commentary7
Immunology Today, voL 7, No. 5, 1986
Proteoglycans in secretory granules of NKcells The cloning of natural killer (NK) cells and cytotoxic T lymphocytes (CTL) has led to significant progress in the understanding of their surface phenotypes and specificities 1, and of their mechanism of killing. There is good evidence that cytolytic lymphocyte clones contain cytotoxic molecules, the so-called perforins (cytolysins) which are stored in cytoplasmic granules 2~. Perforins are proteins which can be activated by Ca 2+ to polymerize on the target cell and form transmembrane pores akin to those generated by the ninth component of complement (C9). Indeed, antigenic cross-reactivity can be demonstrated between perforin and C9 (J. Tschopp et al., unpublished). These findings suggest that the interaction of surface receptors of the cytolytic cell with target cell antigens stimulates exocytosis and release of granule contents, allowing perforins to form pores in the target cell membrane. Cell death would thus ensue from the exchange of solutes through these pores. To date, three granule proteins synthesized specifically by cytolytic lymphocyte clones, but not by helper- or B-cell clones, have been isolated and characterized: the cytolytic perforin and two serine esterases. The role of the latter in cytolysis is not known at present (D. Masson et aL, unpublished). Recently, MacDermott, Schmidt and colleagues s'6 found that granules of cloned NK cells also contain protease-resistant proteoglycans of the chondroitin sulphate A type. Proteoglycans are highly negatively charged molecules, and consist of long unbranched, sulphated polysaccharide chains composed of repeating disaccharide units bound to a core protein. Proteoglycans such as heparin, heparan sulphate, and chondroitin sulphate have been identified in many types of secretory cells including peritoneal mast cells 7 9 chromaffin cells of the adrenal medulla 1°, pituitary cells 11 and zymogen granules 12, and they are secreted by stimulated cells 9'13. Accordingly, MacDermott, Schmidt et al. showed that the incubation of cloned NK cells with appropriate target cells induces the release of up to 50% of 35S04-1abelled proteoglycans s into the medium. Using various target cells, they find a strong positive correlation between the extent of target cell lysis and proteoglycan release 6. Maximal release is achieved within an hour of incubation and at effector to target ratios of 0.5:1 suggesting that a single encounter of an NK with a target cell induces the release of a large fraction of granules. Their results do not allow one to determine whether the secretion is polarized towards the contact site between killer and target, or whether exocytosis occurs all over the surface of the NK cell. However, morphological examination and cinemicrography of the cytolytic process suggest that exocytosis occurs preferentially towards the target cells 14-16. What is the function of chondroitin sulphate A in NK-cell granules? MacDermott and colleagues propose several speculative answers which, however, are not mutually exclusive. First, chondroitin sulphate may comInstitute of Biochemistry, University of Lausanne, CH-1066 Epalinges, Switzerland.
JiirgTschoppand AndreasConzelmann plex with cytolytic molecules and facilitate their transfer to the target cell. This seems unlikely, since purified perforin is as efficient in pore formation as whole granules. Also, chondroitin sulphate A may be involved in the packaging and condensation of granule contents. Protection of NK cells from autolysis by secreted perforin may be another possible role. Indeed, our own results show that proteoglycans bind to and inactivate perforin (J. Tschopp etaL, unpublished). At pH 5, the probable pH of cytoplasmic granules, binding is considerably stronger than at neutral pH. This would be compatible with a hypothetical model drawn in Fig. 1. Chondroitin sulphate A, present in high concentrations within the granules, binds and inactivates perforin. Upon exocytosis, perforin dissociates from the proteoglycans due to the increase in pH and to dilution. Consequently, perforins exert their lytic action on the target cell, whereas molecules diffusing back to the effector cell are trapped by the much slower-diffusing proteoglycans, some of which may remain bound to granule membrane proteins. Regardless of whether proteoglycans have a specific role in cytolysis, the study of MacDermott, Schmidt et al. clearly establishes that cytotoxic activity is associated with regulated secretion. In addition, proteoglycan release provides a quantitative assay for this event by allowing the quantification of the activation of killers by target cells independently of the susceptibility of targets to the lytic activity of granule contents. The massive release of proteoglycans from granules of NK cells is induced not only by target cells, but also by antibodies directed to NK cell recognition structures, such as T3 and NKTa, and this suggests that it might be possible to find recognition structures on other, T3-negative NK cells by screening for antibodies which induce proteoglycan secretion.
Itln-sulfate
A
orin
Fig.1. Possiblerole of chondroitinsulphateA in granulesof NK cells:preventionof autolysis of the attacking cytotoxiccell. 135 © 1986, Elsevier Science Publishers [~.V, Amsterdam
For technical reasonswe are unableto reproducethis figure in colour. Seethe May issueof Immunology Today for full colour illustration.
0167
4919/86/$0200
Immunology Today, vol. 7, No. 5, 1986
References
10rtaldo, J.R. and Herberman, R.B.(1984) Ann. Rev. Immunol. 2, 359-394 2 Blumenthal, R., Millard, P.J., Henkart, M.P. etaL (1984)Proc. NatlAcad. Sci. USA 81, 5551-5555 3 Podack, E.R.,Young, J.D.E. and Cohn, Z.A. (1985) Proc. Natl Acad. Sci. USA 82, 8629-8631 4 Masson, D. and Tschopp, J. (1985)J. Biol. Chem. 260, 9069-9072 5 MacDermott, R.P.,Schmidt, R.E.,Caulfield, J.P.etaL (1985) J. Exp. Med. 162, 1771-1787 6 Schmidt, R.E, MacDermott, R.P.,Bartley,G. etaL (1985) Nature (London) 318, 289-291 7 Stevens,R.L.and Austen, K.F. (1982)./. BioL Chem. 257, 253-259 8 Razin,E., Stevens,R.L.,Akiyama, F. etaL (1982)J. Biol.
Chem. 257, 7229 7236 9 Stevens,R.L.,.Razin,E., Austen, K.F. et aL (1983) J. BioL Chem. 288, 5977-5984 10 Kiang, W., Krusius,T., Finne,J. etal. (1982) J. Biol. Chem. 257, 1651-1659 11 Zanini, A., Giannattasio,G., Nussdorfer, G. etaL (1980) J. Cell. BioL 86, 260-272 12 Reggio, H.A. and Palade,G.E. (1978)J. Cell. BioL 77, 288-314 13 Burgess,T.L. and Kelly, R.B.(1984) J. Cell. BioL 99, 22232230 14 Zagury, D. (1982)Adv, Exp. Med. Biol. 146, 149-168 15 Henkart, P.A. (1985)Ann. Rev. ImmunoL 3, 31-58 16 Yannelli, J.R., Sullivan,J.A., Mandell, G.L. etaL (1986) J. Immunol. 136, 377-382
A functionfor IgD? It is twenty-one years since IgD was first described I yet despite the advance of molecular and cellular immunology its precise function remains unknown. Various suggestions have been made. IgD is present at only low concentration in human serum and is undetectable in mouse serum, but is expressed on the surface of a large proportion of B cells. Its importance may be, therefore, as a membrane receptor 2. Other clues to its function may be found in its structure. Human IgD has a heavy chain consisting of four domains, an Fc region that is highly resistant to proteolysis, and a more labile Fab3. In mice, there appear to be only three heavy chain domains 4. The most remarkable structural feature is the long hinge region which is low in proline content and in inter-heavy chain disulphide bridges but rich in basic amino acid residues 3. This long hinge might be important in the cross-linking of slgD by antigenS; furthermore, because the region is very susceptible to proteolysis slgD may be cleaved after the binding of antigen, and the resulting Fab~-antigen complexes may be involved in the generation of an anti-idiotypic response6. In the light of recent understanding of how B cells process antigen, perhaps the structure of IgD enables it to play a special part in this processing; following the binding of antigen and consequent endocytosis, IgD then undergo proteolysis in the endocytic vesicles, releasing antigen to interact with class II molecules. Another possibility is that IgD is necessary for B cells to respond to T-dependent antigens. Although some of the data conflict, it appears that, while slgM+D + B cells respond to both T-independent and T-dependent antigens 7, cells which express slgD at low density may be tess able than slgD-rich B cells to make a primary response to thymus-dependent and possibly type I thymus-independent antigens 8,9. There is good evidence that the B cells resident in the splenic marginal zones, which appear to be responsible for the antibody response to carbohydrate antigens, express little or no IgD 1°. In a series of recent papers, Thorbecke and her colleagues have investigated how IgD might be involved in the modulation of antibody responses. Mice with IgD plasmacytomas, unlike animals bearing plasmacytomas
136
Department of Pathology, Universityof Newcastle upon Tyne, NE14LP, UK. Q 1986, ElsevierScience Publishers B.V., Amsterdam 0167-4919/86/502.00
Jane E. Calvert of other isotypes, make enhanced antibody responses ~1. Moreover, these augmented responses can be induced by the injection of IgD 11 in normal mice, but not in athymic animals, and can be adoptively transferred by Lytl +2-, L3T4+ T cells from IgD-treated mice 12. Incubation of spleen cells o n IgD-coated dishes in vitro for only one hour can endow the cells with the ability to augment an antibody response ~2. Both types of treatment dramatically increase the proportion of cells able to form rosettes with IgD-coated sheep erythrocytes, implying that a receptor for lgD has been induced ~3. The rosetting cells are T cells, designated TS, and can be induced in Lytl+2 , but not in L3T4- populations 13. The authors propose that these T8 cells are responsible for the enhanced antibody responses in IgD-treated animals. In a related paper Jacobson eta/, 14 propose that T~ cells might also be implicated in the enhancement of antibody responses induced by injection of mice with anti-lgD. Anti-lgD alone had a suppressive effect on the primary antibody response to trinitrophenylated keyhole limpet haemocyanin (TNP-KLH) but this could be overcome by simultaneous injection of a supernatant from cultures of lymph node cells and syngeneic lymphoma cells, which contained interleukin 2 and other lymphokines. Furthermore, if mice were primed with TNP-KLH and then injected with the same antigen four days later, the secondary response was strikingly enhanced by concomitant administration of anti-lgD together with the supernatant. The lymphokine-containing supernatant alone was less effective. When anti-lgD specific for one allotype was injected into allotype heterozygous animals, the enhancement of secondary response was seen predominantly in the IgG of the unlinked allotype. Together these results were interpreted as a suggestion that anti-lgD had at least two effects- direct suppression of B cells and an indirect enhancement dependent on the presence of lymphokines. Because the response of the linked allotype was suppressed by IgD a greater enhancement was observed in the IgG of the alternative allotype. A model is proposed to explain the immunoaugmentation induced by both IgD and anti-lgD. Coico et a/. 13 hypothesize that T8 cells are responsible, being induced in the first instance by the injected IgD myeloma proteins
Immunology Today, voL 7, No. 5, 1986
Proteoglycans in secretory granules of NKcells The cloning of natural killer (NK) cells and cytotoxic T lymphocytes (CTL) has led to significant progress in the understanding of their surface phenotypes and specificities 1, and of their mechanism of killing. There is good evidence that cytolytic lymphocyte clones contain cytotoxic molecules, the so-called perforins (cytolysins) which are stored in cytoplasmic granules 2~. Perforins are proteins which can be activated by Ca 2+ to polymerize on the target cell and form transmembrane pores akin to those generated by the ninth component of complement (C9). Indeed, antigenic cross-reactivity can be demonstrated between perforin and C9 (J. Tschopp et al., unpublished). These findings suggest that the interaction of surface receptors of the cytolytic cell with target cell antigens stimulates exocytosis and release of granule contents, allowing perforins to form pores in the target cell membrane. Cell death would thus ensue from the exchange of solutes through these pores. To date, three granule proteins synthesized specifically by cytolytic lymphocyte clones, but not by helper- or B-cell clones, have been isolated and characterized: the cytolytic perforin and two serine esterases. The role of the latter in cytolysis is not known at present (D. Masson et aL, unpublished). Recently, MacDermott, Schmidt and colleagues s'6 found that granules of cloned NK cells also contain protease-resistant proteoglycans of the chondroitin sulphate A type. Proteoglycans are highly negatively charged molecules, and consist of long unbranched, sulphated polysaccharide chains composed of repeating disaccharide units bound to a core protein. Proteoglycans such as heparin, heparan sulphate, and chondroitin sulphate have been identified in many types of secretory cells including peritoneal mast cells 7 9 chromaffin cells of the adrenal medulla 1°, pituitary cells 11 and zymogen granules 12, and they are secreted by stimulated cells 9'13. Accordingly, MacDermott, Schmidt et al. showed that the incubation of cloned NK cells with appropriate target cells induces the release of up to 50% of 35S04-1abelled proteoglycans s into the medium. Using various target cells, they find a strong positive correlation between the extent of target cell lysis and proteoglycan release 6. Maximal release is achieved within an hour of incubation and at effector to target ratios of 0.5:1 suggesting that a single encounter of an NK with a target cell induces the release of a large fraction of granules. Their results do not allow one to determine whether the secretion is polarized towards the contact site between killer and target, or whether exocytosis occurs all over the surface of the NK cell. However, morphological examination and cinemicrography of the cytolytic process suggest that exocytosis occurs preferentially towards the target cells 14-16. What is the function of chondroitin sulphate A in NK-cell granules? MacDermott and colleagues propose several speculative answers which, however, are not mutually exclusive. First, chondroitin sulphate may comInstitute of Biochemistry, University of Lausanne, CH-1066 Epalinges, Switzerland.
JiirgTschoppand AndreasConzelmann plex with cytolytic molecules and facilitate their transfer to the target cell. This seems unlikely, since purified perforin is as efficient in pore formation as whole granules. Also, chondroitin sulphate A may be involved in the packaging and condensation of granule contents. Protection of NK cells from autolysis by secreted perforin may be another possible role. Indeed, our own results show that proteoglycans bind to and inactivate perforin (J. Tschopp etaL, unpublished). At pH 5, the probable pH of cytoplasmic granules, binding is considerably stronger than at neutral pH. This would be compatible with a hypothetical model drawn in Fig. 1. Chondroitin sulphate A, present in high concentrations within the granules, binds and inactivates perforin. Upon exocytosis, perforin dissociates from the proteoglycans due to the increase in pH and to dilution. Consequently, perforins exert their lytic action on the target cell, whereas molecules diffusing back to the effector cell are trapped by the much slower-diffusing proteoglycans, some of which may remain bound to granule membrane proteins. Regardless of whether proteoglycans have a specific role in cytolysis, the study of MacDermott, Schmidt et al. clearly establishes that cytotoxic activity is associated with regulated secretion. In addition, proteoglycan release provides a quantitative assay for this event by allowing the quantification of the activation of killers by target cells independently of the susceptibility of targets to the lytic activity of granule contents. The massive release of proteoglycans from granules of NK cells is induced not only by target cells, but also by antibodies directed to NK cell recognition structures, such as T3 and NKTa, and this suggests that it might be possible to find recognition structures on other, T3-negative NK cells by screening for antibodies which induce proteoglycan secretion.
Itln-sulfate
A
orin
Fig.1. Possiblerole of chondroitinsulphateA in granulesof NK cells:preventionof autolysis of the attacking cytotoxiccell. 135 © 1986, Elsevier Science Publishers [~.V, Amsterdam
For technical reasonswe are unableto reproducethis figure in colour. Seethe May issueof Immunology Today for full colour illustration.
0167
4919/86/$0200
Immunology Today, vol. 7, No. 5, 1986
References
10rtaldo, J.R. and Herberman, R.B.(1984) Ann. Rev. Immunol. 2, 359-394 2 Blumenthal, R., Millard, P.J., Henkart, M.P. etaL (1984)Proc. NatlAcad. Sci. USA 81, 5551-5555 3 Podack, E.R.,Young, J.D.E. and Cohn, Z.A. (1985) Proc. Natl Acad. Sci. USA 82, 8629-8631 4 Masson, D. and Tschopp, J. (1985)J. Biol. Chem. 260, 9069-9072 5 MacDermott, R.P.,Schmidt, R.E.,Caulfield, J.P.etaL (1985) J. Exp. Med. 162, 1771-1787 6 Schmidt, R.E, MacDermott, R.P.,Bartley,G. etaL (1985) Nature (London) 318, 289-291 7 Stevens,R.L.and Austen, K.F. (1982)./. BioL Chem. 257, 253-259 8 Razin,E., Stevens,R.L.,Akiyama, F. etaL (1982)J. Biol.
Chem. 257, 7229 7236 9 Stevens,R.L.,.Razin,E., Austen, K.F. et aL (1983) J. BioL Chem. 288, 5977-5984 10 Kiang, W., Krusius,T., Finne,J. etal. (1982) J. Biol. Chem. 257, 1651-1659 11 Zanini, A., Giannattasio,G., Nussdorfer, G. etaL (1980) J. Cell. BioL 86, 260-272 12 Reggio, H.A. and Palade,G.E. (1978)J. Cell. BioL 77, 288-314 13 Burgess,T.L. and Kelly, R.B.(1984) J. Cell. BioL 99, 22232230 14 Zagury, D. (1982)Adv, Exp. Med. Biol. 146, 149-168 15 Henkart, P.A. (1985)Ann. Rev. ImmunoL 3, 31-58 16 Yannelli, J.R., Sullivan,J.A., Mandell, G.L. etaL (1986) J. Immunol. 136, 377-382
A functionfor IgD? It is twenty-one years since IgD was first described I yet despite the advance of molecular and cellular immunology its precise function remains unknown. Various suggestions have been made. IgD is present at only low concentration in human serum and is undetectable in mouse serum, but is expressed on the surface of a large proportion of B cells. Its importance may be, therefore, as a membrane receptor 2. Other clues to its function may be found in its structure. Human IgD has a heavy chain consisting of four domains, an Fc region that is highly resistant to proteolysis, and a more labile Fab3. In mice, there appear to be only three heavy chain domains 4. The most remarkable structural feature is the long hinge region which is low in proline content and in inter-heavy chain disulphide bridges but rich in basic amino acid residues 3. This long hinge might be important in the cross-linking of slgD by antigenS; furthermore, because the region is very susceptible to proteolysis slgD may be cleaved after the binding of antigen, and the resulting Fab~-antigen complexes may be involved in the generation of an anti-idiotypic response6. In the light of recent understanding of how B cells process antigen, perhaps the structure of IgD enables it to play a special part in this processing; following the binding of antigen and consequent endocytosis, IgD then undergo proteolysis in the endocytic vesicles, releasing antigen to interact with class II molecules. Another possibility is that IgD is necessary for B cells to respond to T-dependent antigens. Although some of the data conflict, it appears that, while slgM+D + B cells respond to both T-independent and T-dependent antigens 7, cells which express slgD at low density may be tess able than slgD-rich B cells to make a primary response to thymus-dependent and possibly type I thymus-independent antigens 8,9. There is good evidence that the B cells resident in the splenic marginal zones, which appear to be responsible for the antibody response to carbohydrate antigens, express little or no IgD 1°. In a series of recent papers, Thorbecke and her colleagues have investigated how IgD might be involved in the modulation of antibody responses. Mice with IgD plasmacytomas, unlike animals bearing plasmacytomas
136
Department of Pathology, Universityof Newcastle upon Tyne, NE14LP, UK. Q 1986, ElsevierScience Publishers B.V., Amsterdam 0167-4919/86/502.00
Jane E. Calvert of other isotypes, make enhanced antibody responses ~1. Moreover, these augmented responses can be induced by the injection of IgD 11 in normal mice, but not in athymic animals, and can be adoptively transferred by Lytl +2-, L3T4+ T cells from IgD-treated mice 12. Incubation of spleen cells o n IgD-coated dishes in vitro for only one hour can endow the cells with the ability to augment an antibody response ~2. Both types of treatment dramatically increase the proportion of cells able to form rosettes with IgD-coated sheep erythrocytes, implying that a receptor for lgD has been induced ~3. The rosetting cells are T cells, designated TS, and can be induced in Lytl+2 , but not in L3T4- populations 13. The authors propose that these T8 cells are responsible for the enhanced antibody responses in IgD-treated animals. In a related paper Jacobson eta/, 14 propose that T~ cells might also be implicated in the enhancement of antibody responses induced by injection of mice with anti-lgD. Anti-lgD alone had a suppressive effect on the primary antibody response to trinitrophenylated keyhole limpet haemocyanin (TNP-KLH) but this could be overcome by simultaneous injection of a supernatant from cultures of lymph node cells and syngeneic lymphoma cells, which contained interleukin 2 and other lymphokines. Furthermore, if mice were primed with TNP-KLH and then injected with the same antigen four days later, the secondary response was strikingly enhanced by concomitant administration of anti-lgD together with the supernatant. The lymphokine-containing supernatant alone was less effective. When anti-lgD specific for one allotype was injected into allotype heterozygous animals, the enhancement of secondary response was seen predominantly in the IgG of the unlinked allotype. Together these results were interpreted as a suggestion that anti-lgD had at least two effects- direct suppression of B cells and an indirect enhancement dependent on the presence of lymphokines. Because the response of the linked allotype was suppressed by IgD a greater enhancement was observed in the IgG of the alternative allotype. A model is proposed to explain the immunoaugmentation induced by both IgD and anti-lgD. Coico et a/. 13 hypothesize that T8 cells are responsible, being induced in the first instance by the injected IgD myeloma proteins
