THE JOURNAL OF INFECTIOUS DISEASES. VOL. 137, NO.5. MAY 1978 © 1978 by the University of Chicago. 0022-1899/78/3705-0015$00.75
Recombinant DNA and Autoimmune Disease Jonathan King
From the Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts
In trying to assess the hazards associated with the generation of enteric bacteria carrying a wide variety of foreign DNA sequences, a group of us have been concerned with the spread of foreign genes into natural populations of Escherichia coli (i.e., the potential transfer of plasmids into strains of E. coli already established in the environment). In the course of laboratory or industrial practice, laboratory strains could reach an environment where they encounter indigenous E. coli strains and transfer their plasmids to the wild strains. Even with nonconjugal plasmids, mobilization by other plasmids may occur and lead to maintenance of the nonconjugal plasmid in the wild strain [1, 2]. This transfer could take place in the gut of a technician or graduate student-a few days is quite long enough for the transfer to occur [3, 4]-or in whatever reservoir the E. coli reach after exit from the body in the feces. A sewer or the intestines of pigeons, rats, or chickens would serve equally well [5]. (There has been no discussion at this meeting of the ecology of E. coli outside of the human body.) These strains then move back into the human ecosystem -from chickens to humans, for example, as described by Levy et al. [6], by any of the myriad routes through which coliforms reach our gastrointestinal tract. Though it is certainly true that, in general, strains carrying unrelated genetic sequences will be less fit than others, present experiments involve linkage of the foreign DNA to sequences which confer antibiotic resistance on the host organism. This counterselection will often be so potent that genetic linkages will preserve foreign sequences for long periods [1,2]. Given the increasing frequency of nosocomial infections due to E. coli [7], there is reason to be concerned about human infection by indigenous Please address requests for reprints to Dr. Jonathan King, Department of Biology, Building 16, Room 535, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139.
663
E. coli strains. In Boston, for example, Myerowitz et al. [8] reported 10 episodes of gram-negative bacillemia per 1,000 admissions with a 25% mortality rate. The single most commonly identified species was E. coli. Their evidence suggested that 40% of the infections were contracted from the patient's own urinary tract, 8% from the respiratory tract, 19% from the gastrointestinal tract, and 5% from intravenous catheterization. Similarly, McGowan et al. [9] reported two cases of E. coli bacteremia per 1,000 admissions, with a case fatality rate of 35%-45%. Despite claims to the contrary, there is little evidence that these infections stem from highly specialized strains. For example, Wachsmuth et al. [10] isolated E. coli strains from 152 women with urinary tract infections and found that none were invasive, on the basis of the guinea pig eye test, or produced toxin. The evidence presented earlier during this meeting by 0rskov makes it clear that a variety of E. coli strains are capable of generating urinary tract infections and are, therefore, candidates for causing bacteremia in a compromised patient. Although resistance to serum and other factors needed for systemic infections, or pyelonephritis, are probably restricted to a subset of the normal flora, we know that these microorganisms carry plasmids [11, 12] and therefore are potential hosts for foreign DNA linked to plasmids. Autoimmune Disease
Several kinds of pathologies might result from infection by chimeric E. coli strains displaying foreign proteins. One model is an autoimmune condition associated with exposure to antigens cross-reacting with human antigens. It is now evi. dent that presentation to the body of a homologous antigen in an unusual state can generate an immune response which then damages the normal tissue and results in an autoimmune disease. Wellstudied systems include experimental allergic encephalomyelitis [13] and experimental antiglo-
664
merular basement membrane nephritis [14]. In the former case, injection of central nervous system tissue results in immune damage to the nervous system. In the latter case, injection of homologous or heterologous renal tissue results in the induction of antibodies which cause glomerulonephritis. Injection of the antibody alone, in the absence of antigen, induces the disease and establishes the autoimmune nature of the pathology [14]. Both of these are thought to be models for autoimmune disease normally occurring as a secondary consequence of injury to the normal tissue, whether of mechanical, toxic, or infectious origin. If a cross-reacting antigen were to be presented in the background of the gram-negative cell surface, which is a potent inducer of the immune response, there is every reason to suspect that an immune response against the foreign antigen would result. As will be discussed below, many recombinant DNA experiments are likely to generate strains of bacteria that display an antigen cross-reacting with human proteins or antigens. Character and Course of the Syndrome
There are considerable data from a relevant model system (the immune diseases which follow infection by certain group A.streptococci) to show the course of autoimmune disease. Among the well-studied examples are rheumatic fever [15] and acute glomerulonephritis (AGN) following streptococcal infections [5, 16]. Although the exact nature of the pathology is unclear, whether these are immune complex diseases or true autoimmune responses, it is worth noting that the streptococcal strains exhibit antigens cross-reacting with heart tissue in the first case and with basement membrane in the second [5]. Outbreaks of poststreptococcal AGN have occurred in recent years on an Indian reservation in Red Lake, Minn., and in Israel [17]. In the Israeli outbreak there was a five- to 10-fold increase in people hospitalized for AGN, with 50% of these cases occurring in children "younger than age five. The subject contracts a streptococcal sore throat or skin infection, the immune response is triggered, and the subject then recovers from the infection. Some weeks later this person comes down with AGN, due to the presence of antigens in the infecting strains which
King
cross-react with antigens of glomerular basement membrane. The condition also can be induced experimentally by injection of kidney tissue into an animal; apparently the self-nonself recognition system is short-circuited when proteins that should be structural and localized appear in the circulatory system. It is not at all clear that this is actually due to antibodies against streptococci directly attacking the basement membrane, rather than an immune-complex disease [14]. Rapaport and coauthors [18] did report, however, that animals perfused with antistreptococcal serum, and not directly exposed to the antigen, developed kidney damage. However, this finding requires further confirmation. The above models of autoimmune disease involve a normal response to an unusual antigen. There are also pathologies thought to be associated with hypersensitive responses to common antigens; Parish has reviewed this response for bacterial antigens [19]. A particularly interesting example, with respect to E. coli, is Chron's disease. Shorter et al. [20] have proposed that this inflammatory bowel disease is due to an unusual immune response to certain intestinal E. coli strains. One of the characteristics of the introduction of potent new technologies, such as recombinant DNA, is rapid proliferation of its applications. This proliferation means that a continually increasing variety of eucaryotic genes are being spliced to plasmids and introduced into E. coli. With respect to the autoimmune model, each different, exposed foreign antigen has the capacity of inducing a different form of autoimmune damage. Since the symptoms can be expected to appear only after the infection is gone, they can be very difficult to diagnose correctly or track down, much less treat. For example, Lasch et al. [17] reported that antibiotic therapy during the Israeli outbreak neither prevented subsequent AGN nor limited the severity of the disease. Furthermore, Baldwin and Schact [21] found that a substantial fraction of cases of poststreptococcal AGN progressed to chronic dysfunction. Findings contrary to all of these cases exist. A fuller discussion is given by Glassock and Bennett [5]. Note that the induction of these immunity-related diseases does not require an initially virulent infection. For example, the streptococcal skin infections are quite mild.
Autoimmune Disease
Risk Assessment
With respect to chimeric E. coli, the primary population at risk would be the approximately 30,000 people who contract nosocomial infections due to E. coli each year [7, 22]. These individuals have been colonized by E. coli strains that are noncliarrheagenic and would be expected to generate an immune response. In addition, the hospital environment provides strong selection for strains carrying antibiotic-resistance plasmids. Chimeric plasmids generated, for example, in hospital research laboratories might become established if they were to escape containment. My discussion has focused on the problem of exposure to enteric bacteria presenting antigens cross-reacting with human antigens. Major problems might also exist with respect to viruses which have been constructed to incorporate cross-reacting mammalian proteins into their envelope. For example, Goodpasture's disease, which resembles anti-GBM (glomerular basement membrane) disease, is associated with prior influenza infection [14]. Clearly, it is difficult to predict how frequently foreign proteins coded for by recombinant DNA will insert into the bacterial membrane and be exposed. Note, however, that this situation may require only the preservation of general hydrophobic properties and not the full biological function of the protein. In fact, for technical reasons, proteins that do insert into the cell envelope and are available to antibodies will probably be selected. This is an excellent method of screening colonies for the expression of foreign proteins. In addition, the very problem of insertion and transport across membranes will be studied by introduction of eucaryotic proteins that have these properties into E. coli. If transport requires additional genes, someone will introduce the additional genes. Note also that cross-reactivity for many essential proteins, such as the cytochromes, is distributed broadly throughout the eucaryotes. It is quite possible that proteins of Drosophila could induce autoimmune responses in humans. These risks can be dealt with by developing adequate containment procedures and facilities. They can be developed, however, only in a climate in which it is possible to assess seriously the hazards attendant to the failure to employ adequate containment standards.
665
References I. Stewart, F. M., Levin, B. R. The population biology of bacterial plasmids: a priori conditions for the existence of conjugationally transmitted factors. Genetics 87:209-228, 1977. 2. Levin, B. R., Stewart, F. M. The probability of establishing chimeric plasmids in natural populations of bacteria. Science 196:218-220, 1977. 3. Anderson, J. 0., Gillespie, W. A., Richmond, M. H. Chemotherapy and antibiotic resistance transfer between enterobacteriaceae in the human gastrointestinal tract. J. Med. Microbiol. 6:475-486,1974. 4. Anderson, E. S. Viability and transfer of a plasmid from Escherichia coli K12 in the human intestine. Nature 255:502-504,1975. 5. Glassock, R. J., Bennett, C. M. The glomerulopathies. In B. M. Brenner and F. C. Rector, Jr. [ed.]. The kidney. Saunders, Philadelphia, 1976, p. 941-1078. 6. Levy, S. B., Fitzgerald, G. B., Macone, A. B. Changes in intestinal flora of farm personnel after introduction of a tetracycline supplemented feed on a farm. N. Engl. J. Med. 295:583-588,1976. 7. McCabe, W. R. Immunoprophylaxis of gram-negative bacillary infections. Annu. Rev. Med. 27:335-341, 1976. 8. Myerowitz, R. L., Medeiros, A. A., O'Brian, T. F. Recent experience with bacillemia due to gram-negative organisms. J. Infect. Dis. 124:239-246, 1971. 9. McGowan, J. E., Jr., Barnes, M., Finland, M. Bacteremia at Boston City Hospital: occurrence and mortality during 12 selected years (1935-1972), with special reference to hospital-acquired cases. J. Infect. Dis. 132:316-335, 1973. 10. Wachsmuth, 1. K., Stamm, W. E., McGowan, J. E. Prevalence of toxigenic and invasive strains of Escherichia coli in hospital populations. J. Infect. Dis. 132:601-603, 1975. II. Datta, N. R factors in Escherichia coli. Ann. N.Y. Acad, Sci. 182:59-64, 1971. 12. Moorhouse, E. Prevalence of R + bacteria in infants in Ireland. Ann. N.Y. Acad. Sci. 182:65-71, 1971. 13. Paterson, P. Y. Immune processes and infectious factors in central nervous system disease. Annu. Rev. Med. 20:75-100,1969. 14. Wilson, C. B., Dixon, F. J. The renal response to immunological injury. In B. M. Brenner and F. C. Rector [ed.], The kidney. Vol. 2. Saunders, Philadelphia, 1976, p. 838-940. 15. Lyampert, 1. M., Danilova, T. A. Immunological phenomena associated with cross-reactive antigens of microorganisms and mammalian tissues. Prog. Allergy 18:423-477, 1975. 16. Johnson, J. C., Stollerman, G. H. Nephritogenic streptococci. Annu. Rev. Med. 20:315-322,1969. 17. Lasch, E. L., Frankel, V., Vardy, P. A., Bergner-Rabinowitz, S., Ofeck, 1., Rabinowitz, K. Epidemic glomerulonephritis in Israel. J. Infect. Dis. 124:141-147, 1971. 18. Rapaport, F. T., Markowitz, A. S., McCluskey, R. T.,
666
Hanaoka, T., Shimada, T. Induction of renal disease with antisera to group A streptococcal membranes. Transplant. Proc. 1:981-984, 1969. 19. Parish, W. E. Damage from hypersensitivity to bacteria. Microbial pathogenicity in man and animals. 22nd Symposium of the Society for General Microbiology, Cambridge University Press, Cambridge, England, 1972. 20. Shorter, R. G., Huzenga, K. A., Spencer, R. J. A work-
King
ing hypothesis for the etiology and pathogenesis of nonspecific inflammatory bowel diseases. Am. J. Diges. Dis. 17:24-32, 1972. 21. Baldwin, D. S., Schaer, R. G. Late sequelae of poststreptococcal glomerulonephritis. Annu, Rev. Med. 27:49-55,1976. 22. Gangarosa, E. J. Epidemiology of Escherichia coli in the United States. J. Infect. Dis. 137:634-638, 1978.
THE JOURNAL OF INFECTIOUS DISEASES. VOL. 137, NO.5. MAY 1978
Discussion DR. N. PIERCE. The discussions at this meeting have mostly been concerned with noninvasive bacteria that could potentially reside for a brief period of time in the gut, or with the products of these organisms that could react with the host. These products are usually delivered to an intact mucosal surface in an immunologically normal individual. The model proposed by Dr. King doesn't fit this situation for two reasons. First, a streptococcal pharyngeal infection.is not a gut infection. There are anatomical and immunological differences between these organisms. The pharynx lacks a submucosal layer of IgA cells and does not have the same local immunological system as the gut. Thus, an infection in the pharynx is not an ideal analogy for one in the intestine. Second, it appears that the streptococcal infection must produce an inflammatory reaction in order to initiate subsequent nephritis or rheumatic fever. This inflammatory component alters considerably the absorption of antigens and their interaction with host tissue. A major problem with postulating autoimmune disease due to enteric pathogens is the lack of an inflammatory response to normal bowel flora. DR. R. SHORTER. Circulating mononuclear cells from patients with chronic ulcerative colitis are cytotoxic for colonic epithelium. There is strong evidence that the cytotoxicity results from cross-reactivity between human colonic epithelium and Escherichia coli. Patients may have Crohn's disease that is confined to the small bowel, but they have cells cytotoxic for colonic epithelium and not small bowel epithelium. The small bowel epithelium does not cross-react with Enterobacteri aceae.
667
DR. S. GORBACH. Let us return to the issue raised by Dr. King, the possibility that a microorganism could elaborate a product and thereby cause an autoimmune disease in the host. Are there appropriate laboratory tests-for example, the cytotoxicity assays-to answer this question? DR. SHORTER. Yes, it might be possible in a rat model because rat colon cross-reacts with human colon-especially with the neonatal rat model. Cross-reactivity would be the criterion, rather than the pathogenic effect of the E. coli per se. DR. PIERCE. These patients also have antibodies to other tissues, not just to the bowel. So, it's not necessarily a case of a particularly effective antigen; it may be that these individuals have a defect in their immune system. For example, there may be a defect in suppressor T-cell function, rather than bacterial flora, that causes a reaction against their bowel, thyroid, and smooth muscle. DR. SHORTER. Yes, but in contrast to the humoral mechanism, it is thought that anticolon antibodies result from cross-reaction dictated by confrontation with E. coli antigens, rather than from a straight autoimmune response. While humoral antibodies are important in autoimmunity, they don't do much in vitro. Cytotoxicity systems do kill the cell of the target organ, and it appears to be positive evidence that one is reproducing a potential pathogenetic mechanism in a test tube rather than measuring, say, a hemagglutination phenomenon which doesn't have any effect on viability.
Recombinant DNA and Autoimmune Disease Jonathan King
From the Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts
In trying to assess the hazards associated with the generation of enteric bacteria carrying a wide variety of foreign DNA sequences, a group of us have been concerned with the spread of foreign genes into natural populations of Escherichia coli (i.e., the potential transfer of plasmids into strains of E. coli already established in the environment). In the course of laboratory or industrial practice, laboratory strains could reach an environment where they encounter indigenous E. coli strains and transfer their plasmids to the wild strains. Even with nonconjugal plasmids, mobilization by other plasmids may occur and lead to maintenance of the nonconjugal plasmid in the wild strain [1, 2]. This transfer could take place in the gut of a technician or graduate student-a few days is quite long enough for the transfer to occur [3, 4]-or in whatever reservoir the E. coli reach after exit from the body in the feces. A sewer or the intestines of pigeons, rats, or chickens would serve equally well [5]. (There has been no discussion at this meeting of the ecology of E. coli outside of the human body.) These strains then move back into the human ecosystem -from chickens to humans, for example, as described by Levy et al. [6], by any of the myriad routes through which coliforms reach our gastrointestinal tract. Though it is certainly true that, in general, strains carrying unrelated genetic sequences will be less fit than others, present experiments involve linkage of the foreign DNA to sequences which confer antibiotic resistance on the host organism. This counterselection will often be so potent that genetic linkages will preserve foreign sequences for long periods [1,2]. Given the increasing frequency of nosocomial infections due to E. coli [7], there is reason to be concerned about human infection by indigenous Please address requests for reprints to Dr. Jonathan King, Department of Biology, Building 16, Room 535, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139.
663
E. coli strains. In Boston, for example, Myerowitz et al. [8] reported 10 episodes of gram-negative bacillemia per 1,000 admissions with a 25% mortality rate. The single most commonly identified species was E. coli. Their evidence suggested that 40% of the infections were contracted from the patient's own urinary tract, 8% from the respiratory tract, 19% from the gastrointestinal tract, and 5% from intravenous catheterization. Similarly, McGowan et al. [9] reported two cases of E. coli bacteremia per 1,000 admissions, with a case fatality rate of 35%-45%. Despite claims to the contrary, there is little evidence that these infections stem from highly specialized strains. For example, Wachsmuth et al. [10] isolated E. coli strains from 152 women with urinary tract infections and found that none were invasive, on the basis of the guinea pig eye test, or produced toxin. The evidence presented earlier during this meeting by 0rskov makes it clear that a variety of E. coli strains are capable of generating urinary tract infections and are, therefore, candidates for causing bacteremia in a compromised patient. Although resistance to serum and other factors needed for systemic infections, or pyelonephritis, are probably restricted to a subset of the normal flora, we know that these microorganisms carry plasmids [11, 12] and therefore are potential hosts for foreign DNA linked to plasmids. Autoimmune Disease
Several kinds of pathologies might result from infection by chimeric E. coli strains displaying foreign proteins. One model is an autoimmune condition associated with exposure to antigens cross-reacting with human antigens. It is now evi. dent that presentation to the body of a homologous antigen in an unusual state can generate an immune response which then damages the normal tissue and results in an autoimmune disease. Wellstudied systems include experimental allergic encephalomyelitis [13] and experimental antiglo-
664
merular basement membrane nephritis [14]. In the former case, injection of central nervous system tissue results in immune damage to the nervous system. In the latter case, injection of homologous or heterologous renal tissue results in the induction of antibodies which cause glomerulonephritis. Injection of the antibody alone, in the absence of antigen, induces the disease and establishes the autoimmune nature of the pathology [14]. Both of these are thought to be models for autoimmune disease normally occurring as a secondary consequence of injury to the normal tissue, whether of mechanical, toxic, or infectious origin. If a cross-reacting antigen were to be presented in the background of the gram-negative cell surface, which is a potent inducer of the immune response, there is every reason to suspect that an immune response against the foreign antigen would result. As will be discussed below, many recombinant DNA experiments are likely to generate strains of bacteria that display an antigen cross-reacting with human proteins or antigens. Character and Course of the Syndrome
There are considerable data from a relevant model system (the immune diseases which follow infection by certain group A.streptococci) to show the course of autoimmune disease. Among the well-studied examples are rheumatic fever [15] and acute glomerulonephritis (AGN) following streptococcal infections [5, 16]. Although the exact nature of the pathology is unclear, whether these are immune complex diseases or true autoimmune responses, it is worth noting that the streptococcal strains exhibit antigens cross-reacting with heart tissue in the first case and with basement membrane in the second [5]. Outbreaks of poststreptococcal AGN have occurred in recent years on an Indian reservation in Red Lake, Minn., and in Israel [17]. In the Israeli outbreak there was a five- to 10-fold increase in people hospitalized for AGN, with 50% of these cases occurring in children "younger than age five. The subject contracts a streptococcal sore throat or skin infection, the immune response is triggered, and the subject then recovers from the infection. Some weeks later this person comes down with AGN, due to the presence of antigens in the infecting strains which
King
cross-react with antigens of glomerular basement membrane. The condition also can be induced experimentally by injection of kidney tissue into an animal; apparently the self-nonself recognition system is short-circuited when proteins that should be structural and localized appear in the circulatory system. It is not at all clear that this is actually due to antibodies against streptococci directly attacking the basement membrane, rather than an immune-complex disease [14]. Rapaport and coauthors [18] did report, however, that animals perfused with antistreptococcal serum, and not directly exposed to the antigen, developed kidney damage. However, this finding requires further confirmation. The above models of autoimmune disease involve a normal response to an unusual antigen. There are also pathologies thought to be associated with hypersensitive responses to common antigens; Parish has reviewed this response for bacterial antigens [19]. A particularly interesting example, with respect to E. coli, is Chron's disease. Shorter et al. [20] have proposed that this inflammatory bowel disease is due to an unusual immune response to certain intestinal E. coli strains. One of the characteristics of the introduction of potent new technologies, such as recombinant DNA, is rapid proliferation of its applications. This proliferation means that a continually increasing variety of eucaryotic genes are being spliced to plasmids and introduced into E. coli. With respect to the autoimmune model, each different, exposed foreign antigen has the capacity of inducing a different form of autoimmune damage. Since the symptoms can be expected to appear only after the infection is gone, they can be very difficult to diagnose correctly or track down, much less treat. For example, Lasch et al. [17] reported that antibiotic therapy during the Israeli outbreak neither prevented subsequent AGN nor limited the severity of the disease. Furthermore, Baldwin and Schact [21] found that a substantial fraction of cases of poststreptococcal AGN progressed to chronic dysfunction. Findings contrary to all of these cases exist. A fuller discussion is given by Glassock and Bennett [5]. Note that the induction of these immunity-related diseases does not require an initially virulent infection. For example, the streptococcal skin infections are quite mild.
Autoimmune Disease
Risk Assessment
With respect to chimeric E. coli, the primary population at risk would be the approximately 30,000 people who contract nosocomial infections due to E. coli each year [7, 22]. These individuals have been colonized by E. coli strains that are noncliarrheagenic and would be expected to generate an immune response. In addition, the hospital environment provides strong selection for strains carrying antibiotic-resistance plasmids. Chimeric plasmids generated, for example, in hospital research laboratories might become established if they were to escape containment. My discussion has focused on the problem of exposure to enteric bacteria presenting antigens cross-reacting with human antigens. Major problems might also exist with respect to viruses which have been constructed to incorporate cross-reacting mammalian proteins into their envelope. For example, Goodpasture's disease, which resembles anti-GBM (glomerular basement membrane) disease, is associated with prior influenza infection [14]. Clearly, it is difficult to predict how frequently foreign proteins coded for by recombinant DNA will insert into the bacterial membrane and be exposed. Note, however, that this situation may require only the preservation of general hydrophobic properties and not the full biological function of the protein. In fact, for technical reasons, proteins that do insert into the cell envelope and are available to antibodies will probably be selected. This is an excellent method of screening colonies for the expression of foreign proteins. In addition, the very problem of insertion and transport across membranes will be studied by introduction of eucaryotic proteins that have these properties into E. coli. If transport requires additional genes, someone will introduce the additional genes. Note also that cross-reactivity for many essential proteins, such as the cytochromes, is distributed broadly throughout the eucaryotes. It is quite possible that proteins of Drosophila could induce autoimmune responses in humans. These risks can be dealt with by developing adequate containment procedures and facilities. They can be developed, however, only in a climate in which it is possible to assess seriously the hazards attendant to the failure to employ adequate containment standards.
665
References I. Stewart, F. M., Levin, B. R. The population biology of bacterial plasmids: a priori conditions for the existence of conjugationally transmitted factors. Genetics 87:209-228, 1977. 2. Levin, B. R., Stewart, F. M. The probability of establishing chimeric plasmids in natural populations of bacteria. Science 196:218-220, 1977. 3. Anderson, J. 0., Gillespie, W. A., Richmond, M. H. Chemotherapy and antibiotic resistance transfer between enterobacteriaceae in the human gastrointestinal tract. J. Med. Microbiol. 6:475-486,1974. 4. Anderson, E. S. Viability and transfer of a plasmid from Escherichia coli K12 in the human intestine. Nature 255:502-504,1975. 5. Glassock, R. J., Bennett, C. M. The glomerulopathies. In B. M. Brenner and F. C. Rector, Jr. [ed.]. The kidney. Saunders, Philadelphia, 1976, p. 941-1078. 6. Levy, S. B., Fitzgerald, G. B., Macone, A. B. Changes in intestinal flora of farm personnel after introduction of a tetracycline supplemented feed on a farm. N. Engl. J. Med. 295:583-588,1976. 7. McCabe, W. R. Immunoprophylaxis of gram-negative bacillary infections. Annu. Rev. Med. 27:335-341, 1976. 8. Myerowitz, R. L., Medeiros, A. A., O'Brian, T. F. Recent experience with bacillemia due to gram-negative organisms. J. Infect. Dis. 124:239-246, 1971. 9. McGowan, J. E., Jr., Barnes, M., Finland, M. Bacteremia at Boston City Hospital: occurrence and mortality during 12 selected years (1935-1972), with special reference to hospital-acquired cases. J. Infect. Dis. 132:316-335, 1973. 10. Wachsmuth, 1. K., Stamm, W. E., McGowan, J. E. Prevalence of toxigenic and invasive strains of Escherichia coli in hospital populations. J. Infect. Dis. 132:601-603, 1975. II. Datta, N. R factors in Escherichia coli. Ann. N.Y. Acad, Sci. 182:59-64, 1971. 12. Moorhouse, E. Prevalence of R + bacteria in infants in Ireland. Ann. N.Y. Acad. Sci. 182:65-71, 1971. 13. Paterson, P. Y. Immune processes and infectious factors in central nervous system disease. Annu. Rev. Med. 20:75-100,1969. 14. Wilson, C. B., Dixon, F. J. The renal response to immunological injury. In B. M. Brenner and F. C. Rector [ed.], The kidney. Vol. 2. Saunders, Philadelphia, 1976, p. 838-940. 15. Lyampert, 1. M., Danilova, T. A. Immunological phenomena associated with cross-reactive antigens of microorganisms and mammalian tissues. Prog. Allergy 18:423-477, 1975. 16. Johnson, J. C., Stollerman, G. H. Nephritogenic streptococci. Annu. Rev. Med. 20:315-322,1969. 17. Lasch, E. L., Frankel, V., Vardy, P. A., Bergner-Rabinowitz, S., Ofeck, 1., Rabinowitz, K. Epidemic glomerulonephritis in Israel. J. Infect. Dis. 124:141-147, 1971. 18. Rapaport, F. T., Markowitz, A. S., McCluskey, R. T.,
666
Hanaoka, T., Shimada, T. Induction of renal disease with antisera to group A streptococcal membranes. Transplant. Proc. 1:981-984, 1969. 19. Parish, W. E. Damage from hypersensitivity to bacteria. Microbial pathogenicity in man and animals. 22nd Symposium of the Society for General Microbiology, Cambridge University Press, Cambridge, England, 1972. 20. Shorter, R. G., Huzenga, K. A., Spencer, R. J. A work-
King
ing hypothesis for the etiology and pathogenesis of nonspecific inflammatory bowel diseases. Am. J. Diges. Dis. 17:24-32, 1972. 21. Baldwin, D. S., Schaer, R. G. Late sequelae of poststreptococcal glomerulonephritis. Annu, Rev. Med. 27:49-55,1976. 22. Gangarosa, E. J. Epidemiology of Escherichia coli in the United States. J. Infect. Dis. 137:634-638, 1978.
THE JOURNAL OF INFECTIOUS DISEASES. VOL. 137, NO.5. MAY 1978
Discussion DR. N. PIERCE. The discussions at this meeting have mostly been concerned with noninvasive bacteria that could potentially reside for a brief period of time in the gut, or with the products of these organisms that could react with the host. These products are usually delivered to an intact mucosal surface in an immunologically normal individual. The model proposed by Dr. King doesn't fit this situation for two reasons. First, a streptococcal pharyngeal infection.is not a gut infection. There are anatomical and immunological differences between these organisms. The pharynx lacks a submucosal layer of IgA cells and does not have the same local immunological system as the gut. Thus, an infection in the pharynx is not an ideal analogy for one in the intestine. Second, it appears that the streptococcal infection must produce an inflammatory reaction in order to initiate subsequent nephritis or rheumatic fever. This inflammatory component alters considerably the absorption of antigens and their interaction with host tissue. A major problem with postulating autoimmune disease due to enteric pathogens is the lack of an inflammatory response to normal bowel flora. DR. R. SHORTER. Circulating mononuclear cells from patients with chronic ulcerative colitis are cytotoxic for colonic epithelium. There is strong evidence that the cytotoxicity results from cross-reactivity between human colonic epithelium and Escherichia coli. Patients may have Crohn's disease that is confined to the small bowel, but they have cells cytotoxic for colonic epithelium and not small bowel epithelium. The small bowel epithelium does not cross-react with Enterobacteri aceae.
667
DR. S. GORBACH. Let us return to the issue raised by Dr. King, the possibility that a microorganism could elaborate a product and thereby cause an autoimmune disease in the host. Are there appropriate laboratory tests-for example, the cytotoxicity assays-to answer this question? DR. SHORTER. Yes, it might be possible in a rat model because rat colon cross-reacts with human colon-especially with the neonatal rat model. Cross-reactivity would be the criterion, rather than the pathogenic effect of the E. coli per se. DR. PIERCE. These patients also have antibodies to other tissues, not just to the bowel. So, it's not necessarily a case of a particularly effective antigen; it may be that these individuals have a defect in their immune system. For example, there may be a defect in suppressor T-cell function, rather than bacterial flora, that causes a reaction against their bowel, thyroid, and smooth muscle. DR. SHORTER. Yes, but in contrast to the humoral mechanism, it is thought that anticolon antibodies result from cross-reaction dictated by confrontation with E. coli antigens, rather than from a straight autoimmune response. While humoral antibodies are important in autoimmunity, they don't do much in vitro. Cytotoxicity systems do kill the cell of the target organ, and it appears to be positive evidence that one is reproducing a potential pathogenetic mechanism in a test tube rather than measuring, say, a hemagglutination phenomenon which doesn't have any effect on viability.
