Immunology Today August 1985
The immune destruction of pancreatic fl cells from Charles Janeway The /3 cell of the pancreatic islets of Langerhans is the sole source of the circulating polypeptide hormone insulin, required for a number of metabolic effects, particularly for transport of glucose into muscle and adipose cells and for control of hepatic glucose production. In type I diabetes mellitus, or insulin-dependent diabetes (IDDM), the fl cell disappears from the islets, no insulin is produced, and hyperglycemia and ketoacidosis ensue. While the metabolic syndrome can be reversed toward normal with injections of insulin, long-term complications as well as acute imbalances in glucose metabolism plague victims of insulindependent diabetes. A series of workshops, co-sponsored by the United States Government and the American Diabetes Association, have met to consider this problem and to make recommendations for future research. At the last of these, held in 1982, it was proposed that the workgroup considering the mechanism of fl cell destruction, and that dealing with transplantation of islet tissue as a treatment for I D D M , merge in a future workshop, since it seemed apparent that /$ cell destruction was mediated by the same or similar mechanisms in both cases. T o this end, a workshop was held recently under the sponsorship of the Howard Hughes Medical Institute in Miami to consider various aspects of the immunology of the pancreatic fl cell. Evidence favoring an autoimmune pathogenesis for IDDM Several lines of evidence have converged to make it seem almost certain that most, if not all, I D D M results from an autoimmune process. First, there is a strong association of I D D M with HLAD genotype. In particular, heterozygous individuals having H L A - D R 3/4 have a relative risk of about 40 of having I D D M , while individuals
having H L A - D R 2 and lacking HLADR3 and 4 have relative risks around 0.2, suggesting that this genotype confers protection against the development of the disease. It seems likely that these risk calculations will become even more striking in the near future. Such advances will come from examining subtypes of H L A - D defined by restriction-fragment length polymorphisms (A. Lernmark, Copenhagen), by two-dimensional gel electrophoresis of H L A - D proteins, or by T-cell-detected antigens (F. Bach, Minnesota). As the only known role of H L A - D antigens is in T-cell recognition, Bach argued that T-cell typing may detect polyrnorphisms that are not
detectable by other means and cited the well-known example of the class I H-2 mutants, which give rise to pronounced differences in T-cell recognition but to only minor changes in protein mobility on gels or serological specificities. In relation to I D D M , Bach reported that at least five T-cell-defined subsets of H L A - D R 4 exist. More significantly, while H L A - D R 2 is rarely found in patients with IDDM, the T-cell-defined specificity HLA-Dw2, usually associated with HLA-DR2, is found in only three of his series of 13 I D D M patients having HLA-DR2. Thus, such typing appears to be removing anomalous associations of I D D M and H L A - D genotype. Such an analysis, if vigorously pursued, should allow a far more precise definition of H L A genotype, which seems certain to improve the strength of the association with IDDM. Continued on p. 230
Flow microfluorimetry
Achievements and prospects from NeiU M. Mackenzie, David W. Dresser and Andrew C. Pinder New applications which are areas of potential growth for the science of flow microfluorimetry (FMF) and represent a major shift in direction away from its principal current use by cellular immunologists were highlighted at a recent meeting. Its intention was to inform flow mlcrofluorimetrists of current trends in machine and dye technology and new applications in particular chromosome analysis and the analysis of cells isolated from solid tissue, One of the earliest applications of FMF was for the analysis of cell cycle kinetics in isolated cells and nuclei. Z. Darzynkiewicz (New York) discussed the properties of D N A binding dyes on which this work relies, some of which intercalate with the DNA, e.g. propidium iodide and some that do not, e.g. H33258. Until recently such FMF measurements of cellular DNA content were limited to the availability of fresh tissue: now, as demonstrated by B. Schutte (Maastrict) and J. Watson (Cambridge), it is possible to examine cell cycle and ploidy levels by FMF in nuclei previously been extracted from tissues previously fLxed and blocked
for histological studies. Schutte has correlated the degree of propidium iodide staining of DNA in nuclei from archive material derived from patients with colorectal cancers with survival of these patients. He found that the ploidy level was not an adverse prognostic sign in patients with localised tumours (Dukes Stages A and B) but that of those patients with lymph node involvement (Dukes Stage C) and with distant metastases (Dukes Stage D), ones carrying diploid tumours had a better survival rate than those with aneuploid turnouts. Similarly no correlation between the percentage of Continued on p. 232
© 1985, Elsevier Science Publishers B.V., Amsterdam 0167 - 4919/85/$02.00
230 Pancreatic/$ cells--continuedfrom p. 229 Perhaps the most striking evidence favoring an autoimmune pathogenesis of I D D M comes from studies of twins or triplets with I D D M analysed by G. Eisenbarth (Boston), and from studies of identical twin pancreatic segment grafts by D. Sutherland (Minnesota). In Eisenbarth's studies, twins or triplets destined to become diabetic frequently have antibodies to fl cell antigens for years prior to onset, and show a progressive but clinically asymptomatic loss of fl cell function, especially to glucose-stimulated insulin release, several years before becoming frankly diabetic. This slow and unrelenting course is in contrast to the resuks of segmental pancreas grafts between monozygotic twins, one of whom has I D D M of long standing (15-30 years). In each case described by Sutherland, the graft initially restored essentially normal I$ cell function. However, in the absence of immunosuppressive agents, cell loss could be detected within two weeks, and by 10-12 weeks after grafting, all ]] cells were destroyed, while ~r and 6 cells were spared, as in the original disease. As the twins are genetically identical, and as several carried renal grafts from the same donor for many years without immunosuppression, this reaction must be fl-cell specific. Where analysed, islets containing /] cells showed a characteristic infiltrate with lymphocytes, termed insulitis, comprising mainly cells with the T8 surface marker. The finding of insulitis is a third reason for considering I D D M as an autoimmune disease, In recent onset I D D M , islets are frequently iniVfltrated by mononuclear cells, including T lymphocytes, B lymphocytes and macrophages. In addition, Eisenbarth finds activated T lymphocytes in peripheral blood of individuals prior to, or at onset of, I D D M . The specificity o f t cell destruction, and the findings that only islets having residual/3 cells show insulitis, strongly supports the notion of a specific immunological attack on the fl cell. However, the nature of this immunological process is not under:stood. The nature of the autoimmune destruction of fl cells Given that fl cell destruction results from an immunological attack, what might be the mechanism of this attack, and what cells might mediate it? Several mechanisms of/3 cell destruction were discussed at this meeting, but there is at present no consensus as to mechanism. Specific/~ cell surface antibodies are seen in diabetes, often preceding the
Immunology Today, vol. 6, No. 8, 1985
onset of overt diabetes by many years. In particular, an antibody reacting with a 64 Kd protein on the surface of the fl cell is seen early in patients who later have diabetes, and in spontaneously diabetic BB rats before the time at which /] cell loss can first be detected (Lernmark). Whether this antibody causes cell destruction and by what means, is not known. J. Nerup (Copenhagen) has shown that/~ cells can be destroyed in vitro by supernatants of mitogen-activated human mononudear cells, and the active principle appears to be interleukin 1 (IL- 1). This would be a unique function for IL-I, but it was consistent with electron-micrographs ofinsulitis in one experimental animal system, the BB rat, obtained by H. Kolb (Dusseldorf), who observed macrophages as the initial infikrating cells in the islets. However, IL-1 destroys all islet cells in vitro, while fl cells appear to be the only target of autoimmune insulitis in vivo. Two types of cytotoxic T cells (CTL) were discussed. C. Janeway (Yale) described T4 ÷ C T L that could kill targets sensitive to lymphotoxin, presumably by lymphotoxin release. Such cells are restricted in antigen recognition by class II M H C gene products, and can kill bystander targets. Thus/3 cells could be destroyed by such cells if they were particularly sensitive to this mechanism of killing. As such C T L are class II restricted, the association with H L A - D could be accounted for by such cells. In addition, E. Simpson (Harrow) described a correlation between rejection of skin grafts differing for the male antigen H-Y with delayed-type hypersensitivity. Graft rejection is under genetic control by class II M H C genes, while classical eytolytic T-cell responses are controlled both by class I and class II M H C antigens and do not correlate with skin graft rejection. The possible role of class I-restricted killer cells in diabetes was rendered more problematic by the finding reported by F. Sheinvold (Miami) that/3 cells of several species express little or no class I M H C gene products, a finding that was not supported by data from other investigators; nevertheless, it was clear that the level of class I products on the fl cell might be low. The expression of class n products in the islets was the subject of significant discussion. Sheinvold found it on capillary, endothelium, but not on other cell types. Interestingly, A. Cooke (London) found class II antigens on capillary endothelial cells in pre-diabetic BB rats but not in normal pancreas, and only observed class n antigens on fl cells late in the development of diabetes.
Whether/J cell class 11 molecules are endogenously synthesized or passively acquired from infiltrating macrophages and B cells is not known. Finally, A. Rabinovitch (Miami) demonstrated that cells indistinguishable from natural killer cells appeared to be selectively cytotoxic forl$ cells in vitro. This finding was questioned on the ground that nude mice, and mice and rats with suppressed T-cell function, did not develop diabetes, although they have normal to elevated levels of natural killer cells. The relevance of this finding remains to be established. In summary, the nature of the cell that destroys the fl cell selectively, and the mechanism by which it does so, has not been elucidated. Neither has the target structure that allows such effector cells to selectively kill ~ cells, if it exists, been defined. The rapid rate at which syngeneic I] cells are destroyed in the identical twin studies of Sutherland, and the preponderance of T8 cells in these islets, suggests that classic, class I MHC-restricted C T L are the most likely candidate for the destruction o f t cells. This is reinforced by the finding that only islets having fl cells display insulitis, suggesting that the response is ~Jcell specific, and that the fl cells are not uniquely sensitive to a toxic lymphocyte product released in the islets in response to another autoantigen or a virus. However, until such cells are cloned and shown to transfer the disease, questions about the mechanism of fl cell destruction will persist. Experimental models of I D D M The search for an experimental model for I D D M has focused recently o n two animals that have a high propensity to develop diabetes spontaneously with accompanying/3 cell destruction and insulitis: the BB/W rat (E. Marliss, Montreal) and the N O D mouse (S. Makino, Japan). Several studies using these models were presented; these studies illuminated certain issues. In both models, the genetics are complex but have one guiding principle. In breeding studies, susceptibility to I D D M mapped in the M H C in both cases. However, in each model, at least one other gene was important. In the BB rat, the most prominent of these is a gene that causes lymphopenia (Marliss, Eisenbarth) and particularly a deficiency in cells bearing the W3/25 marker. Interestingly, BB rats can be protected from I D D M by transferring T cells carrying the cell surface antigen W3/25 (A. Rossini, Boston), usually associated with T H cells. Janeway and P. Flood (New Haven) both emphasized
231
Immunology Today, voL 6, No. 8, 1985 the importance of referring to such cells in terms that reflected the means by which they were identified rather than the putative function, and Flood pointed out that cells inducing suppression were of the T4 (or helper) phenotype in mice. While T-cell transfers can prevent I D D M in BB rats, thymectomy, anti-lymphocyte serum and cyclosporin A can also prevent the disease, but cannot reverse it after onset. Furthermore, Rossini has also reported passive transfer of I D D M by T cells from acutely diabetic BB rats to various normal, non-diabetic rats. This phenomenon appears to show M H C restriction. As in man, acutely diabetic BB rats have high levels of activated T cells (A. Naji, Philadelphia). These studies, taken together, suggest that T cells are intimately involved in both pathogenesis and protection, a finding that places immunoregulation squarely in the frame of studies of this disease (see below). The N O D mice proved to be an immunogenetic puzzle, lacking detectable class II M H C molecules detectable by a variety of probes (M. Hattori, Joslin). I-E molecules are clearly absent in these mice, as no m R N A was visible on Northern hybridization blots; I-A molecules were not detected with several broadly reactive anti-I-A monoclonal antibodies, but m R N A was present (R. Jackson, Joslin). T h e N O D mouse is not lymphopenic, and has no known immunologic abnormalities. However, nude N O D mice do not develop diabetes unless given n u / + T cells (Makino), a finding consistent with earlier studies in BB/W rats. Interestingly, males are relatively resistant to diabetes unless raised in germ-free conditions, and N O D mice raised at the Walter and Eliza Hall Institute in Melbourne (T. Mandel) had a lower incidence of I D D M than mice raised in Japan. The pathology of experimental I D D M was of interest. In particular, insulitis in the BB rat is far less intense than the inflammatory reaction seen in the N O D mouse. The most striking finding in the N O D mouse was an initial peri-insulitis of great intensity associated with intact islet morphology and function. This suggested that one level of control in autoimmune diabetes might be access of lymphoid cells to the tissue. It is of interest that this should be apparent in mice but not rats, as it has long been known to be more difficult to induce organ-specific autoimmunity in mice than rats, and vascular barriers appear to be more effective in mice than in rats (blood : brain, blood : thymus). Mandel reported a similar peri-insulitis
in islets grafted under the kidney capsule, a n d it is known to be more difficult to graft islets between rats than between mice.
Transplanted/3 ceils as allografts and autografts As I D D M is clearly a disease in which there is loss of a particular cell type, the/3 cell, replacement of/3 cells is the most logical therapy. However, several obstacles confront the clinician who chooses this course of treatment. First, islets need to be prepared in reasonable numbers. It was clear from studies cartied out by D. Mintz, R. Alejandro and colleagues (Miami), that the n u m b e r of: islets grafted to pancreatectomized dogs determined the success of an autograft. A variety of protocols for improving the yield were discussed (P. Morris, London). Second, the i m m u n e response to allografted islets needs to be overcome. Various maneuvers to do so were described. These involved depletion of the islets of stimulatory, class II M H C bearing cells by anti-Ia antibody, by high oxygen or low temperature culture, or by treatment with anti-dendritic cell antibody (J. Davie, St. Louis; H. Lau, Columbia; Mandel: Naji). The host immune system could also be damped using cyclosporin A (Alejandro), antilymphocyte antibody (Davie) or transfusion of anti-Ia treated donor whole blood cells prior to transplant (Davie, Lau). The finding that even diabetic individuals bearing renal grafts from identical twins could specifically reject [J cells from the same genetically identical donor unless immunosuppression was used suggested that grafting islets deliberately mismatched for H L A antigens might be critical in patients with autoi m m u n e I D D M , since recognition of/3 cell-specific antigens, and thus autoimmunity, might be M H C restricted. This was emhasized by W. Silvers (Philadelphia), who finds that acquired tolerance is M H C restricted, in that rats and mice tolerant to allogeneic tissue fail to recognize minor histocompatability or tissue-specific antigens in the context of non-self M H C . Thus, two principles emerge in these studies. First, to study islet transplants as a treatment for diabetes, a spontaneously diabetic animal is the best model. Second, allografts may have major advantages over H L A matched grafts, so prolongation of islet allograft survival is a relevant and important goal of research. A second aspect of transplantation biology that appeared in several presentations involved the nature of tolerance to foreign tissue antigens. In several systems, it was clear that clonal
deletion could not account for the data. For instance, in the studies presented by W. Streilein (Miami), neonatal tolerance could not be broken even by heroic levels of normal T cells, but rather required specifically immune T cells. Furthermore, the doses of cells required to reverse tolerance increased with time, even if the graft being monitored was always given as fresh tissue. These findings are more consistent with active maintenance of the tolerant state than they are with a clonal deletion model of tolerance. Recent data from P. Medawar and from K. Eichmann suggest that G T L precursors are under active regulation in tolerance, consistent with the findings presented here. T h e i m p o r t a n c e of suppressor cells in a u t o i m m u n i t y a n d tolerance This workshop left a very strong impression that whatever the mechanism of/3 cell destruction, in either I D D M or islet-cell transplantation, the critical control parameter was the level of specific immunological suppression. W h a t was most dramatic was the obvious evidence, often mentioned in the past by the late H H M I investigator Richard Gershon, that a phenomenon as fundamental in immunologic dogma as tolerance to M H C antigens was under active immunological control. Furthermore, T ceils are clearly important in the causation of I D D M , as seen by the protection afforded by thymectomy, by the transfer of the disease by activated T cells, and by the infiltration o f T cells in rejection of identical twin pancreas grafts. Yet T cells can also protect against I D D M , which strongly suggests that different populations of T cells performing different functions are responsible for the disease and can also prevent it. Thus, measures that simply decrease total cell levels or activity are unlikely to prevent or 'cure' IDDM. This is dramatically illustrated in recent-onset I D D M patients given cyclosporin A. These patients can frequently be kept off insulin for as long as cyclosporin A is given. However, as soon as treatment is terminated, the disease continues where it left off and the fl cells are eventually totally lost. Likewise, the chronic course ofprimaryfl cell destruction, as opposed to the rapid destruction of transplanted, histocompatible fl cells, suggests that I D D M represents a slight shift in the immunoregulatory balance towards autoimmunity. T h e precarious nature of this balance is further emphasized by the high rate of discordancy for I D D M in identical twins living in the same environment, and the apparent influence of gender, feeding pattern and season (Nerup) on incidence.
Immunology Today, vol. 6, No. 8, 1985
232 If this construction is correct, what is the optimal approach to the prevention of IDDM? This question is highly pertinent, as advances in genetics may soon make it possible to identify high-risk individuals. To paraphrase G. Bernard Shaw, who in turn was mimicking Metchnikov, it seems likely that the key is to learn how to 'stimulate the suppressors', stimulate the suppressors. This is not a simple task. One would like to
stimulate only those suppressors that are relevant, and this requires an understanding of both the pathways of immune suppression and the specific antigen involved. Nevertheless, this analysis clearly shifts the emphasis from the fl cell and the effector cells that destroy it (still critically important issues) to the analysis of suppression mechanisms in general, and to their control and manipulation in I D D M in particular.
Charles A. Janeway, Jr is in the Howard Hughes Medical Institute at Yale University School of Medicine, New Haven, C T 06510, USA.
Flow microfluorimetry--cantinued from p. 229 cells in S phase and survival of patients in Dukes Stages A and B was found, but, in Dukes stages C and D, a low proliferative rate was correlated with a longer survival. Watson used a dual staining protocol with propidium iodide and a FITC labelled monoclonal antibody against a c-myc oncogene product in a retrospective study on testis and on colon cancer. Whilst at the moment insufficient numbers of samples have been analysed to be significant there appears to be a correlation of high anti-c-myc activity with low survival rate. These two studies show that such information allows prognostic judgements to be made from biopsy material, a point emphasised by G. Valet (Martinsried), who gave details of how FMF measurements are currently being used to assess the necessity for surgical intervention in colorectal cancer.
Burkitt's lymphoma patients appeared as a change in the relative fluorescence of the translocated chromosome and its partner. J. A. Aten (Amsterdam) suggested that it may be possible, using the 'slit scan' method, to correlate chromosomal banding patterns using FMF with those obtained by classical cytogenetics to improve the resolution of individual chromosomes in flow karyotype. The large-scale alterations in karyotype mentioned above are fairly easy to detect using FMF. Smaller changes in the chromosomes such as alteration in their higher order structure cannot be visualized using D N A binding fluorochromes alone. One solution t9 this problem is to stain, in conjunction with the traditional D N A dyes, with antibodies directed against specific chromatin structures. Antibodies have the advantage over the DNA-binding dyes in that they may be chromosome specific and the particular site to which they bind can be defined. Green's group used flow karyotype analysis to examine chromosome abnormalities caused by radiation. They found that large doses of radiation (400 fads) altered the flow karyotype (Hoechst 33342 stain) but these levels were lethal! Smaller doses, of between 25 and 50 fads had no obvious effect on the flow karyotype so that detection of any chromosomal changes by this method has proved impossible. Their solution has been to concentrate their search on chromosomes bearing no centromeres. (acentric) or two centromeres (dicentric) using sera from scleroderma patients which contains antibodies which bind to centromeres. A. Fellner, R. Festin and P. Matsson (Uppsala) developed monoclonal antibodies to double stranded and single stranded D N A to study higher order structure in chromosomes. Correlations between the staining activity of propidium iodide and these
monoclonal antibodies reveal that antibody penetration is independent of the degree of the contraction of the chromosomes. B. Turner (Birmingham) also looked at chromosomes in flow using a monoclonal which binds to a highly conserved region (amino acids 1-8) of histone H-2B. Unlike Fellner, Festin and Matsson he found two sub-populations of chromosomes to which the antibody bound - believed to represent differences in higher order chromatin structure. Whilst the ability to analyse chromosomes in the flow microfluorimeter has been with us for some time, sorting out of individual chromosomes now seems to be a practical proposition. M. A. Van Dilla representing the National Laboratory Gene Library Project at Lawrence Livermore (LLNL) and Los Alamos (LANL) USA described the use of FMF technology to produce a chromosome specific gcne library. The first phase of the project, currently under way, is to produce a human DNA library for use in gene mapping, genetic disease diagnosis, linkage and pedigree analysis, using restriction fragment length polymorphism (RFLP) analysis. Later the group hopes to clone larger fragments of up to 100 kb, suitable for the study of gene structure, expression and function. Once these two phases are complete, it is hoped that the mouse genome will be analysed systematically. Whilst in theory the methods for production of these libraries are straightforward, the technical problems on a day to day basis have proved to be immense. Chromosomes are isolated from human fibroblasts (number 13 and smaller) or Chinese hamster/human hybrids (numbers 1-12). The chromosomes of interest (Hoechst 33342 or Chrornomycin A 2 stained) are sorted out using FMF. The 'sorted' chromosomes are digested using Hind III (LLNL) or
Chromosomes Chromosomes can be isolated, stained with DNA-binding fluorochromes and analysed by flow microfluorimetry. The resultant fluorescence profile has been coined the 'flow karyotype'. B. D. Young (London) and D. K. Green (Edinburgh) discussed ways in which flow karyotyping might be used for detecting alterations in chromosome structure or number to assist, for example, cancer detection- which by classical methods is labour-intensive, subjective and slow. The difficulty with application of FMF however is the individual nature of the human flow karyotype. This is due mainly to natural variations in centrometric heterochromatin especially in chromosomes 1, 3, 9, 16, 21, 22 and Y. Even so, using flow karyotype analysis trisomy in Down's syndrome revealed itself as an increased height for the peak representing chromosome 21 and translocations in~ for example, chromosomes from
Acknowledgement The author wishes to thank the participants at the meeting, its sponsor, the Howard Hughes Medical Institute and Drs Rossini, Cahill, Lernmark and Eisenbarth for their help in organizing the meeting and this report. IT]:
The immune destruction of pancreatic fl cells from Charles Janeway The /3 cell of the pancreatic islets of Langerhans is the sole source of the circulating polypeptide hormone insulin, required for a number of metabolic effects, particularly for transport of glucose into muscle and adipose cells and for control of hepatic glucose production. In type I diabetes mellitus, or insulin-dependent diabetes (IDDM), the fl cell disappears from the islets, no insulin is produced, and hyperglycemia and ketoacidosis ensue. While the metabolic syndrome can be reversed toward normal with injections of insulin, long-term complications as well as acute imbalances in glucose metabolism plague victims of insulindependent diabetes. A series of workshops, co-sponsored by the United States Government and the American Diabetes Association, have met to consider this problem and to make recommendations for future research. At the last of these, held in 1982, it was proposed that the workgroup considering the mechanism of fl cell destruction, and that dealing with transplantation of islet tissue as a treatment for I D D M , merge in a future workshop, since it seemed apparent that /$ cell destruction was mediated by the same or similar mechanisms in both cases. T o this end, a workshop was held recently under the sponsorship of the Howard Hughes Medical Institute in Miami to consider various aspects of the immunology of the pancreatic fl cell. Evidence favoring an autoimmune pathogenesis for IDDM Several lines of evidence have converged to make it seem almost certain that most, if not all, I D D M results from an autoimmune process. First, there is a strong association of I D D M with HLAD genotype. In particular, heterozygous individuals having H L A - D R 3/4 have a relative risk of about 40 of having I D D M , while individuals
having H L A - D R 2 and lacking HLADR3 and 4 have relative risks around 0.2, suggesting that this genotype confers protection against the development of the disease. It seems likely that these risk calculations will become even more striking in the near future. Such advances will come from examining subtypes of H L A - D defined by restriction-fragment length polymorphisms (A. Lernmark, Copenhagen), by two-dimensional gel electrophoresis of H L A - D proteins, or by T-cell-detected antigens (F. Bach, Minnesota). As the only known role of H L A - D antigens is in T-cell recognition, Bach argued that T-cell typing may detect polyrnorphisms that are not
detectable by other means and cited the well-known example of the class I H-2 mutants, which give rise to pronounced differences in T-cell recognition but to only minor changes in protein mobility on gels or serological specificities. In relation to I D D M , Bach reported that at least five T-cell-defined subsets of H L A - D R 4 exist. More significantly, while H L A - D R 2 is rarely found in patients with IDDM, the T-cell-defined specificity HLA-Dw2, usually associated with HLA-DR2, is found in only three of his series of 13 I D D M patients having HLA-DR2. Thus, such typing appears to be removing anomalous associations of I D D M and H L A - D genotype. Such an analysis, if vigorously pursued, should allow a far more precise definition of H L A genotype, which seems certain to improve the strength of the association with IDDM. Continued on p. 230
Flow microfluorimetry
Achievements and prospects from NeiU M. Mackenzie, David W. Dresser and Andrew C. Pinder New applications which are areas of potential growth for the science of flow microfluorimetry (FMF) and represent a major shift in direction away from its principal current use by cellular immunologists were highlighted at a recent meeting. Its intention was to inform flow mlcrofluorimetrists of current trends in machine and dye technology and new applications in particular chromosome analysis and the analysis of cells isolated from solid tissue, One of the earliest applications of FMF was for the analysis of cell cycle kinetics in isolated cells and nuclei. Z. Darzynkiewicz (New York) discussed the properties of D N A binding dyes on which this work relies, some of which intercalate with the DNA, e.g. propidium iodide and some that do not, e.g. H33258. Until recently such FMF measurements of cellular DNA content were limited to the availability of fresh tissue: now, as demonstrated by B. Schutte (Maastrict) and J. Watson (Cambridge), it is possible to examine cell cycle and ploidy levels by FMF in nuclei previously been extracted from tissues previously fLxed and blocked
for histological studies. Schutte has correlated the degree of propidium iodide staining of DNA in nuclei from archive material derived from patients with colorectal cancers with survival of these patients. He found that the ploidy level was not an adverse prognostic sign in patients with localised tumours (Dukes Stages A and B) but that of those patients with lymph node involvement (Dukes Stage C) and with distant metastases (Dukes Stage D), ones carrying diploid tumours had a better survival rate than those with aneuploid turnouts. Similarly no correlation between the percentage of Continued on p. 232
© 1985, Elsevier Science Publishers B.V., Amsterdam 0167 - 4919/85/$02.00
230 Pancreatic/$ cells--continuedfrom p. 229 Perhaps the most striking evidence favoring an autoimmune pathogenesis of I D D M comes from studies of twins or triplets with I D D M analysed by G. Eisenbarth (Boston), and from studies of identical twin pancreatic segment grafts by D. Sutherland (Minnesota). In Eisenbarth's studies, twins or triplets destined to become diabetic frequently have antibodies to fl cell antigens for years prior to onset, and show a progressive but clinically asymptomatic loss of fl cell function, especially to glucose-stimulated insulin release, several years before becoming frankly diabetic. This slow and unrelenting course is in contrast to the resuks of segmental pancreas grafts between monozygotic twins, one of whom has I D D M of long standing (15-30 years). In each case described by Sutherland, the graft initially restored essentially normal I$ cell function. However, in the absence of immunosuppressive agents, cell loss could be detected within two weeks, and by 10-12 weeks after grafting, all ]] cells were destroyed, while ~r and 6 cells were spared, as in the original disease. As the twins are genetically identical, and as several carried renal grafts from the same donor for many years without immunosuppression, this reaction must be fl-cell specific. Where analysed, islets containing /] cells showed a characteristic infiltrate with lymphocytes, termed insulitis, comprising mainly cells with the T8 surface marker. The finding of insulitis is a third reason for considering I D D M as an autoimmune disease, In recent onset I D D M , islets are frequently iniVfltrated by mononuclear cells, including T lymphocytes, B lymphocytes and macrophages. In addition, Eisenbarth finds activated T lymphocytes in peripheral blood of individuals prior to, or at onset of, I D D M . The specificity o f t cell destruction, and the findings that only islets having residual/3 cells show insulitis, strongly supports the notion of a specific immunological attack on the fl cell. However, the nature of this immunological process is not under:stood. The nature of the autoimmune destruction of fl cells Given that fl cell destruction results from an immunological attack, what might be the mechanism of this attack, and what cells might mediate it? Several mechanisms of/3 cell destruction were discussed at this meeting, but there is at present no consensus as to mechanism. Specific/~ cell surface antibodies are seen in diabetes, often preceding the
Immunology Today, vol. 6, No. 8, 1985
onset of overt diabetes by many years. In particular, an antibody reacting with a 64 Kd protein on the surface of the fl cell is seen early in patients who later have diabetes, and in spontaneously diabetic BB rats before the time at which /] cell loss can first be detected (Lernmark). Whether this antibody causes cell destruction and by what means, is not known. J. Nerup (Copenhagen) has shown that/~ cells can be destroyed in vitro by supernatants of mitogen-activated human mononudear cells, and the active principle appears to be interleukin 1 (IL- 1). This would be a unique function for IL-I, but it was consistent with electron-micrographs ofinsulitis in one experimental animal system, the BB rat, obtained by H. Kolb (Dusseldorf), who observed macrophages as the initial infikrating cells in the islets. However, IL-1 destroys all islet cells in vitro, while fl cells appear to be the only target of autoimmune insulitis in vivo. Two types of cytotoxic T cells (CTL) were discussed. C. Janeway (Yale) described T4 ÷ C T L that could kill targets sensitive to lymphotoxin, presumably by lymphotoxin release. Such cells are restricted in antigen recognition by class II M H C gene products, and can kill bystander targets. Thus/3 cells could be destroyed by such cells if they were particularly sensitive to this mechanism of killing. As such C T L are class II restricted, the association with H L A - D could be accounted for by such cells. In addition, E. Simpson (Harrow) described a correlation between rejection of skin grafts differing for the male antigen H-Y with delayed-type hypersensitivity. Graft rejection is under genetic control by class II M H C genes, while classical eytolytic T-cell responses are controlled both by class I and class II M H C antigens and do not correlate with skin graft rejection. The possible role of class I-restricted killer cells in diabetes was rendered more problematic by the finding reported by F. Sheinvold (Miami) that/3 cells of several species express little or no class I M H C gene products, a finding that was not supported by data from other investigators; nevertheless, it was clear that the level of class I products on the fl cell might be low. The expression of class n products in the islets was the subject of significant discussion. Sheinvold found it on capillary, endothelium, but not on other cell types. Interestingly, A. Cooke (London) found class II antigens on capillary endothelial cells in pre-diabetic BB rats but not in normal pancreas, and only observed class n antigens on fl cells late in the development of diabetes.
Whether/J cell class 11 molecules are endogenously synthesized or passively acquired from infiltrating macrophages and B cells is not known. Finally, A. Rabinovitch (Miami) demonstrated that cells indistinguishable from natural killer cells appeared to be selectively cytotoxic forl$ cells in vitro. This finding was questioned on the ground that nude mice, and mice and rats with suppressed T-cell function, did not develop diabetes, although they have normal to elevated levels of natural killer cells. The relevance of this finding remains to be established. In summary, the nature of the cell that destroys the fl cell selectively, and the mechanism by which it does so, has not been elucidated. Neither has the target structure that allows such effector cells to selectively kill ~ cells, if it exists, been defined. The rapid rate at which syngeneic I] cells are destroyed in the identical twin studies of Sutherland, and the preponderance of T8 cells in these islets, suggests that classic, class I MHC-restricted C T L are the most likely candidate for the destruction o f t cells. This is reinforced by the finding that only islets having fl cells display insulitis, suggesting that the response is ~Jcell specific, and that the fl cells are not uniquely sensitive to a toxic lymphocyte product released in the islets in response to another autoantigen or a virus. However, until such cells are cloned and shown to transfer the disease, questions about the mechanism of fl cell destruction will persist. Experimental models of I D D M The search for an experimental model for I D D M has focused recently o n two animals that have a high propensity to develop diabetes spontaneously with accompanying/3 cell destruction and insulitis: the BB/W rat (E. Marliss, Montreal) and the N O D mouse (S. Makino, Japan). Several studies using these models were presented; these studies illuminated certain issues. In both models, the genetics are complex but have one guiding principle. In breeding studies, susceptibility to I D D M mapped in the M H C in both cases. However, in each model, at least one other gene was important. In the BB rat, the most prominent of these is a gene that causes lymphopenia (Marliss, Eisenbarth) and particularly a deficiency in cells bearing the W3/25 marker. Interestingly, BB rats can be protected from I D D M by transferring T cells carrying the cell surface antigen W3/25 (A. Rossini, Boston), usually associated with T H cells. Janeway and P. Flood (New Haven) both emphasized
231
Immunology Today, voL 6, No. 8, 1985 the importance of referring to such cells in terms that reflected the means by which they were identified rather than the putative function, and Flood pointed out that cells inducing suppression were of the T4 (or helper) phenotype in mice. While T-cell transfers can prevent I D D M in BB rats, thymectomy, anti-lymphocyte serum and cyclosporin A can also prevent the disease, but cannot reverse it after onset. Furthermore, Rossini has also reported passive transfer of I D D M by T cells from acutely diabetic BB rats to various normal, non-diabetic rats. This phenomenon appears to show M H C restriction. As in man, acutely diabetic BB rats have high levels of activated T cells (A. Naji, Philadelphia). These studies, taken together, suggest that T cells are intimately involved in both pathogenesis and protection, a finding that places immunoregulation squarely in the frame of studies of this disease (see below). The N O D mice proved to be an immunogenetic puzzle, lacking detectable class II M H C molecules detectable by a variety of probes (M. Hattori, Joslin). I-E molecules are clearly absent in these mice, as no m R N A was visible on Northern hybridization blots; I-A molecules were not detected with several broadly reactive anti-I-A monoclonal antibodies, but m R N A was present (R. Jackson, Joslin). T h e N O D mouse is not lymphopenic, and has no known immunologic abnormalities. However, nude N O D mice do not develop diabetes unless given n u / + T cells (Makino), a finding consistent with earlier studies in BB/W rats. Interestingly, males are relatively resistant to diabetes unless raised in germ-free conditions, and N O D mice raised at the Walter and Eliza Hall Institute in Melbourne (T. Mandel) had a lower incidence of I D D M than mice raised in Japan. The pathology of experimental I D D M was of interest. In particular, insulitis in the BB rat is far less intense than the inflammatory reaction seen in the N O D mouse. The most striking finding in the N O D mouse was an initial peri-insulitis of great intensity associated with intact islet morphology and function. This suggested that one level of control in autoimmune diabetes might be access of lymphoid cells to the tissue. It is of interest that this should be apparent in mice but not rats, as it has long been known to be more difficult to induce organ-specific autoimmunity in mice than rats, and vascular barriers appear to be more effective in mice than in rats (blood : brain, blood : thymus). Mandel reported a similar peri-insulitis
in islets grafted under the kidney capsule, a n d it is known to be more difficult to graft islets between rats than between mice.
Transplanted/3 ceils as allografts and autografts As I D D M is clearly a disease in which there is loss of a particular cell type, the/3 cell, replacement of/3 cells is the most logical therapy. However, several obstacles confront the clinician who chooses this course of treatment. First, islets need to be prepared in reasonable numbers. It was clear from studies cartied out by D. Mintz, R. Alejandro and colleagues (Miami), that the n u m b e r of: islets grafted to pancreatectomized dogs determined the success of an autograft. A variety of protocols for improving the yield were discussed (P. Morris, London). Second, the i m m u n e response to allografted islets needs to be overcome. Various maneuvers to do so were described. These involved depletion of the islets of stimulatory, class II M H C bearing cells by anti-Ia antibody, by high oxygen or low temperature culture, or by treatment with anti-dendritic cell antibody (J. Davie, St. Louis; H. Lau, Columbia; Mandel: Naji). The host immune system could also be damped using cyclosporin A (Alejandro), antilymphocyte antibody (Davie) or transfusion of anti-Ia treated donor whole blood cells prior to transplant (Davie, Lau). The finding that even diabetic individuals bearing renal grafts from identical twins could specifically reject [J cells from the same genetically identical donor unless immunosuppression was used suggested that grafting islets deliberately mismatched for H L A antigens might be critical in patients with autoi m m u n e I D D M , since recognition of/3 cell-specific antigens, and thus autoimmunity, might be M H C restricted. This was emhasized by W. Silvers (Philadelphia), who finds that acquired tolerance is M H C restricted, in that rats and mice tolerant to allogeneic tissue fail to recognize minor histocompatability or tissue-specific antigens in the context of non-self M H C . Thus, two principles emerge in these studies. First, to study islet transplants as a treatment for diabetes, a spontaneously diabetic animal is the best model. Second, allografts may have major advantages over H L A matched grafts, so prolongation of islet allograft survival is a relevant and important goal of research. A second aspect of transplantation biology that appeared in several presentations involved the nature of tolerance to foreign tissue antigens. In several systems, it was clear that clonal
deletion could not account for the data. For instance, in the studies presented by W. Streilein (Miami), neonatal tolerance could not be broken even by heroic levels of normal T cells, but rather required specifically immune T cells. Furthermore, the doses of cells required to reverse tolerance increased with time, even if the graft being monitored was always given as fresh tissue. These findings are more consistent with active maintenance of the tolerant state than they are with a clonal deletion model of tolerance. Recent data from P. Medawar and from K. Eichmann suggest that G T L precursors are under active regulation in tolerance, consistent with the findings presented here. T h e i m p o r t a n c e of suppressor cells in a u t o i m m u n i t y a n d tolerance This workshop left a very strong impression that whatever the mechanism of/3 cell destruction, in either I D D M or islet-cell transplantation, the critical control parameter was the level of specific immunological suppression. W h a t was most dramatic was the obvious evidence, often mentioned in the past by the late H H M I investigator Richard Gershon, that a phenomenon as fundamental in immunologic dogma as tolerance to M H C antigens was under active immunological control. Furthermore, T ceils are clearly important in the causation of I D D M , as seen by the protection afforded by thymectomy, by the transfer of the disease by activated T cells, and by the infiltration o f T cells in rejection of identical twin pancreas grafts. Yet T cells can also protect against I D D M , which strongly suggests that different populations of T cells performing different functions are responsible for the disease and can also prevent it. Thus, measures that simply decrease total cell levels or activity are unlikely to prevent or 'cure' IDDM. This is dramatically illustrated in recent-onset I D D M patients given cyclosporin A. These patients can frequently be kept off insulin for as long as cyclosporin A is given. However, as soon as treatment is terminated, the disease continues where it left off and the fl cells are eventually totally lost. Likewise, the chronic course ofprimaryfl cell destruction, as opposed to the rapid destruction of transplanted, histocompatible fl cells, suggests that I D D M represents a slight shift in the immunoregulatory balance towards autoimmunity. T h e precarious nature of this balance is further emphasized by the high rate of discordancy for I D D M in identical twins living in the same environment, and the apparent influence of gender, feeding pattern and season (Nerup) on incidence.
Immunology Today, vol. 6, No. 8, 1985
232 If this construction is correct, what is the optimal approach to the prevention of IDDM? This question is highly pertinent, as advances in genetics may soon make it possible to identify high-risk individuals. To paraphrase G. Bernard Shaw, who in turn was mimicking Metchnikov, it seems likely that the key is to learn how to 'stimulate the suppressors', stimulate the suppressors. This is not a simple task. One would like to
stimulate only those suppressors that are relevant, and this requires an understanding of both the pathways of immune suppression and the specific antigen involved. Nevertheless, this analysis clearly shifts the emphasis from the fl cell and the effector cells that destroy it (still critically important issues) to the analysis of suppression mechanisms in general, and to their control and manipulation in I D D M in particular.
Charles A. Janeway, Jr is in the Howard Hughes Medical Institute at Yale University School of Medicine, New Haven, C T 06510, USA.
Flow microfluorimetry--cantinued from p. 229 cells in S phase and survival of patients in Dukes Stages A and B was found, but, in Dukes stages C and D, a low proliferative rate was correlated with a longer survival. Watson used a dual staining protocol with propidium iodide and a FITC labelled monoclonal antibody against a c-myc oncogene product in a retrospective study on testis and on colon cancer. Whilst at the moment insufficient numbers of samples have been analysed to be significant there appears to be a correlation of high anti-c-myc activity with low survival rate. These two studies show that such information allows prognostic judgements to be made from biopsy material, a point emphasised by G. Valet (Martinsried), who gave details of how FMF measurements are currently being used to assess the necessity for surgical intervention in colorectal cancer.
Burkitt's lymphoma patients appeared as a change in the relative fluorescence of the translocated chromosome and its partner. J. A. Aten (Amsterdam) suggested that it may be possible, using the 'slit scan' method, to correlate chromosomal banding patterns using FMF with those obtained by classical cytogenetics to improve the resolution of individual chromosomes in flow karyotype. The large-scale alterations in karyotype mentioned above are fairly easy to detect using FMF. Smaller changes in the chromosomes such as alteration in their higher order structure cannot be visualized using D N A binding fluorochromes alone. One solution t9 this problem is to stain, in conjunction with the traditional D N A dyes, with antibodies directed against specific chromatin structures. Antibodies have the advantage over the DNA-binding dyes in that they may be chromosome specific and the particular site to which they bind can be defined. Green's group used flow karyotype analysis to examine chromosome abnormalities caused by radiation. They found that large doses of radiation (400 fads) altered the flow karyotype (Hoechst 33342 stain) but these levels were lethal! Smaller doses, of between 25 and 50 fads had no obvious effect on the flow karyotype so that detection of any chromosomal changes by this method has proved impossible. Their solution has been to concentrate their search on chromosomes bearing no centromeres. (acentric) or two centromeres (dicentric) using sera from scleroderma patients which contains antibodies which bind to centromeres. A. Fellner, R. Festin and P. Matsson (Uppsala) developed monoclonal antibodies to double stranded and single stranded D N A to study higher order structure in chromosomes. Correlations between the staining activity of propidium iodide and these
monoclonal antibodies reveal that antibody penetration is independent of the degree of the contraction of the chromosomes. B. Turner (Birmingham) also looked at chromosomes in flow using a monoclonal which binds to a highly conserved region (amino acids 1-8) of histone H-2B. Unlike Fellner, Festin and Matsson he found two sub-populations of chromosomes to which the antibody bound - believed to represent differences in higher order chromatin structure. Whilst the ability to analyse chromosomes in the flow microfluorimeter has been with us for some time, sorting out of individual chromosomes now seems to be a practical proposition. M. A. Van Dilla representing the National Laboratory Gene Library Project at Lawrence Livermore (LLNL) and Los Alamos (LANL) USA described the use of FMF technology to produce a chromosome specific gcne library. The first phase of the project, currently under way, is to produce a human DNA library for use in gene mapping, genetic disease diagnosis, linkage and pedigree analysis, using restriction fragment length polymorphism (RFLP) analysis. Later the group hopes to clone larger fragments of up to 100 kb, suitable for the study of gene structure, expression and function. Once these two phases are complete, it is hoped that the mouse genome will be analysed systematically. Whilst in theory the methods for production of these libraries are straightforward, the technical problems on a day to day basis have proved to be immense. Chromosomes are isolated from human fibroblasts (number 13 and smaller) or Chinese hamster/human hybrids (numbers 1-12). The chromosomes of interest (Hoechst 33342 or Chrornomycin A 2 stained) are sorted out using FMF. The 'sorted' chromosomes are digested using Hind III (LLNL) or
Chromosomes Chromosomes can be isolated, stained with DNA-binding fluorochromes and analysed by flow microfluorimetry. The resultant fluorescence profile has been coined the 'flow karyotype'. B. D. Young (London) and D. K. Green (Edinburgh) discussed ways in which flow karyotyping might be used for detecting alterations in chromosome structure or number to assist, for example, cancer detection- which by classical methods is labour-intensive, subjective and slow. The difficulty with application of FMF however is the individual nature of the human flow karyotype. This is due mainly to natural variations in centrometric heterochromatin especially in chromosomes 1, 3, 9, 16, 21, 22 and Y. Even so, using flow karyotype analysis trisomy in Down's syndrome revealed itself as an increased height for the peak representing chromosome 21 and translocations in~ for example, chromosomes from
Acknowledgement The author wishes to thank the participants at the meeting, its sponsor, the Howard Hughes Medical Institute and Drs Rossini, Cahill, Lernmark and Eisenbarth for their help in organizing the meeting and this report. IT]:
