Increased proliferation, lytic activity, and purity of human natural killer cells cocultured with mitogen-activated feeder cells

CELLULAR

IMMUNOLOGY

135,454-470

(1991)

Increased Proliferation, Lytic Activity, and Purity of Human Natural Killer Cells Cocultured with Mitogen-Activated Feeder Cells’ HANNAH

RABINOWICH,*,~

PETER SEDLMAYR,*‘? RONALD AND THERESA L. WHITESIDE**?

*Pittsburgh Cancer Institute and tDepartments University of Pittsburgh School of Medicine, Received

November

1, 1990; accepted

of Pathology

and #Medicine, Pennsylvania

Pittsburgh, February

B. HERBERMAN,*,?*

13, 1991

The addition of mitogen-prestimulated periferal blood lymphocytes (PBL) or Epstein-Barr virus (EBV)-transformed lymphoblastoid cell lines (LCL) cultures to enriched populations of natural killer (NK) cells obtained from PBL of normal donors in the presence of rIL-2 resulted in highly significant increases in proliferation, purity, and cytolytic activity of cultured NK cells. Two sources of enriched NK cell preparations were used: (i) Adherent-lymphokine activated killer (A-LAK) cells obtained by adherence to plastic during 24 hr activation with lo3 Cetus U/ml rIL2; and (ii) NK cells negatively selected from PBL by removal of high-affinity rosette-forming cells and CD3+ lymphocytes. Coculture of A-LAK cells for 14 days with autologous or allogeneic Con A-activated PBL ( lo6 cells/ml) or selected EBV-transformed LCL (2 X 1O5 cells/ml) as feeder cells increased fold expansion by a mean + SEM of 629 fold f 275 (P i 0.019) and 267 fold + 54 (P < O.OOOl), respectively, compared to 55 + 20 in A-LAK cultures without feeder cells. The addition of either activated PBL or EBV lines to A-LAK cultures also led to a significant increase in the percentage of NK cells (CD3-CD56+) (84 f 2.4 and 84 + 2.6%, respectively, P < 0.0001 for both), compared to 53 f 7.2% in cultures without feeders. The presence of feeder cells in cultures of A-LAK cells also led to significantly higher anti-tumor cytolytic activity compared to control cultures, as measured against NK-sensitive (K562) and NK-resistant (Daudi) target cells. Mitogen-stimulated CD4+ PBL purified by positive selection on antibody-coated flaskswere better feeders than CD8+ or unseparated PBL. In the presence of feeder cells, it was possible to generate up to 6 X lo9 activated NK cells from 2 X 10’ fresh PBL by Day 13 of culture. Enhanced NK cell proliferation in the presence of feeder cells was not attributable to a detectable soluble factor. The improved method for generating A-LAK or activated-NK cells should facilitate cellular adoptive immunotherapy by providing sufficient numbers of highly enriched CD3-CD56+ effector cells with high anti-tumor activity. 0 1991 Academic PISS, IX.

INTRODUCTION Recent clinical trials of adoptive immunotherapy (AIT) with a combination of rIL2 and lymphokine activated killer (LAK) cells in patients with metastatic renal cell carcinoma (RCC), melanoma, and colerectal cancer resulted in overall response rates of 10 to 20% (1). The need to improve AIT of cancer patients has led recent investigations to focus on the most tumoricidal effector cell populations. Most investigators now agree that LAK activity is attributable to a quantitatively minor subpopulation ’ This research was supported in part by ACS Grant IM588A to T.L.W. 454 0008-8749/9 I $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form resewed.

PROPERTIES

OF

HUMAN

ACTIVATED

NK

CELLS

455

of lymphocytes (10-I 5%) which are CD3-CD56+ natural killer (NK) cells (2-4). In our laboratory, a method has been developed for selective enrichment of rll-Zactivated NK effector cells from rat splenocytes and from human peripheral blood mononuclear cells (PBMNC) (5, 6). The method depends on the propensity of IL-2-activated NK cells to adhere to plastic surfaces before most T lymphocytes do. In this way, activated NK cells can be captured on plastic and separated from T lymphocytes, whose activation and expression of surface adherence molecules requires a longer period of time (7, 8). Cultivation of these adherent NK cells in the presence of rIL-2 and conditioned medium for 14 to 2 1 days results in proliferation of a cell population containing a high percentage of cells with morphologic characteristics of large granular lymphocytes and a CD3-CD56+ phenotype (9, 10). Adherent-lymphokine activated killer (A-LAK) cells derived from normal humans have significantly higher cytolytic activity on a per cell basis and greater proliferative capacity than LAK cells prepared in a conventional fashion (6). In animal tumor models of established metastases, A-LAK cells have superior antitumor effects compared to unfractionated LAK cells (11). Recent experiments in our laboratory utilizing human A-LAK cells plus rlL-2 for local immunotherapy of human squamous cell carcinoma of the head and neck growing in nude mice also indicated that up to IO-fold more LAK than A-LAK cells were needed to achieve complete growth inhibition of established 3- or 7-day tumors (12). These results suggested that therapy with purified or enriched populations of rlL-2-activated NK cells may be more advantageous, may allow the use of fewer cells for therapy, and may require lower doses of rIL-2 than those used with unfractionated LAK cells ( 13). Since only less than 4% of peripheral blood lymphocytes (PBL) adhere to plastic after 24 hr of rlL-2 activation (6), a high proliferation rate is obligatory to obtain a sufficient number of cells for adoptive transfer. In this report we describe an improved approach to generation of A-LAK or activated-NK cells from human PBL. Although Epstein-Barr virus (EBV)-transformed B cell lines have been used in the past by several groups to improve growth of human NK cells (14, 15) Con A and rIL-2-activated T cell subpopulations have not been carefully evaluated as a possible source of growth factors for human NK cells. The use of T cell blasts as feeder cells should facilitate future studies of human NK cells and cellular AIT by providing sufficient numbers of highly enriched CD3-CD56’ effector cells with high antitumor activity. MATERIALS

AND

METHODS

Peripheral blood Iymphocytes. PBMNC were isolated from heparinized venous blood obtained from normal volunteers or leukapheresed blood obtained from the Central Blood Bank of Pittsburgh by centrifugation on Ficoll-Hypaque gradients. Cells collected from the gradient interface were washed twice in RPM1 1640 medium (GIBCO, Grand Island, NY) and incubated in medium containing phenylalanine-methyl ester (PME, 1 mg/ml/ 10’ cells, duPont de Nemours, Glenolden, PA) for 40 min to remove monocytes (16). After three washes, the cell concentration was adjusted to lO’/ml in complete tissue culture medium (TCM) consisting of RPM1 supplemented with 2 mM L-glutamine, 50 U/ml penicillin, 50 pg/ml streptomycin, 25 mM Hepes buffer, and 10% heat-inactivated pooled AB human serum (Central Blood Bank of Pittsburgh). Cell lines. The EBV-transformed lymphoblastoid cell lines, DEM, DBB, YAR, MOU, JVM, QBL, JHAF, DUCAF, MZ0707, COX, BOLETH, KOSE (obtained from

456

RABINOWICH

ET AL.

the XI International Histocompatibility Workshop), the locally transformed EBVlymphoblastoid cell line (LCL), DT, and RUSH, the human myeloid leukemia cell line, K562, and the human Burkitt lymphoma-derived cell line, Daudi, were maintained in culture in RPM1 1640 medium supplemented with 10% heat-inactivated fetal calf serum (FCS; GIBCO). The cell lines were subcultured as needed, and cells in the log phase of growth were used as feeders and for cytotoxicity assays. To be used as feeders, EBV-LCL were irradiated at a dose of 25,000 R which was necessary to abrogate proliferation, and added at a concentration of 2 X lo5 cells/ml to enriched NK cell preparations. All cell lines were mycoplasma free, as determined by DNA hybridization using Gen-Probe kit (San Diego, CA). Generation ofA-LAIY ceE1.r.Monocyte-depleted PBL at a concentration of 5 X lo6 cells/ml were incubated in TCM containing 1O3Cetus U/ml rIL-2 (Cetus, Emeryville, CA) in plastic culture flasks (T25 to T150, depending on number of cells available; Corning, Ithaca, NY) for 24 hr. The flasks were maintained in a horizontal position at 37°C in humidified atmosphere of 5% COZ. After 24 hr activation in rIL-2, nonadherent cells were decanted and centrifuged. Adherent cells were washed three times with prewarmed TCM to remove cells that were not firmly attached to plastic. The initial number of A-LAK cells per flask was estimated as described earlier (6). The cell-free supernatant of the nonadherent cell population was collected and used as autologous conditioned medium (AuCM). A-LAK cells were supplemented with TCM containing 50% AuCM and lo3 Cetus U/ml’ rIL-2. A-LAK cultures were incubated at 37°C in CO2 in air and maintained at a concentration of l-5-2.0 X lo6 cells/ml by supplying fresh TCM containing 10’ Cetus U/ml rIL-2 as needed. A-LAK cells were cultured for at least 3 weeks and tested weekly for cytotoxicity and surface markers. For consistency, proliferative, phenotypic, and cytotoxicity data for 1Cday A-LAK cell cultures are presented in this manuscript. PuriJication of NK cells from PBL by depletion of E-RFC and CD3+ cells. Purification of NK cells from monocyte-depleted PBL was performed by two-step depletion of T cells in which high-affinity rosette-forming cells (E-RFC) and then CD3+ lymphocytes were removed by gradient centrifugation and magnetic beads separation, respectively. E-RFC were removed from PBL as described by West et al. (17). Briefly, monocytedepleted PBL at a concentration of 4 X lo6 cells/ml were mixed with FCS and 0.5% sheep red blood cells at a ratio of 1:2:2 (V/V), respectively. The mixture was centrifuged at 300g for 10 min and incubated for 1 hr at 29°C. Cells forming E-rosettes were separated from nonrosetted cells on Ficoll-Hypaque density gradients. Further removal of T cells from lymphocytes depleted of E-RFC was accomplished by sensitization with anti-CD3 monoclonal mouse antibodies (DAKO, Carpentia, CA), followed by a negative selection with magnetic beads coated with goat antimouse immunoglobulins (Advanced Magnetics, Cambridge, MA). Briefly, E-RFCdepleted PBL were resuspended in Hanks’ balanced salt solution (HBSS) containing 2% FCS and incubated in the presence of anti-CD3 monoclonal antibodies for 30 min at 0°C. The cells were washed twice and then incubated with magnetic beads (1 cell: 30 beads) for 30 min at 0°C. After two successive incubations with beads, a magnet was used to separate beads with attached CD3+ T cells from CD3- NK cells. The negatively selected E-RFCCD3cells were cultured in the presence of lo3 Cetus U/ ml rIL-2 with and without feeder cells at a 1: 1 cell ratio (E:S) with mitogen-stimulated PBL or a 5: 1 cell ratio (E:S) with EBV lines.

PROPERTIES

OF

HUMAN

ACTIVATED

NK

CELLS

457

Selection of enriched CD4+ and CD8’ lymphocytes from peripheral blood. Enriched populations of CD4+ and CD8 T lymphocytes were obtained by positive selection on anti-CD4 or anti-CD8 monoclonal antibody-coated flasks (Generously supplied by Applied Immune Sciences, Menlo Park, CA). To capture T-lymphocytes on antibody-coated surfaces, PBL (100 X 1O6cells in 5 ml RPMI) were placed in T25 culture flasks and incubated at 4°C for 1 hr. Following incubation, noncaptured cells were removed by extensive washing, and TCM containing rIL-2 (100 Cetus U/ml) was added. To be used as feeders, captured CD4+ and CD8 lymphocytes were stimulated by various combinations of T-mitogens during 72 hr incubation at 37°C in humidified atmosphere of 5% CO*. Preparation of mitogen-activated PBL feeder cells. Unseparated lymphocytes or enriched CD4+ or CD8+ T cells obtained from normal human peripheral blood were incubated at a concentration of 106/ml for 3-7 days in the presence of 50 Cetus U/ ml rIL-2 in T25 to T150 tissue culture flasks with one of the following activators: 10 Kg/ml Con A (Calbiochem, La Jolla, CA); 10 rig/ml OKT3 MoAb (Ortho Biotech Division, Raritan, NJ); 700 rig/ml ionomycin (Sigma, St. Louis, MO); 1 rig/ml PMA (Sigma); 10 pg/ml Con A + 1 rig/ml PMA; 10 rig/ml OKT3 MoAb + 1 rig/ml PMA; 700 rig/ml ionomycin + 1 rig/ml PMA. Activated PBL were washed three times, irradiated at 5000 R and added at a concentration of lo6 cells/ml to A-LAK cultures on Day 1 and to E-RFC-CD3- PBL on Day 0. In several experiments, Con A-stimulated PBL were washed three times with 0.1 M methyl-1-a-D-mannopyranoside (Sigma) in RPM1 to competitively remove Con A from the cell surface (18). Proliferative activity. Fold expansion of A-LAK cells was determined by cell counts performed in the presence of trypan blue. The total cell number in culture was divided by the number of adherent cells observed after the first 24 hr of rIL-2 activation (6). In some experiments, proliferative responses were measured by uptake of [methyl3H]thymidine (6.7 Ci/mmol; NEN, Boston, MA) during 18 hr of incubation prior to cell harvest. Cytotoxicity assay. Cytotoxic activity of A-LAK cells was tested against NK-sensitive myeloid leukemia line, K562, and the NK-resistant B-lymphoblastoid cell line, Daudi, in a 4-hr “Cr release assay as described earlier (6). The target cells were labeled with 1OO- 150 PC1 of [S’Cr]sodium chromate (sp act 5 mCi/mmol; NEN, Boston, MA) for l-2 hr at 37°C. Washed target cells were added to wells of U-bottomed 96-well plates (Costar) and incubated with effector cells at effector:target (E:T) ratios ranging from 6: 1 to 0.375: 1. Maximal release and spontaneous release were determined by incubating the tumor cells with 5% T&on-X or medium alone, respectively. All determinations were made in triplicate. Radioactivity was counted in a gamma counter, and percentage of specific lysis determined according to the formula % Specific Lysis = mean cpm experimental release - mean cpm spont. release x 100. mean cpm maximal release - mean cpm spont. release Lytic units were calculated using a computer program based on the formula developed by Pross et al. (19). One lytic unit was defined as the number of effector cells needed to lyse 20% of 5 X 1O3target cells, and lytic units per lo7 effector cells were calculated. Flow cytometry. To determine relative proportions of NK and T cells in fresh and cultured cell preparations, two-color flow cytometry was performed on a FACScan.

458

RABINOWICH

ET AL.

Cells were adjusted to a concentration of 0.5 X lo6 in 0.2 ml PBL-0.1% sodium azide and stained with fluorescein- or phycoerythrin-labeled monoclonal antibodies with specificity for different surface markers on human mononuclear cells. The cells were incubated with monoclonal antibodies, pretitered to give optimal staining, for 15 min at 4°C. The monoclonal antibodies were purchased from Becton-Dickinson (San Jose, CA) and included: Leu4 (anti-CD3), Leu19 (antiCD56), Leu 1 la (antXDl6), Leu 12 (antiCD20), and LeuM3 (antXDl4). The stained cells were washed twice with the PBS-sodium azide buffer and resuspended in 2% (w/v) paraformaldehyde. Controls were unstained cells in PBS-sodium azide, and isotype controls (IgGl and IgG2a) were used to set markers. Cell suspensions were stained with monoclonal antibody to the pan-leukocyte antigen, anti-HLe- 1 (anti-CD45, Becton-Dickinson), to confirm the exclusion of debris. Statistical analysis. Significance of differences in fold expansion, cytotoxicity, and percentage of cells stained for surface markers were calculated using Student’s t test or Wilcoxon’s rank sum test. The Statview 5 12+ program was used for statistical analysis. RESULTS Propagation of A-LAK cells in the presence of mitogen-activated PBL. In initial experiments, A-LAK cultures were expanded in the presence of 1.0 X lo6 irradiated PBL which had been stimulated with various combinations of mitogens for 72 hr. Increased proliferation was observed in the presence of feeder cells, regardless of the type of agent used to preactivate them (Table 1). However, best results were obtained TABLE 1 Proliferation of A-LAK Cells Cocultured with Allogeneic PBL Preactivated with rIL-2 and Various Activating Agents [‘H]Thymidine uptake * (cpm X 10e3) Feeder cells

Activator” (of feeder cells)

Expt 1

rIL-2 Con A PMA Con A + PMA OKT3 OKT3 + PMA Ionomycin Ionomycin + PMA

75* 15 60 + 20 186 -c 18 92* 4 219+ 3 172 -t 10 177* 3 149f 4 163+ 1

Fold expansion’

Expt 2 14t 39* 172 k 65k 207 + 114f 171+ ND 126zh

3 2 12 1 15 5 6 5

Expt 1

Expt 2

51 128 176 91 286 87 102 57 76

3 26 155 17 257 50 61 17 44

n Allogeneic PBL used as feeders were prestimulated with lo3 Cetus U/ml rIL-2 or various combinations of activating agents in the presence of 50 Cetus U/ml rIL2, as described under Materials and Methods. Preactivated feeder cells were irradiated and added to A-LAK cultures on Day 1. * Proliferation of 10’ A-LAK cells was measured on Day 14 by incorporation of t3H]thymidine. Data are means * SD in triplicate wells. Two experiments are shown out of three performed. ‘Fold expansion was determined by dividing total cell number on Day 14 by the initial number of adherent cells at 24 hr.

PROPERTIES

OF HUMAN

ACTIVATED

459

NK CELLS

using Con A- or Con A + PMA-stimulated PBL as feeders. A-LAK cells also proliferated better in the presence of allogeneic PBL which had been preactivated with rIL-2 for 3 days (Table l), but not of monocyte-depleted, irradiated fresh PBL (Table 2). Addition of 50% conditioned medium, which had been reported necessary for growth of ALAK cells (20) was not needed when A-LAK cells were cocultured with activated PBL as feeders (data not shown). Improved proliferation of A-LAK cells in the presence of feeder cells was not due to the direct effect of activators on A-LAK lymphocytes (Table 2). No increase in A-LAK cell proliferation was observed when various activating agents, singly or in combination, were directly added on Day 1 to A-LAK cell cultures. Preferential expansion of CD3-CD56+ cells was demonstrated only in A-LAK cell cultures containing irradiated PBL prestimulated with Con A or Con A + PMA (Table 2). Treatment of Con A-stimulated feeders with 0.1 A4 methyl- 1-cu+mannopyranoside to competitively remove Con A from the cell surface membrane had no effect on the growth-promoting activity of feeder cells (data not shown). The increased expansion observed in presence of OKT3 or OKT3 + PMA (Table 2) was due to activation and proliferation of T lymphocytes, which adhere to plastic surfaces and initially constitute a part of the A-LAK cell population (Table 3). Thus, coculture of A-LAK cells with

TABLE 2 Expansion and Phenotypic Characteristics of A-LAK Cells Cocultured with Allogeneic Preactivated PBL Feeder Cells or Cultured in the Presence of Activating Agents ?k Positive cells’ Activator” (of A-LAK cell culture) None Fresh unstimulated PBL Con A stimulated PBL Con A + PMA stimulated PBL OKT3 OKT3 + PMA Con A; PMA; Con A + PMA; Ionomycin; Ionomycin + PMA

Fold expansionh (Day 14) 48 +- 19 (II = 19) 2 1 0.4 (n = 6) 629 +- 275 (n = 25) 283 + 65 (n = 17) 33 f 8 (n = 3) 27 2 10 (n = 3)

CD3CD56+

CD3+ CD56 ’

CD3 ’ CD56

53 k 7 (n = 19) QNSd

15+4

27 t 6

84 + 2 (n = 24) 83 + 3 (n = 17) 5k2 (n = 3) 9+-4 (n = 3)

QNS

QNS

7 -r 0.8

8t3

7*1

9t2

I3 + 3

81 tj

II _t5

79 t 9

2.8 4 1.4 (n = 21)

’ Following the initial 24 hr activation with IL-2, A-LAK cells were incubated in the presence of IO’ Cetus U/ml of IL-2 plus one of the listed activating agents. Alternatively, A-LAK cells were cocultured with fresh or prestimulated PBL irradiated at 5000 R and added to A-LAK cell cultures at a concentration of lo6 cells/ml on Day 1. Concentrations of activators used are described under Materials and Methods. ’ Fold expansion was determined as described in a footnote to Table I. Data are means -CSEM. ‘ Two-color flow cytometry on a FACScan was used to determine proportions (shown as means k SEM) of NK and T cells in A-LAK cell cultures. d Quantity not sufficient.

RABINOWICH

460

ET AL.

TABLE 3 Phenotype of Fresh PBL and A-LAK Cells before and after Culture with Prestimulated Feeder Cells” % Positive cells Expt No.

Cells

CD3CD56+

CD3+ CD56+

CD3+ CD56-

1

Fresh PBL Detached A-LAKb Expanded A-LAK Fresh PBL Detached A-LAKb Expanded A-LAK Fresh PBL Detached A-LAKb Expanded A-LAK Fresh PBL Detached A-LAKb Expanded A-LAK Fresh PBL Detached A-LAK b Expanded A-LAK Fresh PBL Detached A-LAK b Expanded A-LAK Fresh PBL Detached A-LAKb Expanded A-LAK

24 33 86 9 13 95 16 92 94 5 65 54 28 89 84 7 71 72 9 81 85

23 20 5 2 1 3 2 3 3 6 20 17 2 0 3 4 8 5 2 5 7

40 36 4 68 11 2 12 3 2 78 14 20 55 0 2 77 17 19 II 8 6

2 3 4 5 6 7

’ Two-color flow cytometry was used to determine proportions of NK and T cells in fresh PBL (Day 0), 24 hr activated A-LAK cells and expanded A-LAK cells cocultured for 14 days with prestimulated and irradiated allogeneic PBL as feeder cells. b A-LAK cells were detached from plastic surfacesby scraping with a rubber policeman after 24 hr of rIL2 activation.

Con A-stimulated PBL as feeder cells led to preferential proliferation of NK cells, while under different culture conditions (i.e., OKT3 or OKT3 + PMA) T lymphocytes proliferated preferentially in the same culture. Phenotypic characterization of A-LAK cells detached from plastic surfaces by scraping with a rubber policeman after 24 hr of adherence in the presence of rIL-2 indicated that considerable (up to 92%, Table 3) enrichment in CD3-CD56+ was achieved by the adherence step. However, the presence of Con A-stimulated PBL feeder cells induced preferential proliferation of NK cells, when the initial adherent population contained high proportions of CD3+ cells (see Experiment 1. in Table 3). A-LAK cells after 24 hr of activation consisted mainly of NK cells with CD 16+CD56+ phenotype (Table 4). However, the initially minor subpopulation of NK cells with CD 16CD56+ phenotype was found to be prevalent in A-LAK cell cultures expanded for 14 days in the presence of feeder cells. In a separate study (data not shown), significant correlation between the percentage of CD3CD56+ in fresh PBL and in ALAK cultures expanded without feeder cells (n = 22 r = 0.47 p = 0.02) was observed. However, when A-LAK cells were grown in presence of Con A-stimulated PBL as

PROPERTIES

OF HUMAN

ACTIVATED

461

NK CELLS

TABLE 4 Phenotypic Characteristics of A-LAK Cells before and after Culture with Prestimulated Feeder Cells” % Positive cells Cell population Detached A-LAK (Day IJh Expanded A-LAK (Day 14)

No. of experiments 3 13

CD56CD16+

CD56’ CDl6+

CD56’ CD16-

CD56 + CD3

9 k6

66 t 4

17 t 5

87 _t 3

1.3 * 0.4

21 t_5

46 f 5

81 23

’ Two-color flow cytometry was used to determine proportions of different NK subpopulations in 24 hr activated A-LAK cells as well as expanded A-LAK cells cocultured for 14 days with prestimulated and irradiated allogeneic PBL. Data are means + SEM. b A-LAK cells were detached from plastic surface by scraping with a rubber policeman after 24 hr of activation.

feeders this correlation was no longer significant (n = 23 r = 0.19 p = 0.38). This indicated that the use of feeder cells made the generation of cultures highly enriched in CD3-CD56’ A-LAK cells possible, even from donors with low proportions of peripheral blood NK cells or with a low capability of their IL-2-activated NK cells to adhere to plastic. In order to better characterize the feeder cells responsible for sustaining proliferation of A-LAK cells, enriched CD4+ and CD8+ populations were obtained from normal PBL by selection on antibody-coated flasks. The enriched populations of T cells were cultured with various combinations of activating agents for 72 hr, irradiated, and used as feeders in human A-LAK cell cultures. A-LAK cells grown in the presence of CD4’

TABLE 5 Expansion of A-LAK Cells Cocultured with Enriched and Activated Allogeneic CD4’ or CDS+ T Lymphocytes as Feeder Cells” A-LAK cell cultures fold expansionh Feeder cells

Feeder cell activation

CD4+ feeders

CD8 + feeders

+ + + + + +

rIL-2 Con A OKT3 PMA Con A + PMA OKT3 + PMA

26 788 3477 1623 165 2670 1085

26 489 574 884 173 343 163

a Enriched CD4’ and CD8+ T lymphocytes were obtained by positive selection from normal peripheral blood on anti-CD4 and anti-CD8 coated flasks. Captured cells were activated by various stimuli (as listed) for 72 hr, irradiated, and used as feeder cells in A-LAK cultures. Data shown are from a representative experiment. ’ Fold expansion was determined as described in a footnote to Table I.

462

RABINOWICH

ET AL.

or CD8+ feeders demonstrated comparable enrichment in cells with the CD3-CD56+ phenotype as well as levels of cytolytic activity (data not shown). However proliferation of A-LAK cells in the presence of stimulated CD4+PBL was superior to that of cultures containing CD8 feeder cells, as shown in Table 5. The highest expansion fold was obtained with CD4+ T cells which had been cultured with Con A or Con A + PMA (3477 and 2670, respectively, on Day 14). Phenotype of A-LAK cells cocultured with feeder cells. To determine the relative proportions of NK cells (CD3-CD56+) and T cells (CD3+CD56- and CD3+CD56+) in A-LAK cell cultures grown without or with feeder cells, two-color flow cytometry was performed. In A-LAK cell cultures supplemented with 50% autologous conditioned medium only, the mean percentage * SEM of CD3-CD56+ NK cells (53 -+ 7.2) was significantly lower than that in cultures containing either Con A-activated PBL (P < 0.000 1) or EBV-LCL (P < 0.000 1) as feeder cells (Fig. 1A). Cultures grown in the presence of Con A-activated feeder cells contained 84% + 2.4 (mean f SEM; n = 24) of CD3CD56+ cells and those in the presence of EBV-LCL, 84% f 2.6 (n = 27). The remaining cells in these cultures were equally divided between CD3+CD56+ and CD3+CD56- T lymphocytes. Both these populations of T cells represented a substantial proportion of A-LAK cell cultures grown without feeder cells (Fig. IA). When paired data for control (no feeder cells) cultures vs either cultures with Con A-activated feeder cells or with EBV-LCL as feeders are examined (Fig. lB), the advantage of feeder cells becomes even more apparent. In most cases A-LAK cultures supplemented with feeder cells contained much higher proportions of CD3CD56+ NK cells than paired control cultures (Fig. 1B). Proliferation of A-UK cells cocultured with feeder cells. Proliferation of A-LAK cells without added feeder cells, or in the presence of Con A-stimulated PBL, or EBVLCL was tested on Day 14 of culture. Figure 2 compares the fold expansion obtained with differently fed A-LAK cell cultures. Significantly higher expansion was demonstrated in cultures containing Con A-PBL 629 & 275 (mean + SEM; n = 25; range 9-5477; P < 0.019) or EBV-LCL 267 f 54 (mean +- SEM; n = 22; range 18-1066; P < 0.0001) as compared to 55 + 20 (mean f SEM; n = 34; range l-565) in A-LAK cell cultures without feeder cells. Enhanced proliferation of A-LAK cells was observed only in the presence of one of the feeder cells and rIL-2. In the absence of rIL-2, neither Con A-PBL nor EBV-LCL were able to sustain growth of enriched NK cell cultures. Cytolytic activity ofA-LAK cells cocultured with feeder cells. Antitumor effector cell function of A-LAK cells grown with or without feeder cells was measured against the NK-sensitive myeloid leukemia cell line, K562, and against the NK-resistant B-lymphoblastoid target, Daudi, on Day 14 of culture. As shown in Fig. 3, A-LAK cells expanded in the presence of activated PBL or EBV-transformed feeder cells were significantly more cytotoxic against tumor cell targets than A-LAK cells cultured without feeder cells. A-LAK cocultured with Con A-PBL feeders had mean f SEM cytotoxicity of 8037 + 868 LUzo/107 cells (n = 19) against K562, which was significantly different (P < 0.005) from the activity {mean f SEM of 4076 -t 80 1 LU2,,/107 cells (n = 1 l)} of A-LAK cells grown without feeder cells. Also, mean + SEM activity against Daudi (7754 & 1896 LU20/107 cells; n = 17) was higher than that (2897 f 1935 L&,-J 10’ cells; n = 10) in control A-LAK cultures, but the difference was not significant. Feeding of A-LAK cultures with EBV-LCL significantly increased their cytolytic activity as compared to cultures without feeder cells. A-LAK cultures cocultured with

PROPERTIES lOO-

OF HUMAN

ACTIVATED

NK CELLS

463

A

eo-

m = 8

60

s ‘5 a 0” 8

40-

20-

O-

No

feeders n=19

cT2-e~:L

Ete.si%L

nz24 pco.ooo1

11.27 pco.ooo1

1”

Control

ConA-PBL feeders

Control

EBV-LCL feeders

FIG. 1. Phenotypic analysis of A-LAK cells grown without feeder cells, with Con A-stimulated, irradiated PBL, or with B lymphoblastoid cell lines. The P values refer to comparisons between cultures grown with or without feeder cells. (A) Summary of nonpaired data from (n) number of normal donors in means k SEM (B) Paired data for control cultures vs either cultures with Con A-activated feeders or with EBV-LCL as feeders.

EBV-LCL had mean + SEM cytotoxicity of 7258 2 883 LU2,,/107 cells (n = 19; P < 0.01) against K562 and of 11078 f 2186 LU20/107 cells (n = 17; P < 0.01) against Daudi. As both fold expansion and cytolytic activity on a per cell basis of A-LAK cells cocultured with feeders were significantly superior to control cultures (Figs. 1 and 3), the total lytic units of cytotoxicity calculated for cultures with feeders were significantly (P < 0.001) greater than those in A-LAK cultures grown without feeder cells (data not shown).

Characteristicsof activated NK cellsexpandedfrom ERFC- and CD3-depleted PBL. Highly-enriched preparations of fresh NK cells were obtained by removal of highaffinity SRBC rosette-forming cells and then of CD3’ PBL using antibody-coated

464

RABINOWICH

ET AL.

:

:

. . .

:

I

: i : .

. . . . . . 1I No

feeders

n134

ConA-PBL

EBV-LCL

Feeders

Feeders

n.25 P’O.019

nz22 p<0.0001

FIG. 2. Fold expansion of A-LAK cells cultured without feeder cells, with Con A-stimulated PBL or with B lymphoblastoid cell lines. Horizontal bars represent mean values. The P values refer to comparisons between cultures grown with or without feeder cells. n = number of donors.

magnetic beads. The percentage of lymphocytes with high-affinity receptor for SRBC varied between 34 and 46. The yield of nonrosetted cells was in the range of 35-40% of the initial PBL used. The purity of NK cell preparations (CD3-CD56+) obtained by this procedure ranged from 64 to 99% (n = 23,83 f 10 mean It_ SD) with yield in the range of 0.5-5% of initial PBL. Eighty to one-hundred percent of the recovered cells had LGL morphology on May-Grunwald-Giemsa (MGG)-stained smears. Enriched NK-cell preparations were cultured in the presence of lo3 Cetus U/ml rIL-2, with or without feeder cells, and monitored for fold expansion, cell surface markers, and cytolytic activity. Purified NK cells (>90% CD3-CD56+) failed to grow in the presence of rIL-2 alone or rIL-2-activated feeders alone, as shown in Table 6. However, in the presence of CD4+ PBL activated with Con A or with OKT3, fold expansion was 848 and 350, respectively. T-depleted, enriched NK cells grown in the presence of allogeneic Con A-prestimulated PBL consisted of 74 a 8.8% CD3-CD56+ (Fig. 4)

PROPERTIES 14ow

OF

n

HUMAN

ACTIVATED

NK

CELLS

465

K562 DAUDI

12000

(17)

10000

(191 P. .z ‘3 0 (II

TT

8ooo

g s0

4ooo

2ooo

0 No

feeders

ConA-PBL

EBV-LCL

feeders

feeders

p
pro.1

p4.01p
FIG, 3. Cytolytic activity of A-LAK cells grown without feeder cells, with Con A-stimulated PBL or with B lymphoblastoid cell lines. The data are means 2 SEM. In parentheses are numbers of experiments. Cytotoxicity is expressed in LU20/107 cells. The P values refer to comparisons between cultures grown with or without feeder cells.

and had mean _CSEM cytolytic activity of 6923 t- 1689 LU& IO7 cells against K562 and 3614 + 2029 against Daudi (Fig. 5). Growth of the same cell preparations with EBV-LCL also gave rise to equally highly enriched and even more cytotoxic NK cells (86 -t 3% CD3CD56+; 8676 + 1929 LU20/107 against K562; 8035 2 2743 against Daudi). The preferential induction of proliferation of CD3-CD56+ and NK activity in the presence of feeder cells was also demonstrated for ERFC CD3- PBL. Table 6 (bottom) presents an overgrowth of T cells (96% CD3+CD56-) observed on Day 14 in an enriched preparation of ERFC-CD3- PBL propagated in rIL-2 only. The same cell preparation gave rise to 87% CD3-3CD56+ when cocultured with an irradiated EBV cell line (DEM). DISCUSSION Populations highly enriched in IL-2-activated NK cells selected from PBL by their adherence to plastic (A-LAK cells) have been recently used for AIT of cancer instead

466

RABINOWICH

ET AL.

TABLE 6 Expansion, Phenotype and Cytolytic Activity of Human NK Cells Obtained as ERFC and CD3- Cells from PBL and Cocultured with Different Feeder Cells” % Positive cells’ CD4+ feeders

Activator (of feeder cells)

Fold expansion (Day 14)

rIL-2 Con A OKT3 Ionomycin PMA

0.7 2.3 848 350 0.9 0.8

+ + + + +

CD3CD56+

CD3+ CD56+

CD3+ CD56-

QNS=

QNS

QNS QNS

QNS QNS

96

2

1

Q!S

Q:S

QiiS

QNS

QNS

QNS

Cytolytic activityb (LUzo/107 cells) EBV-LCL feeders

+d

% Positive cellsb

K562

Daudi

CD3CD56+

100 12306

285 3209

1 87

CD3+ CD56+ 1 1

CD3+ CD5696 7

’ Preparations enriched in human NK cells were obtained by removal of high-affinity-SRBC-rosette forming cells and then of CD3+ PBL using antibody-coated magnetic beads. Negatively selected cells were cultured in the presence of lo3 Cetus U/ml rIL-2 with different feeder cells added on Day 0 of culture. Data shown are from a representative experiment. b Cell surface markers and anti-tumor activity were measured on Day 14. ‘Quantity not sufficient. d The EBV-lymphoblastoid cells, DEM, were used as feeders.

of unfractionated, rIL-2-stimulated lymphocytes (LAK cells). The rationale for this alternative approach to AIT has been that highly purified, rIL-2-activated NK cells are potent anti-tumor effecters, while only a small proportion of LAK cells mediate anti-tumor cytotoxicity (2-4). However, cultures enriched in human NK cells have been difficult to proliferate and maintain. Highly purified NK cells, sorted as CD3-CD56+, failed to proliferate adequately even in the presence of high doses of rIL-2, while partially enriched NK cultures tended to be overgrown by T cells (2 1). It has been particularly difficult to obtain sufficient numbers of A-LAK cells for therapy from patients with metastatic melanoma and renal cell carcinoma participating in a phase I trial at our institution. Only 42 of 144 patients (29%) prescreened for the ability to generate A-LAK cells in vitro could be considered as candidates for A-LAK therapy, i.e., their NK cells supplied with rIG2 and conditioned medium prepared from rIL-2-activated PBL proliferated sufficiently well, so that enough effector cells were available for therapy (unpublished data). Based on this experience, we concluded that improvements in methodology for growth of populations enriched in rIL-Zactivated NK cells were necessary. In this study, we report that addition of feeder cells (either mitogen-activated PBL or LCL) allowed us to grow large numbers of relatively homogeneous, activated NK cell preparations with enhanced cytolytic activity from PBL of normal volunteers.

PROPERTIES

OF HUMAN

ACTIVATED

NK CELLS

467

100

80

20

0 ConA-PBL

feeders n=6

EBV-LCL

feeders IlS?

FIG. 4. Phenotypic analysis of PBL depleted of ERFC+ and CD3+ cells and cocultured with Con Aprestimulated PBL or with B lymphoblastoid cell lines. The data are means + SEM; n = number of donors.

The mechanism(s) by which the mitogen-activated PBL or EBV-LCL used as feeders induce preferentially proliferation of NK cells are largely unknown. In the presence of either type of feeder cells, proliferation of activated NK cells was still dependent on the presence of rIL-2: the feeder cell preparations were unable to induce proliferation of enriched NK cells in the absence of exogenous rlL-2. Also, stimulation by resting PBL did not facilitate proliferation of highly purified CD3-CD56+ lymphocytes. Furthermore, rIL-2-activated, enriched NK cell preparations did not respond by proliferation to any of the commonly used activation agents: Con A, PHA, OKT3 mAb, PMA, or calcium ionophore, added singly or in combination. Enhanced proliferation of activated NK cells in the presence of feeder cells might be due to a growth factor, either secreted or membrane-bound, which synergizes with rIL-2 in sustaining NK proliferation. We have previously shown that cytokines such as interferons, interleukins 1 or 4, and TNF (Yor /3 did not synergize with rIL-2 to promote growth of enriched NK cell preparations (22). Our previous data also showed that the addition of rIL-2-preactivated and irradiated tumor infiltrating lymphocytes (TIL) to cultures of PBL or autologous TIL increased significantly the percentage of cells with CD3-CD56+ phenotype in these cultures (22). Cell sorting experiments identified the stimulator subpopulation as the CD4+ T subset, while CD@ lymphocytes were less efficient in supporting proliferation of NK cells. In the present study, we compared the growth-promoting effect of CD4+ and CD8+ lymphocytes separated from normal PBL and stimulated by T-cell activators. Again,

468

RABINOWICH q

K562

q

DAUDI

ET AL.

- ConA-PEL

feeders n-5

EEV-LCL

feeders n+7

FIG. 5. Cytolytic activity of PBL depleted of ERFC+ and CD3’ cells and cocultured with Con A-prestimulated PBL or with B lymphoblastoid cell lines. The data are means f SEM; n = number of donors. Cytotoxicity is expressed in LUzo/ 10’ cells.

CD4+ lymphocytes augmented proliferation of enriched NK cells better than CD8’ T cells. Similarly, Perussia et al. demonstrated that proliferation of NK cells stimulated by the Daudi cell line required the presence of CD4+ T cells (23). Thus, proliferation of NK cells appears to be dependent on a yet unknown factor produced by activated CD4+ T cells. However, preliminary experiments in our laboratory with culture supematants of activated PBL or of A-LAK cells failed to demonstrate the presence of a growth-promoting soluble factor other than rIG2. Additional experiments, in which Con A-stimulated feeder cells were separated by filters (3.0-pm pores) from proliferating NK cells, confirmed a lack of detectable soluble components capable of supporting growth of NK cells. In line with these observations, it has been shown that glutaraldehyde-treated T cell blasts, which are metabolically inactive, were effective in stimulating the generation of NK-like cells in the presence of crude lymphokine supernatants (24). All these observations suggested a possibility that a membrane-associated factor or determinant might be involved in NK proliferation. Using the two types of feeders, activated PBL or EBV-LCL, NK cells with comparable levels of proliferation, purity, and anti-tumor activity were obtained. The ability of B lymphoblastoid cell lines to augment proliferation of NK from peripheral blood and facilitate the growth of NK cell clones has been described in the past (25, 26). By coculturing total PBL with two selected LCL, Daudi and RPM1 8866, Perussia and colleagues obtained an average of 25-fold increase in NK cells, whereas the number

PROPERTIES

OF

HUMAN

ACTIVATED

NK

CELLS

469

of T cells was increased by only 3-fold in lo-day cultures (23). Using a different approach, we cultured enriched NK celI preparations, rather than total PBL, with Con A-activated feeder cells. This approach enabled us to obtain NK cell preparations consisting of 84% -+ 2.4 (mean + SEM) cells with CD3-CD56+ phenotype and expansion fold of 629 +- 275 (mean -t SEM). By two-color flow cytometry, we determined that these proliferating NK cells were mainly CDS6 b”gh’CD 16-, which have been described before as the NK cell subset with high proliferative capability (8, 27, 28). In our hands, not only Con A-activated PBL but also most of the EBV-LCL tested (12 out of 14) were able to facilitate the growth of human CD56+CD16- NK cells. It has been shown earlier that in cocultures of PBL with allogeneic EBV-LCL in the absence of rIL-2, surface activation antigens, the MHC class II antigens, transferrin receptors, and IL-2 receptors were rapidly induced (23, 29). Also, EBV-LCL have been reported to function as excellent accessory cells for IL-2 production (30). These observations implied that IL-2 may be the factor necessary for NK cell growth. However, IL-2 alone could not induce preferential proliferation of NK cells in our experiments (22). Although, in experiments reported earlier by Melder et al. (6) A-LAK cells from six normal donors proliferated from 130- to 800-fold without feeder cells, addition of conditioned media was obligatory, and rIL2 alone was not sufficient. The growth-promoting ability of conditioned media is variable, however, and feeder cells provide a more consistent source of growth factors. Nevertheless, supernatants obtained from EBV-LCLs, which were demonstrated to induce proliferation of enriched NK cells, or from cocultures of EBV-LCL and NK cells failed to support NK proliferation. Also, NK proliferation was not induced when a filter was used to separate NK cells from EBV-LCL. These experiments indicated that cell-to-cell contact is necessary for optimal NK cell growth. Kobayashi et al. reported recently the purification of a new lymphokine, NK cell stimulatory factor (NKSF), from the supematant fluid of a phorbol-ester induced EBV-LCL. NKSF, purified 9200-fold, induced IFN-y production, augmented NK cell cytotoxic activity and synergized with PHA and phorbol-ester to enhance PBL proliferation (3 1). However, unlike EBV-LCL, NKSF alone did not induce preferential proliferation of NK cells. Furthermore, it was not possible to demonstrate NKSF production by the Daudi cell line reported to be a powerful inducer of preferential NK proliferation in PBL cultures (23). It appears that NKSF may represent a possible or even necessary mediator for activation of NK cells. But, by itself, NKSF does not seem to account for the stimulatory signals necessary for NK proliferation and cytotoxicity provided by the EBV-LCL. Altogether, the data suggest that a contact between the feeder cells and NK cells may be required to allow for efficient NK proliferation. It is thus possible that an NK target cell structure or a nonsecreted membrane-bound factor may play a direct role in stimulation of NK cell proliferation. In the present study, we have described different types of feeder cells, including CD4+ T lymphocytes subpopulations, which display comparable growth-promoting effect on activated NK cells. This approach, which amplifies the generation of activated NK cells in vitro, is being currently extended to include characterization and purification of growth factors that facilitate growth of NK cells. Further studies to identify the functional components that are involved in generating the proliferative response of NK cells may lead to a better understanding of the mechanism underlying NK cell activation and growth and thus to an improved approach to cancer therapy.

470

RABINOWICH

ET AL.

ACKNOWLEDGMENT We thank Dr. Thomas Okarma of Applied Immune Sciences (Menlo Park, CA) for providing anti-CD4 and anti-CD8 monoclonal antibody-coated flasks.

REFERENCES 1. Rosenberg, S. A., Lotze, M. T., Muul, L. M., Chang, A. E., Avis, F. P., Leitman, S., Linehan, W. M., Robertson, C. N., Lee, R. E., Rubin, J. T., Seipp, C. A., Simpson, C. G., and White, D. E., N. Engl. J. Med.

316, 889, 1987.

2. Herberman, R. B., Hiserodt, J. C., Vujanovic, N., Balch, C., Lotzova, E., Bolhuis, R., Golub, S., Lanier, L. L., Philips, J. H., Riccardi, C., Ritz, J., Santoni, A., Schmidt, R. E., and Uchida, A., Immunol. Today& 178, 1987. 3. Ortaldo, J. R., Mason, A., and Overton, R., J. Exp. Med. 164, 1193, 1986. 4. Phillips, J. H., and Lanier, L. L., J. Exp. Med. 164, 814, 1986. 5. Vujanovic, N. L., Herberman, R. B., Maghazachi, A. A., and Hiserodt, J. C., J. Exp. Med. 167, 15, 1987. 6. Melder, R. J., Whiteside, T. L., Vujanovic, N. L., Hiserodt, J. C., and Herberman, R. B., Cancer Res. 48,3461, 1988. 7. Melder, R. J., Walker, E., Herbcrman, R. B., and Whiteside, T. L., J. Leuk. Biol. 48, 163, 1990. 8. Caligiuri, M. A., Zmuidzinas, A., Manley, T. J., Levine, H., Smith, K. A., and Ritz, J., J. Exp. Med. 171, 1509, 1990. 9. Vujanovic, N. L., Herberman, R. B., and Hiserodt, J. C., Cancer Res. 48, 878, 1988. 10. Vujanovic, N. L., Herberman, R. B., Olszowy, M. W., Cramer, D. V., Salup, R. R., Reynolds, C. W., and Hiserodt, J. C., Cancer Res. 48, 884, 1988. 11. Schwarz, R. E., Vujanovic, N. L., and Hiserodt, J. C., Cancer Res. 49, 1441, 1989. 12. Sacchi, M., Vitalo, D., Sedlmayr, P., Rabonowich, H., Johnson, J., Herberman, R. B., and Whiteside, T. L., Int. J. Cancer 199 1, in press. 13. Whiteside, T. L., and Herberman, R. B., Ann. Ital. Inst. Health 1990, in press. 14. Van de Griend, R. J., Van Krimpen, B. A., Ronteltap, C. P. M., and Bolhuis, R. L. H., J. Immunol. 132,3185,

1984.

15. Roberts, K., and Moore, M. A., Eur. .I. Immunol. 15,448, 1985. 16. Hoyer, M., Meineke, T., Lewis, W., Zwilling, B., and Rinehart, J., Cancer Res. 46, 2834, 1986. 17. West, W. H., Cannon, G. B., Kay, H. D., Bonnard, G. D., and Herberman, R. B., .I. Immunol. 118, 355, 1977. 18. Fineman, M. S., Mudawar, F. B., and Geha, R. S., Cell. Immunol. 45, 120, 1979. 19. Pross, H. F., Baines, M. T., Rubin, P., Shragg, E. P., and Patterson, M. S., J. Clin. Immunol. 1, 5 1, 1981.

20. Schwarz, R. E., Iwatsuki, S., Herberman, R. B., and Whiteside, T. L., Cancer Immunol. Immunother. 30, 312, 1989. 21. Whiteside, T. L., Emstoff, M. S., Nair, S., Kirkwood, J. M., and Herberman, R. B., In “Natural Killer Cells: Biology and Clinical Application” (R. E. Schmidt, Ed.) p. 293. 6th Int. Natural Killer Cells Workshop, Goslar, Basel, 1990. 22. Takagi, S., Whiteside, T. L., and Herberman, R. B., The growth-promoting effect of 112-activated lymphocytes on human natural killer (NK) cells. FASEB J. 3, A364, 1989. 23. Peru&a, B., Ramoni, C., Anegon, I., Cuturi, M. C., Faust, J., and Trinchieri, G., Nat. Immun. Cell Growth Regul. 6, 171, 1987. 24. Warren, H. S., J. Immunol. 132,2888, 1984. 25. London, L., Perussia, B., and Trinchieri, G., J. Immunol. 137, 3845, 1986. 26. Hercend, T., Meuer, S. C., Reinherz, E. L., Schlossman, S. F., and Riz, J., J. Immunol. 129, 1299, 1982. 27. Nagler, A., Lanier, L. L., and Phillips, J. H., J. Exp. Med. 171, 1527, 1990. 28. Nagler, A., Lanier, L. L., Cwirla, S., and Phillips, J. H., J. Immunol. 143, 3183, 1989. 29. Phillips, J. H., Le, A. M., and Lanier, L. L., J. Exp. Med. 159, 993, 1984. 30. Vie, J. H., and Miller, R. A., .I. Immunol. 136, 3292, 1986. 31. Kobayashi, M., Fitz, L., Ryan, M., Hewick, R. M., Clark, S. C., Chan, S., Loudon, R., Sherman, F., Perussia, B., and Trinchieri, G., J. Exp. Med. 170, 827, 1989.