C for 1 to 24 h, the cells which had migrated through the filtes into the lower compartment of the Boyden chamber were counted, and the cell composition was determined from Giemsa-stained cytocentrifuge smears or by immune phenotyping (see below). The cells which had retained on the Boyden chamber filters after migration were in some experiments trypsinized briefly, counted and the cell composition determined from cytocentrifuge smears.
2.6 Cell culture conditions The target cells K-562 [26], Raji [27] and COLO [28] were obtained from American Type Culture Collection, Rockville, MD, and cultured in RPMI 1640 (Gibco) supplemented with 10% heat-inactivated FCS (Gibco), 0.29 mg/ml L-glutamine (Gibco), 100 IU/ml penicillin, and 10 pglml streptomycin in a humidified air atmosphere with 5% COz at 37 "C. 2.7 Cytotoxicity assays Target cells ( I x loh cells) were labeled with 20 pCi of sodium 51Cr(Radiochemical Centre, Amersham, GB).The cell count was adjusted to 1 X 10s/ml. Various concentrations of effector cells were added in 100 pl volume to 100 pl of target cells to give ratios of 20: 1, 10 : 1 , 5 : 1,2.5 : 1 and 1.25 : 1 in microtitration wells (Titertek; Flow Labs.. Irvine, Scotland). After 4 or 18 h of incubation at 37"C, 100 pl of SN from each well was collected and counted in a gamma counter (Wallac, Turku, Finland). The percentage of radioactivity released was calculated as follows: % Rclcasc
=
W - S )
100
(TOT - S )
2.3 Scanning electron microscopy (SEM) For SEM analysis the cells were incubated for 2 h in the Boyden chamber. The filters were then fixed in glutaralde-
where EX: cprn released from experimental cells; S: cpm released from cells in the medium control during incubation; and TOT: total cpm incorporation into cells.
The migration of human natural killer cells
Eur. J. Immunol. 1991. 21: 35-42
2.8 Statistical methods To evaluate the statistical significance of the cell migration assay results the paired t-test was used.
3 Results 3.1 LGL morphology on the filters Highly purified LGL (containing 78.8 +. 5.6% of CD16+ cells) were placed on either untreated or fibronectin-coated polycarbonate filters in Boyden invasion chambers. Immunostaining results revealed that the cells which attached and spread on the filters were CD16+ cells.The cells which were in the process of migration through the 3-pm pores of untreated filter were of round and spread forms (Fig. 1A). The cells on the fibronectin-coated filters (Fig. 1B) were found to have many more microspikes than cells on the
37
untreated filters and the spreading was more complete. Lamellipodia typical of migrating cells were seen among the cells on both uncoated and fibronectin-coated filters. The cells which had migrated through the filter remained for a short while attached in spread form on the lower surface of the filter before falling down t o the lower compartment of the chamber. Microvilli density on the surface of LGL decreased markedly among the cells which had migrated through the filter (not shown). 3.2 Effect of fibronectin coating in adhesion and migration
The effect of increasing concentrations of fibronectin coating on filters was examined. Human Ficoll- and Percoll-fractionated lymphocytes (58.0 k 4.2% of LGL) freshly isolated from blood were placed on the filters in the Boyden chambers. The number of cells which migrated through filters into the lower compartment of the chamber during incubation was counted. Cell composition in the lower compartment was determined from cytocentrifuge smears by morphological criteria, and the proportion of migrated LGL from the original input placed on the filters was calculated. As shown in Fig. 2 both adhesion on the filter and migration through the filter were found t o be dependent upon the concentration of the fibronectin coating. The increase in migration produced by coating concentrations between 10 and 20 pg/ml was much more prominent than the increase in adhesion. The numbers of cells which migrated through untreated or fibronectin-coated filters into the lower compartment of the chamber during different incubation times were also compared. Significantly more (p < 0.001) cells migrated through the fibronectin-coated filters than through the untreated filters (Fig. 3A), and a significantly higher @
Eur. J. Immunol. 1991. 21: 35-42
Kristina Somersalo and Eero Saksela DeDartment of Patholoev, Uiversity of Helsinki, Helsinki
35
Fibronectin facilitates the migration of human natural killer cells* The interaction of lymphocytes with the extracellular matrix plays an important role in the immune defence-against tumor cells and virus-infected cells. We have examined the effect of matrix proteins on the migration of large granular lymphocytes (LGL) through 3-pm pores in Nuclepore filters in a Boyden invasion chamber. Fibronectin bound on the filter surface significantly increased (p < 0.001) the capacity of LGL to migrate, whereas soluble fibronectin did not. In addition, a significantly higher (p < 0.001) percentage of LGL was capable of migration through fibronectin-coated filters than through untreated filters. With fibronectin-coated filters, a strong enrichment of CD16+ and CD56+CD3- cells with LGL morphology and reduction of CD3+ cells was found among migrating cells when the incubation time was 4 h or less. Later agranular lymphocytes, mainly CD3+ T lymphocytes, also started to migrate. Laminin coating of filters also facilitated migration, and when filters were coated with both fibronectin and laminin the increase in migration was equal to the sum of the increases induced by each protein alone. Interactions between cell surface and the Arg-Gly-Asp (RGD) peptide sequence of many matrix proteins had no role in the LGL migration through untreated flters. However,when filters were coated with either fibronectin or laminin, or with both, peptide containing the RGD sequence reduced migration to the level of untreated filters, whereas an Arg-Gly-Glu control peptide had no effect. Our results show that unstimulated LGLhatural killer cells are capable of rapid migration through matrix-coated porous membranes, and that interactions between cell surface receptors and the RGD sequence of fibronectin and probably laminin are utilized in this process.
1 Introduction
fibronectin and laminin [6]. When migrating towards a tumor or infected target cells, lymphocytes have to crawl NK cells have large granular lymphocyte (LGL) morpho- along these matrix proteins. For this process lymphocytes logy and they comprise 2%-12% of blood lymphocytes [ 11. need adhesion receptors, which do not anchor them. but They are non-T, non-B lymphocytes, and do not exhibit allow reversible binding. A migrating cell is supposed to phagocytic activity. NK cells are considered to function in attach to the matrix protein with receptors at its leading the first line of defence against cancer and many virus- edge, and to dissolve the contracts at the trailing end of the infected cells [2,3], and are able to recognize the target cell [7]. The cell surface receptors which mediate matrix cells without prior sensitization in a non-MHC-restricted binding are primarily and p 3 integrins; the f i ~subunit is manner. Almost all NK cells have FcR (CD16), which also associated with one of at least six different a subunits. while enable them to kill via a specific antibody-dependent p 3 subunit associates with one of the two a subunits. Thus, cell-mediated cytotoxicity (ADCC). Nearly all CD16+ cells at least eight integrin subtypes exist, which have different also express CD56+ antigen [4]. The spontaneous NK selectivities for distinct matrix proteins. At least two activity is, thus, mediated by CD16+ and CD56+CD3- membersof the family [8,9] and both p 3 integrins [8, 101 recognize the cell-binding sequence Arg-Gly-Asp (RGD) cells. present in fibronectin [ l l , 121 and many other matrix To reach the tissue sites of defence, the circulating NK cells proteins [13-161. In addition to RGD, the presence of other have to penetrate through the endothelium, the basement cell-adhesion sites within the fibronectin molecule have membrane around the capillary and the matrix in the been reported [ 17-20]. Several cells produce slightly differperivascular tissue space. The major components of base- ent types of fibronectin [21], which have distinct adhesion ment membranes are collagen IVand laminin [5], while the sites and are recognized by different cells [ 191. Laminin also matrix is composed of collagen (I-IV), proteoglycans, has at least two cell-adhesion sites [22].Thus, extracellular matrixes of different composition and cells with different fiI and p3 integrins provide the basis for the selective sorting of lymphocytes to their various tissue destinations. It is. [I 85811 therefore, important to know which proteins in the matrix and which cell surface receptors are preferred when N K * This work was supported by grants from the Finnish Cultural cells perform the early phase of host defence. Foundation, the Finnish Cancer Society and the Paulo Foundation. Previous unpublished findings from this laboratory (J. Tarkkanen) have shown that unstimulated LGL separate Correspondence: Kristina Somersalo, Department of Pathology, University of Helsinki, Haartmaninkatu 3, SF-0290 Helsinki, Fin- from the suspension of blood mononuclear cells by migrating through untreated porous polycarbonate filters. In this land study we show that this process is greatly facilitated by fibronectin and to a lesser extent by laminin coating of the Abbreviation: LGL: Large granular lymphocytes 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991
36
K. Somersalo and E. Saksela
filters. LGL use their cell surface fibronectin receptor and the RGD-binding site in the migration.
Eur. J. lmrnunol. 1991. 21: 35-42 hyde and dehydrated in a series of ethanol followed by critical-point drying.The filters were covered with gold and examined in a Jeol JSM-820 SEM.
2 Materials and methods 2.1 Lymphocyte purification
2.4 Immunostaining
Buffy coats from healthy blood donors were obtained from the Finnish Red Cross Transfusion Service. Mononuclear cells were isolated by Ficoll-Isopaque (Pharmacia Fine Chemicals AB, Uppsala, Sweden) gradient centrifugation and subsequent filtration through nylon wool columns. LGL were further enriched by discontinuous four-step density gradient centrifugation on Percoll (Pharmacia) as previously described [23] with minor modifications.The cell composition was defined from Giemsa-stained cytocentrifuge smears.
Polycarbonate filters bearing the spread cells were treated with anti-CD16 mAb (Janssen Chimica. Beerse, Belgium) and as control normal mouse IgG including all subclasses (Coulter Immunology, Hialeah, FL) for 45 min at 4"C.The filtes were washed and exposed to FITC-labeled F(ab'1 fragments of goat anti-mouse IgG (Cappel Laboratories, Melvren, PA) for 45 min at 4°C. After washing, the cells were fixed with 1% paraformaldehyde and analyzed by fluorescence microscopy.
2.2 Migration assay
2.5 FCM
A Boyden chamber similar to that described by Axelsson et al. [24] was used. Polycarbonate filters with 3-pm pores (Nuclepore Corporation, Pleasanton, CA) were placed between the upper and the lower compartment of the chamber. The filters were either left untreated or coated overnight usually with 20 pg/ml, but where indicated with 5. 10, 20, 40 or 80pg/ml, human plasma fibronectin (courtesy of Dr. T. Vartio of this institute) purified by affinity chromatography on gelatin-Sepharose [25], or with 20 pg/ml laminin (Collaborative Research, Bedford, MA), 20 pg/ml collagen I , 20 pg/ml collagen IV (from Prof. I. Virtanen, Department of Anatomy, University of Helsinki) or both fibronectin and laminin.
For cell surface marker analysis, 2 x lo5 lymphocytes were suspended in PBS. mAb antLCD16 (Janssen), anti-CD3. anti-CD8, antLCD56 (Becton Dickinson, Mountain View. CA), anti-y/6 (from Dr. J.Tarkkanen, Dr. L. Lanier. Becton Dickinson Monoclonal Center) and as control normal mouse IgG including all subclasses (Coulter Immunology) were added (from 1 : 10-1 : 40 vol : vol) and incubated for 45 min at 4°C. The cells were washed and resuspended in FITC-labeled F(ab')2 fragments of goat anti-mouse IgG (Cappel) for 45 min at 4°C. After washing the cells were analyzed for fluorescence with a FACScan (Becton Dickinson).
The microcentrifuge tubes in the lower compartment of the chamber were filled with 400 pl RPMI 1640 (Gibco, Grand Island, NY) supplemented with 5% humanAB serum (Finnish Red Cross Transfusion Service) from which fibronectin had been depleted by absorption on a gelatinSepharose (Pharmacia) column at room temperature in PBS. Two hundred microliters of lymphocyte suspension (10 x loh cells/ml) in the same buffer was added to the upper compartment. In chemotaxis experiments 0, 5 , 10, 20, 40, 80 and 160 pg/ml of fibronectin was added to the lower compartment. Where indicated, 0.5 mg/ml of pentapeptide Gly-Arg-Gly-Asp-Ser (GRGDS) or control pentapeptide Gly-Arg-Gly-Glu-Ser (GRGES; produced by the peptide synthesizer at the Department of Biochemistry, University of Helsinki) was added to the lymphocyte suspension 30 min before experiments. After incubation in humidified air with 5% of CO2 at 37°C for 1 to 24 h, the cells which had migrated through the filtes into the lower compartment of the Boyden chamber were counted, and the cell composition was determined from Giemsa-stained cytocentrifuge smears or by immune phenotyping (see below). The cells which had retained on the Boyden chamber filters after migration were in some experiments trypsinized briefly, counted and the cell composition determined from cytocentrifuge smears.
2.6 Cell culture conditions The target cells K-562 [26], Raji [27] and COLO [28] were obtained from American Type Culture Collection, Rockville, MD, and cultured in RPMI 1640 (Gibco) supplemented with 10% heat-inactivated FCS (Gibco), 0.29 mg/ml L-glutamine (Gibco), 100 IU/ml penicillin, and 10 pglml streptomycin in a humidified air atmosphere with 5% COz at 37 "C. 2.7 Cytotoxicity assays Target cells ( I x loh cells) were labeled with 20 pCi of sodium 51Cr(Radiochemical Centre, Amersham, GB).The cell count was adjusted to 1 X 10s/ml. Various concentrations of effector cells were added in 100 pl volume to 100 pl of target cells to give ratios of 20: 1, 10 : 1 , 5 : 1,2.5 : 1 and 1.25 : 1 in microtitration wells (Titertek; Flow Labs.. Irvine, Scotland). After 4 or 18 h of incubation at 37"C, 100 pl of SN from each well was collected and counted in a gamma counter (Wallac, Turku, Finland). The percentage of radioactivity released was calculated as follows: % Rclcasc
=
W - S )
100
(TOT - S )
2.3 Scanning electron microscopy (SEM) For SEM analysis the cells were incubated for 2 h in the Boyden chamber. The filters were then fixed in glutaralde-
where EX: cprn released from experimental cells; S: cpm released from cells in the medium control during incubation; and TOT: total cpm incorporation into cells.
The migration of human natural killer cells
Eur. J. Immunol. 1991. 21: 35-42
2.8 Statistical methods To evaluate the statistical significance of the cell migration assay results the paired t-test was used.
3 Results 3.1 LGL morphology on the filters Highly purified LGL (containing 78.8 +. 5.6% of CD16+ cells) were placed on either untreated or fibronectin-coated polycarbonate filters in Boyden invasion chambers. Immunostaining results revealed that the cells which attached and spread on the filters were CD16+ cells.The cells which were in the process of migration through the 3-pm pores of untreated filter were of round and spread forms (Fig. 1A). The cells on the fibronectin-coated filters (Fig. 1B) were found to have many more microspikes than cells on the
37
untreated filters and the spreading was more complete. Lamellipodia typical of migrating cells were seen among the cells on both uncoated and fibronectin-coated filters. The cells which had migrated through the filter remained for a short while attached in spread form on the lower surface of the filter before falling down t o the lower compartment of the chamber. Microvilli density on the surface of LGL decreased markedly among the cells which had migrated through the filter (not shown). 3.2 Effect of fibronectin coating in adhesion and migration
The effect of increasing concentrations of fibronectin coating on filters was examined. Human Ficoll- and Percoll-fractionated lymphocytes (58.0 k 4.2% of LGL) freshly isolated from blood were placed on the filters in the Boyden chambers. The number of cells which migrated through filters into the lower compartment of the chamber during incubation was counted. Cell composition in the lower compartment was determined from cytocentrifuge smears by morphological criteria, and the proportion of migrated LGL from the original input placed on the filters was calculated. As shown in Fig. 2 both adhesion on the filter and migration through the filter were found t o be dependent upon the concentration of the fibronectin coating. The increase in migration produced by coating concentrations between 10 and 20 pg/ml was much more prominent than the increase in adhesion. The numbers of cells which migrated through untreated or fibronectin-coated filters into the lower compartment of the chamber during different incubation times were also compared. Significantly more (p < 0.001) cells migrated through the fibronectin-coated filters than through the untreated filters (Fig. 3A), and a significantly higher @
