Human eosinophils are more toxic than neutrophils in antibody-independent killing Robert L. Roberts, MD, PhD, Bonnie J. Ank, BA, and E. Richard Stiehm, MD Los Angeles, Calif. Eosinophils (EOSs) are implicated in damaging host tissues in diseases such as asthma and eosinophilic gastroenteritis. In the present study, we assessed the cytotoxicity of human EOSs j?om peripheral blood of patients with eosinophilia and from peritoneal fluid of patients undergoing continuous peritoneal dialysis and compared them to normal neutrophils. Cytotoxicity was measured by the release of %zhromium from cultured tumor cells and chicken erythrocytes. Both EOSs and neutrophils were separated on discontinuous Percoll gradients with >95% purity. The granulocytes were activated by preincubation in an ice bath with phorbol myristate acetate and washed before incubation with the target cells. The EOSs lysed signi&antly more tumor cells (K562, Raji, ana’ CEh4 lines) in an 18-hour assay than did neutrophils, and no sign$cant difference was found between the peritoneal and blood EOSs. The EOSs were also much more efficient than neutrophils in lysing chicken erythrocytes when they were activated by granulocyte-macrophage colony-stimulating factor instead of phorbol myristate acetate. Cytolysis by EOSs is mediated by both oxidative and nonoxidative mechanisms, as indicated by experiments with cells from patients with chronic granulomatous disease. Thus, EOSs are much more cytotoxic than neutrophils and potentially much more damaging to patients with eosinophilia. (J ALLERGY CLIN IUMJNOL 1991;87:1105-15.)
EOSs and neutrophils differ greatly in their predominance in particular clinical disorders. Eosinophilia is associated with parasitic infestations, allergic disease, and malignancies, whereas neutrophilia is induced by bacterial infections or endotoxins. These differences indicate these cells are responding to different endogenous factors released by the host in response to the pathogen or allergen or by the pathogen itself. These cells also differ in their functions in host defense. Neutrophils are more efficient at phagocytizing bacteria and killing them within phagolysosomes in contrast to EOSs that rely primarily on extracellular mechanisms to combat the much larger parasites invading body tissues.‘. * In the present study, we compared activated human From the Division of Immunology/Allergy, Department of Pediatrics, University of California-Los Angeles School of Medicine, Los Angeles, Calif. Supported in part by National Institutes of Health Grants HD-09800 and AI-15332, and Cystic Fibrosis Foundation Grants GI 51702 and COO8-08. Received for publication Feb. 6, 1990. Revised Feb. 19, 1991. Accepted for publication Feb. 20, 1991. Reprint requests: R. L. Roberts, MD, Dept. of Pediatrics, UCLA Medical Center, Los Angeles, CA 90024. 1/1/2!3008
I
Abbreviations
EOS: UCLA: CRBC: PMA: ROI: CGD: GM-CSF: PBS: MNC: FMLP CL: NK: Ab: EDTA: FCS: E:T:
used Eosinophil University
1 of California-Los
Angeles
Chicken erythrocyte Phorbol myristate acetate Reactive oxygen intermediates Chronic granulomatous disease Granulocyte-macrophagecolonystimulating factor Phosphate-buffered saline Mononuclear cell N-fonnyl-methionyl-leucylphenylalanine Chemiluminescence Natural killer AtltibOdy Ethylenediaminetetraaceticacid Fetal calf semm Effector : target
EOSs and neutrophils in their ability to lyse cultured tumor cell and erythrocyte targets. We also investigated the role of ROI released by the cells and the influence of cytokines on cytotoxic processes. The activated EOS was much more toxic than the neutrophi1 to all target cells tested, regardless of the origin 1105
1106
Roberts
J. ALLERGY CLIN. IMMiJNOi ‘i ‘NE igy?
et al.
of the EOS or the method used for separation. These studies indicate that the activated EOS may be a strong rival to the neutrophil in both host defense and in-
flammation . MATERlAL AND METHODS Patient population Patients undergoing continuous peritoneal dialysis were recruited from the outpatient population followed by the Division of Nephrology at UCLA. Studies were done in accordance with the UCLA Human Subjects Protection Committee. The patients ranged from 1.5 to 11 years in age (median, 2 years), and eight patients were studied on one or more occasions. Peripheral blood EOSs were obtained from normal volunteers (with EOS counts 45% of peritoneal cells). None of the patients had peritonitis or other infections when they were studied. Specimens were obtained after the peritoneal cavity had been fully drained and the collection bag had been disconnected from the patient. The cause of eosinophiJia in peritoneal dialysis patients, as noted previously, is not known and is not in response to parasitic infestation or identifiable allergens. To obtain highly enriched EOS populations, the pentoneal cells were concentrated by centrifugation, washed in PBS, and then layered over discontinuous Percoli gradients in a manner similar to those prepared for the separation of blood neutrophils. Four concentrations of Percolt (45%. 55%. 65%, and 75%) were layered into a 15 ml polystyrene tube to form a discontinuous gradient with 3 ml of each concentration of Percoll per layer. Peritoneal cells (2 to 4 x IO8 in 3 ml of PBS) were layered over the gradient and then spun at 400 g for 15 minutes at room temperature. Erythrocytes (if these were present) pelleted to the bottom of the tube, and the peritoneal leukocytes fomled bands at the interfaces of Percoll layers. Hypotonic lysis was sometimes needed to remove erythrocytes from the leukocyte populations. The cell types recovered from each band at successive Percoll interfaces and their functional properties have been
VQLUMF a7 NUMBER 6
EOS
K562
LYSIS
BY NEUTROPHILS
AND
cytotoxicity
1107
EOSINOPHILS
80
60
q q
NEUTROPHILS GRANULOCYTES
fY=J BLOOD
q
PERITONEAL
*
EFFECTOR
: TARGET
EOS EOS
p95%) from normal adult donors (n = 12). (2) mixed granulocyte populations (20% to 30% EOSs) from eosinophilic patients (n = 6). (3) purified EOSs (>95%) from same patients, and (4) peritoneal EOSs (n = 6). All the effector cells had been activated by prior exposure to PMA and washed before placing with K562 cells. The p values are in comparison to normal neutrophils at same E : T ratio.
previously reported., It should be noted that separation of the peritoneal EOSsdiffers from separation of blood EOSs in that treatment with FMLP is not required becauseof the high starting percentageof EOSsin the peritoneal fluid. The different bands are collected separately by pipette and washed in PBS before use in the assays. In these experiments, only the densest EOS population (between the 75% and 65% Percoll layers)were used for assays.Previous studies found that this population is most similar to peripheral blood EOSs in function and density and was >95% EOSsin ~urity.~ CL Assaysfor CL were done on the LKB (LKB, Bromma, Sweden) 1250 luminometer at 37” C with full scale representing 10 mV. Cells were suspendedin 1 ml of N-2hydroxyethylpiperazine-N’-2-ethanesulfonic acid buffer containing calcium, magnesium (unless it is otherwise stated), and luminol (10m5mol/L). Cells were incubatedfor 10 minutes at 37” C before adding any stimuli. The peak (or peaksfor biphasicresponses)was recordedin millivolts.s Cytotoxicity
assays
Cytotoxicity of EOSs and neutrophils was measuredby releaseof chromium 51 from target cells. The tumor cell lines (K562, Raji, and CEM) and CRBCs were used as target cells, and assayswere performed in %-well microtiter
plates aspreviously described.9When tumor cells were used as targets, 1 x 104tumor cells were added to each well (final concentration, 5 x 104iml) with RPM1 1640 with 10% FCS as medium. To achieve a 25: 1 E:T ratio, 2.5 x lo5 granulocyteswere added toeach well, 1.O X 10’ foralO:lratioand5.0 X 104fora5:1ratio.(Finalvolume was 0.2 ml for all.) When CRBCs were used as targets, 2.5 X 104of CRBCs were added to each well with a proportional increase in the effecters. (Final volume still 0.2 ml per well.) In addition “cold” sheeperythrocytes(0.025% by volume) were added to each well when CRBCs were used to decreasespontaneous lysis with RPM1 1640 with 2.0% FCS as medium. Granulocyteswere activated in most experiments by incubating in 100 nglml of PMA for 10 minutes at 4” C. The cellswere then washed in cold PBS and resuspendedin cold RPM1 1640 media supplementedwith heat-inactivatedFCS (10% for tumor cells studies and 2% for CRBC studies). The cell suspensionswere kept in an ice bath until the start of the assaywhen granulocyteswere placed in the well with the target cells. The granulocytesand target cells were then incubated at 37” C for 18 hours unlessit is otherwise stated. EOSsand neutrophils exposed to PMA by this method appear to be fully activated when they are warmed to 37” C in CL assays. Specific lysis of target cells was calculated by the formula, Specific lysis = [(experimental cpm-spontaneous
1108
J. ALLERGY CLIN. IMMUNOL. JUNE 1991
Roberts et al.
RAJI
CELL
LYSIS
BY
PMA-AETlVATeD
5:l
IO:1 EFFECTOR
CELL
tM%AMUL~~S
LYSIS
: TARGET
BY
25:l RATIO
PMA-ACTIVAT~ED
ORA?WLUCYTt5S +
60
0 1O:l EFFEGTOR FIG. 2. A, Lysis of in comparison to exposure to PMA, are in comparison
25:l : TARGET
RATIO
Raji cells by purified blood EOSs. B, Lysis of CEM cells by peritoneal EOSs lysis by normal neutrophils. The effector cells had been activated by prior and incubation of effecters with targets was 18 hours at 37” C. The p vdues to normal neutrophils at same ratio.
cpm)/(total cpm-spontaneouscpm)] X 100. The meantotal release in counts per minute and mean spontaneous release in an l&hour assay for target cells used in this series of experiments are as follows: K562, total release 10,566 rt 1245, spontaneous 3104 * 465; CRBC, total release 1673 -+ 10, spontaneous 148 + 23; Raji, total release3195 -I- 492, spontaneous861 -+ 137; andGEM, total release8981 f 1227, spontaneous2851 ? 298. These results are expressedas mean counts per minute with standard error with a machine background of approximately 100
cpm. The Student’s t test (two-tailed, unpaired) was used to compare the reported means (with sumdard error) for the different variables. RESULTS Lysis of lc!m! cetk
In Fig. 1, the killing of KS2 cells by mmqdmutrophils (35% neutmphih; n = 12) is c t0 that of unfractionated granulcqte populatians from
VOLUME 67 NUMBER 6
EOS
CRBC
LYSIS
BY
PMA-ACTIVATED
cytotoxicity
1109
GRANULOCYTES
50 ........................................................................................................................................... 1 40
10 . . .. ... . .. 1q
NEUTROPHILS EOSlNOPHlLS
1
2
t
1 :lO
1 :l
EFFECTOR
CRBC
LYSIS
40--
BY GM-CSF
q
1O:l
: TARGET
RATIO
ACTIVATED
GRANULOCYTES
NEUT 51
20 -’
10 -*
B BUFFER ALONE
GM-CSF 500
FIG. 3. A, Lysis of CRBC by normal neutrophils and peritoneal EOSs activated by prior exposure to PMA. 9, Lysis of CRBC by normal neutrophils and purified blood EOSs activated by GM-CSF (500 nmol/L). The GM-CSF was present throughout the entire 18-hour incubation.
patients (n = 6) with eosinophilia (20% to 30% EOSs), peripheral blood EOSs isolated from these same patients (>95% EOSs), and peritoneal EOSs isolated from the peritoneal fluid of patients undergoing continuous peritoneal dialysis (n = 6). The blood EOSs were isolated by exposure of whole blood to PMLP; the peritoneal EOSs were isolated by concentrating the cells followed by separation on a Percoll gradient (see MATERIAL AND METHODS). All cell populations had been activated by incubation with PMA before the assay, and the effecters
and targets were incubated together for 18 hours at 37” c. The EOSs were more toxic than the neutrophils at all E:T cell ratios tested regardless of the origin of the EOS or method used for separation. The discrepancy between EOSs and neutrophils was more notable at lower E: T ratios, such as at 5 : 1 in which both the blood and peritoneal EOSs killed five times more targets than did the neutrophils. The unfractionated granulocytes from patients that contained a mixture of neutrophils (70% to 80%) and EOSs (20% to 30%)
1110
Roberts
et al.
K562
J ALLERGY CLIN. IMMWNOC. JUNE 1991
LYSIS
BY NGMMHAL
CGE) GRAWkUCYl-ES
A
q
60
EFFECTOR
: TARGET
FIG. 4. Lysis of K562 cells by neutrophils neutrophils and EOSs from a patient with incubation was for 18 hours.
and CGD.
also killed more target cells than normal neutrophils at the 5 : 1 ratio (7.4% r?~1.7% for neutrophils compared to 17.1% -t 3.3% for mixed granulocytes) but not at higher ratios. No significant lysis (
EOSs and neutrophils differ greatly in their predominance in particular clinical disorders. Eosinophilia is associated with parasitic infestations, allergic disease, and malignancies, whereas neutrophilia is induced by bacterial infections or endotoxins. These differences indicate these cells are responding to different endogenous factors released by the host in response to the pathogen or allergen or by the pathogen itself. These cells also differ in their functions in host defense. Neutrophils are more efficient at phagocytizing bacteria and killing them within phagolysosomes in contrast to EOSs that rely primarily on extracellular mechanisms to combat the much larger parasites invading body tissues.‘. * In the present study, we compared activated human From the Division of Immunology/Allergy, Department of Pediatrics, University of California-Los Angeles School of Medicine, Los Angeles, Calif. Supported in part by National Institutes of Health Grants HD-09800 and AI-15332, and Cystic Fibrosis Foundation Grants GI 51702 and COO8-08. Received for publication Feb. 6, 1990. Revised Feb. 19, 1991. Accepted for publication Feb. 20, 1991. Reprint requests: R. L. Roberts, MD, Dept. of Pediatrics, UCLA Medical Center, Los Angeles, CA 90024. 1/1/2!3008
I
Abbreviations
EOS: UCLA: CRBC: PMA: ROI: CGD: GM-CSF: PBS: MNC: FMLP CL: NK: Ab: EDTA: FCS: E:T:
used Eosinophil University
1 of California-Los
Angeles
Chicken erythrocyte Phorbol myristate acetate Reactive oxygen intermediates Chronic granulomatous disease Granulocyte-macrophagecolonystimulating factor Phosphate-buffered saline Mononuclear cell N-fonnyl-methionyl-leucylphenylalanine Chemiluminescence Natural killer AtltibOdy Ethylenediaminetetraaceticacid Fetal calf semm Effector : target
EOSs and neutrophils in their ability to lyse cultured tumor cell and erythrocyte targets. We also investigated the role of ROI released by the cells and the influence of cytokines on cytotoxic processes. The activated EOS was much more toxic than the neutrophi1 to all target cells tested, regardless of the origin 1105
1106
Roberts
J. ALLERGY CLIN. IMMiJNOi ‘i ‘NE igy?
et al.
of the EOS or the method used for separation. These studies indicate that the activated EOS may be a strong rival to the neutrophil in both host defense and in-
flammation . MATERlAL AND METHODS Patient population Patients undergoing continuous peritoneal dialysis were recruited from the outpatient population followed by the Division of Nephrology at UCLA. Studies were done in accordance with the UCLA Human Subjects Protection Committee. The patients ranged from 1.5 to 11 years in age (median, 2 years), and eight patients were studied on one or more occasions. Peripheral blood EOSs were obtained from normal volunteers (with EOS counts 45% of peritoneal cells). None of the patients had peritonitis or other infections when they were studied. Specimens were obtained after the peritoneal cavity had been fully drained and the collection bag had been disconnected from the patient. The cause of eosinophiJia in peritoneal dialysis patients, as noted previously, is not known and is not in response to parasitic infestation or identifiable allergens. To obtain highly enriched EOS populations, the pentoneal cells were concentrated by centrifugation, washed in PBS, and then layered over discontinuous Percoli gradients in a manner similar to those prepared for the separation of blood neutrophils. Four concentrations of Percolt (45%. 55%. 65%, and 75%) were layered into a 15 ml polystyrene tube to form a discontinuous gradient with 3 ml of each concentration of Percoll per layer. Peritoneal cells (2 to 4 x IO8 in 3 ml of PBS) were layered over the gradient and then spun at 400 g for 15 minutes at room temperature. Erythrocytes (if these were present) pelleted to the bottom of the tube, and the peritoneal leukocytes fomled bands at the interfaces of Percoll layers. Hypotonic lysis was sometimes needed to remove erythrocytes from the leukocyte populations. The cell types recovered from each band at successive Percoll interfaces and their functional properties have been
VQLUMF a7 NUMBER 6
EOS
K562
LYSIS
BY NEUTROPHILS
AND
cytotoxicity
1107
EOSINOPHILS
80
60
q q
NEUTROPHILS GRANULOCYTES
fY=J BLOOD
q
PERITONEAL
*
EFFECTOR
: TARGET
EOS EOS
p95%) from normal adult donors (n = 12). (2) mixed granulocyte populations (20% to 30% EOSs) from eosinophilic patients (n = 6). (3) purified EOSs (>95%) from same patients, and (4) peritoneal EOSs (n = 6). All the effector cells had been activated by prior exposure to PMA and washed before placing with K562 cells. The p values are in comparison to normal neutrophils at same E : T ratio.
previously reported., It should be noted that separation of the peritoneal EOSsdiffers from separation of blood EOSs in that treatment with FMLP is not required becauseof the high starting percentageof EOSsin the peritoneal fluid. The different bands are collected separately by pipette and washed in PBS before use in the assays. In these experiments, only the densest EOS population (between the 75% and 65% Percoll layers)were used for assays.Previous studies found that this population is most similar to peripheral blood EOSs in function and density and was >95% EOSsin ~urity.~ CL Assaysfor CL were done on the LKB (LKB, Bromma, Sweden) 1250 luminometer at 37” C with full scale representing 10 mV. Cells were suspendedin 1 ml of N-2hydroxyethylpiperazine-N’-2-ethanesulfonic acid buffer containing calcium, magnesium (unless it is otherwise stated), and luminol (10m5mol/L). Cells were incubatedfor 10 minutes at 37” C before adding any stimuli. The peak (or peaksfor biphasicresponses)was recordedin millivolts.s Cytotoxicity
assays
Cytotoxicity of EOSs and neutrophils was measuredby releaseof chromium 51 from target cells. The tumor cell lines (K562, Raji, and CEM) and CRBCs were used as target cells, and assayswere performed in %-well microtiter
plates aspreviously described.9When tumor cells were used as targets, 1 x 104tumor cells were added to each well (final concentration, 5 x 104iml) with RPM1 1640 with 10% FCS as medium. To achieve a 25: 1 E:T ratio, 2.5 x lo5 granulocyteswere added toeach well, 1.O X 10’ foralO:lratioand5.0 X 104fora5:1ratio.(Finalvolume was 0.2 ml for all.) When CRBCs were used as targets, 2.5 X 104of CRBCs were added to each well with a proportional increase in the effecters. (Final volume still 0.2 ml per well.) In addition “cold” sheeperythrocytes(0.025% by volume) were added to each well when CRBCs were used to decreasespontaneous lysis with RPM1 1640 with 2.0% FCS as medium. Granulocyteswere activated in most experiments by incubating in 100 nglml of PMA for 10 minutes at 4” C. The cellswere then washed in cold PBS and resuspendedin cold RPM1 1640 media supplementedwith heat-inactivatedFCS (10% for tumor cells studies and 2% for CRBC studies). The cell suspensionswere kept in an ice bath until the start of the assaywhen granulocyteswere placed in the well with the target cells. The granulocytesand target cells were then incubated at 37” C for 18 hours unlessit is otherwise stated. EOSsand neutrophils exposed to PMA by this method appear to be fully activated when they are warmed to 37” C in CL assays. Specific lysis of target cells was calculated by the formula, Specific lysis = [(experimental cpm-spontaneous
1108
J. ALLERGY CLIN. IMMUNOL. JUNE 1991
Roberts et al.
RAJI
CELL
LYSIS
BY
PMA-AETlVATeD
5:l
IO:1 EFFECTOR
CELL
tM%AMUL~~S
LYSIS
: TARGET
BY
25:l RATIO
PMA-ACTIVAT~ED
ORA?WLUCYTt5S +
60
0 1O:l EFFEGTOR FIG. 2. A, Lysis of in comparison to exposure to PMA, are in comparison
25:l : TARGET
RATIO
Raji cells by purified blood EOSs. B, Lysis of CEM cells by peritoneal EOSs lysis by normal neutrophils. The effector cells had been activated by prior and incubation of effecters with targets was 18 hours at 37” C. The p vdues to normal neutrophils at same ratio.
cpm)/(total cpm-spontaneouscpm)] X 100. The meantotal release in counts per minute and mean spontaneous release in an l&hour assay for target cells used in this series of experiments are as follows: K562, total release 10,566 rt 1245, spontaneous 3104 * 465; CRBC, total release 1673 -+ 10, spontaneous 148 + 23; Raji, total release3195 -I- 492, spontaneous861 -+ 137; andGEM, total release8981 f 1227, spontaneous2851 ? 298. These results are expressedas mean counts per minute with standard error with a machine background of approximately 100
cpm. The Student’s t test (two-tailed, unpaired) was used to compare the reported means (with sumdard error) for the different variables. RESULTS Lysis of lc!m! cetk
In Fig. 1, the killing of KS2 cells by mmqdmutrophils (35% neutmphih; n = 12) is c t0 that of unfractionated granulcqte populatians from
VOLUME 67 NUMBER 6
EOS
CRBC
LYSIS
BY
PMA-ACTIVATED
cytotoxicity
1109
GRANULOCYTES
50 ........................................................................................................................................... 1 40
10 . . .. ... . .. 1q
NEUTROPHILS EOSlNOPHlLS
1
2
t
1 :lO
1 :l
EFFECTOR
CRBC
LYSIS
40--
BY GM-CSF
q
1O:l
: TARGET
RATIO
ACTIVATED
GRANULOCYTES
NEUT 51
20 -’
10 -*
B BUFFER ALONE
GM-CSF 500
FIG. 3. A, Lysis of CRBC by normal neutrophils and peritoneal EOSs activated by prior exposure to PMA. 9, Lysis of CRBC by normal neutrophils and purified blood EOSs activated by GM-CSF (500 nmol/L). The GM-CSF was present throughout the entire 18-hour incubation.
patients (n = 6) with eosinophilia (20% to 30% EOSs), peripheral blood EOSs isolated from these same patients (>95% EOSs), and peritoneal EOSs isolated from the peritoneal fluid of patients undergoing continuous peritoneal dialysis (n = 6). The blood EOSs were isolated by exposure of whole blood to PMLP; the peritoneal EOSs were isolated by concentrating the cells followed by separation on a Percoll gradient (see MATERIAL AND METHODS). All cell populations had been activated by incubation with PMA before the assay, and the effecters
and targets were incubated together for 18 hours at 37” c. The EOSs were more toxic than the neutrophils at all E:T cell ratios tested regardless of the origin of the EOS or method used for separation. The discrepancy between EOSs and neutrophils was more notable at lower E: T ratios, such as at 5 : 1 in which both the blood and peritoneal EOSs killed five times more targets than did the neutrophils. The unfractionated granulocytes from patients that contained a mixture of neutrophils (70% to 80%) and EOSs (20% to 30%)
1110
Roberts
et al.
K562
J ALLERGY CLIN. IMMWNOC. JUNE 1991
LYSIS
BY NGMMHAL
CGE) GRAWkUCYl-ES
A
q
60
EFFECTOR
: TARGET
FIG. 4. Lysis of K562 cells by neutrophils neutrophils and EOSs from a patient with incubation was for 18 hours.
and CGD.
also killed more target cells than normal neutrophils at the 5 : 1 ratio (7.4% r?~1.7% for neutrophils compared to 17.1% -t 3.3% for mixed granulocytes) but not at higher ratios. No significant lysis (
