Effect of Granulocyte-Macrophage Colony Stimulating Factor on Endothelial Antigenicity Dariusz Leszczynski and Pekka Hiiyry
ABSTRACT: We have investigated if recombinant granulocyte-macrophage colony stimulating factor (GM-CSF), alone or in concert with recombinant 5, interferon, affects the endothelial cell expression of class I major histocompatibility complex antigen. Results obtained show that the GM-CSF increases class I expression on the endothelial cell in a time- and dose-dependent manner. The effect of interferon 5' on class I expression is diminished in the presence of GM-CSF, cAMP, or prostaglandin E2, but is increased in the presence of cGMP. N-(2-guanidinoethyl)-5-isoquinolinesulfonamide (HA 1004), an inhibitor of cAMP- and cGMP-dependent protein kinases, abolished GM-CSF-induced class I expression, while indomethacin increased it. When added to the endothelial cell cultures together with interferon 5, G M-CSF, HA 1004 as well as indomethacin abolished the inhibitory effect of GM-CSF on interferon 5,-induced class I expression. The results suggest that GM-CSF diminishes effect of interferon 5, on class I major histocompatibility complex expression on the endothelial cell by inducing production of rostacyclin. This, in turn, induces cAMP as a second messenger, which then leads to the events inhibiting expression of class I major histocompatibility complex antigen. ABBREVIATIONS
GM-CSF
granulocyte-macrophage colony stimulating factor HA 1004 N-(2-guanidinoethyl)-5isoquinolinesulfonamide
IFN-5, interferon 5, MHC major histocompability complex PGs prostaglandins TNFa tumor necrosis factor
INTRODUCTION Granulocyte-macrophage colony stimulating factor (GM-CSF) is a glycoprotein growth factor regulating proliferation and differentiation of hematopoetic and nonhematopoetic cells [1,2]. It has been shown that not only all bone marrowderived cells able to produce this cytokine, but such cells as endothelial cells continuously secrete GM-CSF [3]. Therefore, it is possible that endotheliumderived GM-CSF, similarly to interleukin l a and prostaglandins (PGs), can participate in autocrine regulation of endothelial functions. We have examined the effect of recombinant GM-CSF, alone or in combination with interferon y (IFN-7), on the class I major histocompatibility complex (MHC antigen expression by endothelial cells isolated from rat heart. MATERIALS AND METHODS
Materials and methods used were described in detail previously [4]. From the Transplantation Laboratory, University of Helsinki, Helsinki, Finland. Address reprint requests to Dr. Dariusz Leszczynski, Transplantation Laboratory, University of Helsinki, Haartmaninkatu 3, SF-00290 Helsinki, Finland. Received December 1, 1989; acceptedJanuary 3, 1990.
Human Immunology 28, 175-178 (1990) © American Society for Histocompatibility and Immunogenetics, 1990
175 0198-8859/90/$3.50
176
D. Leszczynski and P. H/iyry % of class 100
expressing EC
Treatment glFN
80 glFN/GM-CSF
60 40
20
GM-CSF
0 1
3
a
5
7
days of treatment
Treatment nil
Time
1I
3 5 I
glFN
i
3 5 I
GM-CSF
3 5
glFN/GM-CSF
3 5
1
0
b
200
400
600
--t 800
6-keto-PCFle (pg)
(A) Time-dependent modulation of the class I antigen expression by endothelial cells treated with GM-CSF (1000 U/ml) or IFN-7 (100 U/ml) or both. (B) Regulation of prostacyclin production by endothelial cells in the presence of GM-CSF (1000 U/ml), IFN-7 (100 U/ml), or both. FIGURE 1
RESULTS A N D D I SC U SSI O N GM-CSF increased expression of class I antigen in dose- (not shown) and timedependent manner [Figure I(A)]. However, this effect was far lower than effect of IFN-7 [Figure I(A)]. When GM-CSF and IFN-7 were used together for the treatment of endothelium, GM-CSF (at optimal concentration of 1000 U/ml) did not have any effect on IFN-7-induced (100 U/ml) class I expression during first 3 days of treatment. However, there was marked reduction in the number of class I-expressing endothelial cells after 5-7 days of the treatment as compared with cells treated with IFN-5, alone [Figure I(A)]. Simultaneously, we observed a decrease in the class I antigen density on the cell surface (not shown). These results suggest that GM-CSF may provide some inhibitory signal diminishing IFN-7 induced class I MHC antigen expression by endothelial cells. Moreover, it suggests that GM-CSF secreted by endothelial cells at the site of inflammation
Granulocyte-Macrophage CSF Endothelial Effect
177
might be involved in the negative feedback mechanism regulating endothelial antigenicity. It has been shown by us [5] and by others [6] that cytokine-induced antigen expression can be diminished by prostaglandins. Therefore, we examined the possibility that endothelium-secreted PGs might diminish the observed effect of GM-CSF on the induction of class I MHC antigen by the endothelium. First, we examined the ability of GM-CSF to induce the secretion of prostaglandins by endothelium [Figure I(B)]; such a phenomenon has been already observed for macrophages [7]. GM-CSF increased the secretion of prostaglandins by endothelium, but this effect was slow and required up to 5 days to obtain significant increase in amount of PGs present in the culture medium from 200 to 400 pg/ml of culture medium. IFN-~/did not induce PG synthesis. When IFN-'y and GMCSF were combined, they had a synergistic effect on PG secretion, and maximal effect was observed on day 5 (700 pg/ml), which correlates in time with the decrease of class I expression. These results show that on day 5 of treatment of endothelial cells with GM-CSF/IFN-~/, PGs are present in large quantities in the culture medium, and they may alter some of the endothelial cell properties such as antigenicity. To demonstrate the role of endothelial cell-secreted PGs on GM-CSF-induced class I MHC antigen expression we used indomethacin. This inhibitor of PG synthesis had no effect on the GM-CSF-induced class I expression after 3 days of treatment. This points out a possibility that basal level of PGs secreted by endothelium (200 pg/ml) is too low to control endothelial expression of class I antigen stimulated by optimal concentration of GM-CSF. However, after 5 days of treatment of the EC with GM-CSF the presence of indomethacin in the culture medium had a strong enhancing effect on the class I antigen expression. Similarly, indomethacin did not have any effect on class I expression induced by combination of IFN-~/ and GM-CSF after 3 days of treatment. However, it abolished an inhibitory effect of GM-CSF on IFN-~/-induced class I expression observed on days 5 and 7 of the treatment. The effect of indomethacin is most probably due to the decrease in the level of PGs in the culture medium, which otherwise are present in large amounts after 5 days of treatment of endothelium with GM-CSF or GM-CSF/IFN-~/. It has been shown that prostaglandins use cAMP as a second messenger and transduce their signals via cAMP-dependent protein kinases [8]. On the other hand GM-CSF has been shown in neutrophils to act via cGMP-dependent protein kinases [9]. Therefore, we examined the effects of the HA 1004, blocking specifically the activity of cAMP- and cGMP-dependent protein kinases [10], on the class I induction by the GM-CSF or GM-CSF/IFN-% As a control we performed experiments where endothelial cells were treated with cytokine(s) in the presence of cAMP or cGMP or prostaglandin E2 (PGE2). Results of the experiments are presented in Table 1. HA 1004 abolished class I antigen expression induced by GM-CSF. Effects of the cyclic nucleotides were opposite to each other: cAMP inhibited class I expression whereas cGMP had a stimulatory effect. Increase of class I expression induced by IFN-~ was effectively inhibited by cAMP and PGE2, whereas cGMP increased class I expression. Finally, HA 1004 abolished the inhibitory effect of the GM-CSF on the expression of class I antigen induced by IFN-'y. This suggests that most probably HA 1004 abolished both the inhibitory effect of GM-CSF-induced prostaglandins and the stimulatory effect of GM-CSFinduced intracellular cGMP on the IFN-y-induction of class I antigen. These findings demonstrate that PGs, acting via cAMP-dependent protein kinase, diminish the effect of GM-CSF and GM-CSF/IFN-~/on the class I expres-
178
D. Leszczynski and P. H~iyry
TABLE
1
P e r c e n t a g e s o f class I a n t i g e n - e x p r e s s i n g a e n d o t h e l i a l cells after t r e a t m e n t with d i f f e r e n t c o m p o u n d s Class I-expressing endothelial cells, %
Treatment b Nil GM-CSF GM-CSF GM-CSF GM-CSF GM-CSF GM-CSF
+ + + + +
(db)cAMP (db)cGMP PGE2 IM HA 1004
Without IFN-5,
With IFN-y
li
92 75 75 100 78 90 90
32 24 40 nd 42 16
Measured by fluorescenceactivatedcell sorter. hTreatment for 5 days:GM-CSF,1000 U/ml; IFN-'y,100 U/ml; dibutyryl(db)cAMPand (db)cGMP, 2 g.M;PGE2, 10-7 M; indomethacin(IM), 1 /zg/ml;HA 1004, 2 /a,M.
sion. M o r e o v e r , they suggest that s t i m u l a t i o n o f class I e x p r e s s i o n i n d u c e d by G M - C S F is m e d i a t e d via c G M P - d e p e n d e n t p r o t e i n kinase.
ACKNOWLEDGMENTS This work was supported by the grants from the Academy of Finland, the Finnish Cancer Foundation, the Finnish Kidney Foundation, the University of Helsinki, Helsinki, and the Research and Science Foundation of Farmos, Turku, Finland.
REFERENCES
1. MetcalfD: Blood 67:257, 1986. 2. Dedhar S, Gaboury L, Galloway P, et al: Proc Natl Acad Sci USA 85:9253, 1988. 3. Malon DG, Pierce JH, Falko JP, et al: Blood 71:684, 1988. 4. Leszczynski D, H~iyry P: Submitted. 5. Leszczynski D, H/iyry P: FEBS Lett 242:383, 1989. 6. Snyder DS, Belier DI, Unanue ER: Nature 299:163, 1982. 7. Hancock WW, Pleau ME, Kobzik L: J Immunol 140:3021, 1988. 8. Franks DJ, Macmanus JP, Whitfield JF: Biochem Biophys Res Commun 44:1177, 1971. 9. Coffey GR, Davis JS, Djeu JY: J Immunol 140:2695, 1988. 10. Asano T, Hidaka H: J Pharmacol Exp Therap 231:141, 1984.
ABSTRACT: We have investigated if recombinant granulocyte-macrophage colony stimulating factor (GM-CSF), alone or in concert with recombinant 5, interferon, affects the endothelial cell expression of class I major histocompatibility complex antigen. Results obtained show that the GM-CSF increases class I expression on the endothelial cell in a time- and dose-dependent manner. The effect of interferon 5' on class I expression is diminished in the presence of GM-CSF, cAMP, or prostaglandin E2, but is increased in the presence of cGMP. N-(2-guanidinoethyl)-5-isoquinolinesulfonamide (HA 1004), an inhibitor of cAMP- and cGMP-dependent protein kinases, abolished GM-CSF-induced class I expression, while indomethacin increased it. When added to the endothelial cell cultures together with interferon 5, G M-CSF, HA 1004 as well as indomethacin abolished the inhibitory effect of GM-CSF on interferon 5,-induced class I expression. The results suggest that GM-CSF diminishes effect of interferon 5, on class I major histocompatibility complex expression on the endothelial cell by inducing production of rostacyclin. This, in turn, induces cAMP as a second messenger, which then leads to the events inhibiting expression of class I major histocompatibility complex antigen. ABBREVIATIONS
GM-CSF
granulocyte-macrophage colony stimulating factor HA 1004 N-(2-guanidinoethyl)-5isoquinolinesulfonamide
IFN-5, interferon 5, MHC major histocompability complex PGs prostaglandins TNFa tumor necrosis factor
INTRODUCTION Granulocyte-macrophage colony stimulating factor (GM-CSF) is a glycoprotein growth factor regulating proliferation and differentiation of hematopoetic and nonhematopoetic cells [1,2]. It has been shown that not only all bone marrowderived cells able to produce this cytokine, but such cells as endothelial cells continuously secrete GM-CSF [3]. Therefore, it is possible that endotheliumderived GM-CSF, similarly to interleukin l a and prostaglandins (PGs), can participate in autocrine regulation of endothelial functions. We have examined the effect of recombinant GM-CSF, alone or in combination with interferon y (IFN-7), on the class I major histocompatibility complex (MHC antigen expression by endothelial cells isolated from rat heart. MATERIALS AND METHODS
Materials and methods used were described in detail previously [4]. From the Transplantation Laboratory, University of Helsinki, Helsinki, Finland. Address reprint requests to Dr. Dariusz Leszczynski, Transplantation Laboratory, University of Helsinki, Haartmaninkatu 3, SF-00290 Helsinki, Finland. Received December 1, 1989; acceptedJanuary 3, 1990.
Human Immunology 28, 175-178 (1990) © American Society for Histocompatibility and Immunogenetics, 1990
175 0198-8859/90/$3.50
176
D. Leszczynski and P. H/iyry % of class 100
expressing EC
Treatment glFN
80 glFN/GM-CSF
60 40
20
GM-CSF
0 1
3
a
5
7
days of treatment
Treatment nil
Time
1I
3 5 I
glFN
i
3 5 I
GM-CSF
3 5
glFN/GM-CSF
3 5
1
0
b
200
400
600
--t 800
6-keto-PCFle (pg)
(A) Time-dependent modulation of the class I antigen expression by endothelial cells treated with GM-CSF (1000 U/ml) or IFN-7 (100 U/ml) or both. (B) Regulation of prostacyclin production by endothelial cells in the presence of GM-CSF (1000 U/ml), IFN-7 (100 U/ml), or both. FIGURE 1
RESULTS A N D D I SC U SSI O N GM-CSF increased expression of class I antigen in dose- (not shown) and timedependent manner [Figure I(A)]. However, this effect was far lower than effect of IFN-7 [Figure I(A)]. When GM-CSF and IFN-7 were used together for the treatment of endothelium, GM-CSF (at optimal concentration of 1000 U/ml) did not have any effect on IFN-7-induced (100 U/ml) class I expression during first 3 days of treatment. However, there was marked reduction in the number of class I-expressing endothelial cells after 5-7 days of the treatment as compared with cells treated with IFN-5, alone [Figure I(A)]. Simultaneously, we observed a decrease in the class I antigen density on the cell surface (not shown). These results suggest that GM-CSF may provide some inhibitory signal diminishing IFN-7 induced class I MHC antigen expression by endothelial cells. Moreover, it suggests that GM-CSF secreted by endothelial cells at the site of inflammation
Granulocyte-Macrophage CSF Endothelial Effect
177
might be involved in the negative feedback mechanism regulating endothelial antigenicity. It has been shown by us [5] and by others [6] that cytokine-induced antigen expression can be diminished by prostaglandins. Therefore, we examined the possibility that endothelium-secreted PGs might diminish the observed effect of GM-CSF on the induction of class I MHC antigen by the endothelium. First, we examined the ability of GM-CSF to induce the secretion of prostaglandins by endothelium [Figure I(B)]; such a phenomenon has been already observed for macrophages [7]. GM-CSF increased the secretion of prostaglandins by endothelium, but this effect was slow and required up to 5 days to obtain significant increase in amount of PGs present in the culture medium from 200 to 400 pg/ml of culture medium. IFN-~/did not induce PG synthesis. When IFN-'y and GMCSF were combined, they had a synergistic effect on PG secretion, and maximal effect was observed on day 5 (700 pg/ml), which correlates in time with the decrease of class I expression. These results show that on day 5 of treatment of endothelial cells with GM-CSF/IFN-~/, PGs are present in large quantities in the culture medium, and they may alter some of the endothelial cell properties such as antigenicity. To demonstrate the role of endothelial cell-secreted PGs on GM-CSF-induced class I MHC antigen expression we used indomethacin. This inhibitor of PG synthesis had no effect on the GM-CSF-induced class I expression after 3 days of treatment. This points out a possibility that basal level of PGs secreted by endothelium (200 pg/ml) is too low to control endothelial expression of class I antigen stimulated by optimal concentration of GM-CSF. However, after 5 days of treatment of the EC with GM-CSF the presence of indomethacin in the culture medium had a strong enhancing effect on the class I antigen expression. Similarly, indomethacin did not have any effect on class I expression induced by combination of IFN-~/ and GM-CSF after 3 days of treatment. However, it abolished an inhibitory effect of GM-CSF on IFN-~/-induced class I expression observed on days 5 and 7 of the treatment. The effect of indomethacin is most probably due to the decrease in the level of PGs in the culture medium, which otherwise are present in large amounts after 5 days of treatment of endothelium with GM-CSF or GM-CSF/IFN-~/. It has been shown that prostaglandins use cAMP as a second messenger and transduce their signals via cAMP-dependent protein kinases [8]. On the other hand GM-CSF has been shown in neutrophils to act via cGMP-dependent protein kinases [9]. Therefore, we examined the effects of the HA 1004, blocking specifically the activity of cAMP- and cGMP-dependent protein kinases [10], on the class I induction by the GM-CSF or GM-CSF/IFN-% As a control we performed experiments where endothelial cells were treated with cytokine(s) in the presence of cAMP or cGMP or prostaglandin E2 (PGE2). Results of the experiments are presented in Table 1. HA 1004 abolished class I antigen expression induced by GM-CSF. Effects of the cyclic nucleotides were opposite to each other: cAMP inhibited class I expression whereas cGMP had a stimulatory effect. Increase of class I expression induced by IFN-~ was effectively inhibited by cAMP and PGE2, whereas cGMP increased class I expression. Finally, HA 1004 abolished the inhibitory effect of the GM-CSF on the expression of class I antigen induced by IFN-'y. This suggests that most probably HA 1004 abolished both the inhibitory effect of GM-CSF-induced prostaglandins and the stimulatory effect of GM-CSFinduced intracellular cGMP on the IFN-y-induction of class I antigen. These findings demonstrate that PGs, acting via cAMP-dependent protein kinase, diminish the effect of GM-CSF and GM-CSF/IFN-~/on the class I expres-
178
D. Leszczynski and P. H~iyry
TABLE
1
P e r c e n t a g e s o f class I a n t i g e n - e x p r e s s i n g a e n d o t h e l i a l cells after t r e a t m e n t with d i f f e r e n t c o m p o u n d s Class I-expressing endothelial cells, %
Treatment b Nil GM-CSF GM-CSF GM-CSF GM-CSF GM-CSF GM-CSF
+ + + + +
(db)cAMP (db)cGMP PGE2 IM HA 1004
Without IFN-5,
With IFN-y
li
92 75 75 100 78 90 90
32 24 40 nd 42 16
Measured by fluorescenceactivatedcell sorter. hTreatment for 5 days:GM-CSF,1000 U/ml; IFN-'y,100 U/ml; dibutyryl(db)cAMPand (db)cGMP, 2 g.M;PGE2, 10-7 M; indomethacin(IM), 1 /zg/ml;HA 1004, 2 /a,M.
sion. M o r e o v e r , they suggest that s t i m u l a t i o n o f class I e x p r e s s i o n i n d u c e d by G M - C S F is m e d i a t e d via c G M P - d e p e n d e n t p r o t e i n kinase.
ACKNOWLEDGMENTS This work was supported by the grants from the Academy of Finland, the Finnish Cancer Foundation, the Finnish Kidney Foundation, the University of Helsinki, Helsinki, and the Research and Science Foundation of Farmos, Turku, Finland.
REFERENCES
1. MetcalfD: Blood 67:257, 1986. 2. Dedhar S, Gaboury L, Galloway P, et al: Proc Natl Acad Sci USA 85:9253, 1988. 3. Malon DG, Pierce JH, Falko JP, et al: Blood 71:684, 1988. 4. Leszczynski D, H~iyry P: Submitted. 5. Leszczynski D, H/iyry P: FEBS Lett 242:383, 1989. 6. Snyder DS, Belier DI, Unanue ER: Nature 299:163, 1982. 7. Hancock WW, Pleau ME, Kobzik L: J Immunol 140:3021, 1988. 8. Franks DJ, Macmanus JP, Whitfield JF: Biochem Biophys Res Commun 44:1177, 1971. 9. Coffey GR, Davis JS, Djeu JY: J Immunol 140:2695, 1988. 10. Asano T, Hidaka H: J Pharmacol Exp Therap 231:141, 1984.
