182
Concise Communications
4. Jonsson S. Musher OM. Chapman A. Goree G. Lawrence EC. Phagocytosis and killing of common bacterial pathogens of the lung by human alveolar macrophages. 1 Infect Dis 1985; 152:4-13. 5. Sutton A. Schneerson R, Kendall-Morris S. Robbins lB. Differential complement resistance mediates virulence of Haemophilus influenzae type b. Infect Immun 1982;35:95-104. 6. Musser 1M. Barenkamp Sl, Granoff OM. Selander RK. Genetic relationships of serologically nontypable and serotype b strains of Haemophilus influenzae. Infect Immun 1986;52: 183-91. 7. Kimura A. Hansen El. Antigenic and phenotypic variations of Haemophilus influenzae type b lipopolysaccharide and their relationship to virulence. Infect Immun 1986;51 :69-79. 8. Weiser IN. Williams A. Moxon ER. Phase-variable lipooligosaccharide structures enhance the invasive capacity of Haemophilus influenzae. Infect Immun 1990;58:3455-7. 9. Hoiseth SK. Moxon ER. Silver RP. Genes involved in Haemophilus influenzae type b capsule expression are part of an 18 kilobase tandem duplication. Proc Natl Acad Sci USA 1986;83: 1106-10.
110 1992; 166 (July)
10. Hoiseth SK. Connelly CJ. Moxon ER. Genetics of spontaneous. high frequency loss of b capsule expression in Haemophilus infiuenzae. Infect Immun 1985;49:389-95. II. Moxon ER. Deich RA. Connelly C. Cloning of chromosomal DNA from Haemophilus infiuenzae. Its use for studying expression of type b capsule and virulence. 1 Clin Invest 1984;73:298-306. 12. Williams AE. Maskell 01. Moxon ER. Relationship between intracellular survival in macrophages and virulence of Haemophilus influenzaetype b. 1 Infect Dis 1991;63:1366-9. 13. Gnehm HE. Pelton SI, Gulati S. Rice PA. Characterization of antigens from nontypable Haemophilus infiuenzae recognized by human bactericidal antibodies. 1 Clin Invest 1985;75: 1645-58. 14. Noel Gl. Mosser OM. Edelson Pl. Role of complement in mouse macrophage binding of Haemophilus infiuenzae type b. 1 Clin Invest 1990;85:208-18. 15. Noel Gl, Katz S. Edelson Pl. Complement-mediated early clearance of Haemophilus influenzae type b from blood is independent of serum lytic activity. 1 Infect Dis 1988; 157:85-90.
Graham H. Bothamley, John Swanson Beck, Robert C. Potts, John M. Grange, Thomas Kardjito, and J uraj Ivanyi
Medical Research Council Tuberculosis and Related Infections Unit. Royal Postgraduate Medical School. and Department ofMicrobiology. Royal Brompton Hospital. London. and Department of Pathology. Ninewells Hospital and Medical School. Dundee. United Kingdom; Department of Pulmonology. Airlangga University. Surabaya. Indonesia
Specific antibody levels and delayed-type hypersensitivity skin responses to antigens of Mycobacterium tuberculosis in 39 hospital staff who were heavily exposed to tuberculosis (TB) were compared with those in 36 factory employees from Indonesia. Antibody levels to the TB68 epitope of the 14-kDa antigen were significantly greater, while titers to the TB23 (19-kDa) and TB72 (38-kDa) epitopes and lipoarabinomannan (LAM) were lower in exposed than in nonexposed subjects (all P < .02). The intensity of tuberculin responses correlated positively with anti-LAM and negatively with anti-19-kDa antibody levels. Possible reasons for the selective humoral response of chronically exposed healthy subjects to the 14-kDa antigen, but not to other antigens immunogenic in patients with tuberculosis, are discussed. Despite recent advances in the immunologic study of mycobacterial antigens, little progress has been made in the allocation of either protective or pathogenic function to any antigenic constituent [I]. Patients with tuberculosis (TB) have abundant T cell-mediated and humoral immunity, but it is not known whether disease occurs because the immune repertoire fails in its effector properties or whether the speciReceived 14 November 1991; revised 27 January 1992. Informed consent was obtained from all patients and controls; ethical approval was obtained from the National Heart and Chest Hospitals and Surabaya University. Financial support: the Wellcome Trust. Reprints or correspondence (present address): Dr. Graham Botharnley, Department of Chest Medicine. University College Hospital. Gower St., London WCIE 6AU. UK. The Journal of Infectious Diseases 1992;166:182-6 © 1992 by The University of Chicago. All rights reserved. 0022-1899/92/6601-0030$01.00
ficity of its repertoire is associated with the pathogenic outcome. Definition of the antigens recognized by the immune system in healthy individuals would be of particular interest but is hampered by the complexity of immunogenic stimuli due to environmental or commensal mycobacteria, selfhealing exposure to tubercle bacilli in childhood, and Mycobacterium bovis-bacille Calmette-Guerin (BCG) vaccination. A special opportunity for study is represented by occupational exposure to TB, where adults remain healthy despite prolonged contact with infectious patients. Such exposure to TB has been associated with delayed-type hypersensitivity (DTH) to tuberculin and a greater risk of developing TB [2]. However, a high rate of tuberculin conversion overestimates the incidence ofTB [3]. An early (6-8 h) erythematous reaction to intradermal tuberculin in hospital staff exposed to patients with active pulmonary TB has been interpreted as
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
Specificity of Antibodies and Tuberculin Response after Occupational Exposure to Tuberculosis
1ID 1992; 166 (July)
Concise Communications
the expression of particularly strong DTH [4]. Moreover, a reciprocal relationship between the intensity of skin tuberculin reactions and IgM, IgG, and IgA antibodies to M. bovisBeG in hospital contacts has been reported [5]. This report describes differences between healthy exposed and nonexposed individuals in terms of antigenic specificity or levels of antibodies and their relationships to DTH.
Materials and Methods
Solid-phase antibody competition test. The method used was essentially that of Hewitt et al. [7]. Briefly, wells of polyvinyl flexible microtiter plates (M24; Dynatech) were coated with 50 ~l of a soluble extract of M. tuberculosis H37Rv at 30 JoLg/ml overnight at 4°e. Nonspecific binding was blocked with 3%bovine serum albumin (BSA) for 1 h at 20°e. Four dilutions of serum (1:5,1:25,1:125, and 1:625) were prepared in 3% BSA and 25 JoLI incubated in duplicate wells at 20°C for 4 h. Without washing, 25 JoLI of 125I-labeled MAb, diluted to give 1000-2000 cpm when incubated in wells without competing serum (the high control), was added to each well and plates incubated overnight at 4°e. After being washed repeatedly, wells were counted in a 1260 Multigamma counter (LKB Instruments, Rockville, MD). Antibody (inhibition) titers were calculated as the dilution of serum that gave 50% inhibition of binding of the labeled MAb. Statistical analysis. Antibody levels were compared using the Mann-Whitney U test. The diameters of tuberculin responses in the two groups were compared by Student's t test. Correlations were made using Spearman's rank correlation test.
Results Antibody levels to the TB68 epitope were significantly greater in hospital staffwhile antibody titers to the TB23 and TB72 epitopes and IgO binding to LAM were lower in hospital staff than in factory employees (table I). The difference Table 1. Serum antibody and delayed-type hypersensitivity (DTH) responses to tuberculin in hospital and factory workers.
Epitope. antigen
Hospital workers (n = 39)
Factory workers (n = 36)
p*
Inhibition antibody titer. median (geometric mean). range TB23 (19 kDa) TB68 (14 kDa) TB72 (38 kDa)
19.5 (19.6). 6-68 5 (5.3). 0-57 t 2 (1.7).0-7
26 (27.4). 11-76 2 (2.4). 0-15 4.5 (4.5). 0-13
.01 .02 .001
Antibody binding titer. median (geometric mean). range 19 kDa (IgG) 65 kDa (IgG) LAM (IgG)
0(2.3).0-30 20 (18.3). 0-237 9.5 (12.0). 0-448
0(3.4).0-125 41 (31.7).0-161 23 (30.8). 0-256
NS NS .015
No. of individuals Tuberculin DTH 20mm NOTE.
o
12
3
9
\I 10
27
o
NS. not significant; LAM. lipoarabinomannan.
* Mann-Whitney U test. t Greater proportion of patients with tuberculosis had antibody titers of >5 to TB68 epitope (79% vs. 43%). but only 8 of 118 patients had titers of >57 (data not shown).
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
Subjects. Thirty-nine staff of the William Booth Hospital (Surabaya, Indonesia) with occupational exposure to TB and 36 factory employees from the same socioeconomic group in that town with no such exposure were examined. Thirty-two hospital staff (82%) and 14 factory workers (39%) had received BCG vaccination as children. All subjects were residents of East Java, had normal chest radiographs, and were clinically well. Serum samples were obtained at the time of skin testing and were stored at - 20°e, transported to London on solid carbon dioxide (-40°C), and stored at -20°C until analyzed. DTH. Tuberculin PPD (purified protein derivative)-RT23 (10 IU) was injected intradermally on the volar aspect of the forearm. A ballpoint pen was used to define the edge of induration at 48 h; a ruler measured the diameter of induration, to the nearest millimeter, in the long axis of the forearm and at right angles, and the geometric mean of these two measurements was used as the tuberculin response. Monoclonal antibodies (MAbs). Three MAbs, TB23 (19-kDa antigen), TB68 (14-kDa antigen), and TB72 (38-kDa antigen), were used to prepare purified antigens and as reagents in the antibody competition assay (see below). These MAbs had been evaluated for their binding to extracts from different mycobacterial species and relative molecular mass of the respective antigens by independent laboratories in a World Health Organization workshop: TB68 and TB72 showed binding to Mycobacterium tuberculosis complex alone, while TB23 showed binding to M. tuberculosis. Mycobacterium kansasii, Mycobacterium marinum, and Mycobacterium duvalii [I]. Antigens. The I 9-kDa antigen was purified by affinity chromatography using MAb TB23 as previously described [6]. The 65-kDa antigen is a recombinant protein and lipoarabinomannan (LAM) is a glycolipid common to mycobacteria. ELISA. Microtiter plates (M 129B; Dynatech, Billingshurst, UK) were coated with 0.2 ~g/ml 19-kDa antigen, 1 ~g/ml 65kDa antigen, 0.1 ~g/ml LAM, or PBS overnight at 4°e. Nonspecific binding was blocked with PBS containing 0.05% Tween 20 (Sigma, Poole, UK) and 1% skimmed milk (Sainsbury, London; PBSTM). Four serum dilutions (1:5, 1:25, 1:125, and I:625) in PBSTM were incubated for I h at 37°C in duplicate wells and, after washing, with goat anti-human IgG peroxidase conjugate. Bound peroxidase activity was measured using O. I mg/ml tetramethylbenzidine in 0.1 M citrate buffer, pH 5.2, containing 0.01% hydrogen peroxide for 10 min. The reaction was stopped by the addition of0.5 M sulfuric acid, and optical densities were measured at 450 nm. The antibody binding titer was calculated as the dilution of serum giving 30% of the plateau value of a standard hyperimmune serum.
183
184
Concise Communications
between hospital and factory staff in TB68 titers was also clearly demonstrated when groups with and without BCG vaccination were compared separately (figure IA). The decrease in TB23, TB72, and LAM antibodies remained significant when only BCG-vaccinated hospital staff and factory employees were compared. However, there were no significant differences in any of the antimycobacterial antibody levels between BCG-vaccinated and non vaccinated factory employees. In BCG-vaccinated factory workers, IgG antibody levels to the 19-kDa antigen showed a significant correlation with anti-LAM IgG (r = .70, P < .05) but not with titers to the TB23 epitope. The diameters of the tuberculin skin reaction were much greater in hospital staff than in factory workers (table I). A tuberculin reaction of induration ~ 15 mm in diameter was found in 3 hospital staff and in 23 factory employees (70%; X2 = 8.7, P < .0 I). However, no significant difference in tuberculin reactions between BCG-vaccinated and nonvaccinated factory employees was observed. Three vaccinated factory employees did not respond to tuberculin.
BCG+
Discussion This study demonstrated significant associations of antibody levels to distinct antigens of M. tuberculosis and skin DTH responses with occupational exposure to TB. Antibody titers to the TB68 (14-kDa) epitope were higher in hospital staff than in factory workers. Selectively raised antibody levels to either the 14-kDa antigen or its TB68 epitope have previously been described in sera of healthy subjects after occupational exposure to TB [8], in family contacts of patients with smear-positive pulmonary TB [6], in children
BCG15
20
o ~
fI)
o
Factory (n=14) Hospital (n=30)
f:J
t5 Q) E ~
The intensity of the DTH response to tuberculin was positively correlated only with antibody levels to LAM in BCGvaccinated hospital staff (r = .43, P = .00 I; figure I B), but there was no relationship between anti- TB68 antibody levels and the diameter ofthe tuberculin response (r = -.05). Moreover, a significant negative correlation was found between DTH reactions and anti-19-kDa antibody levels in factory employees (r = -.51, P = .02; figure IB).
Factory (n=22) Hospital (n=7)
10 10
'0 c:i
z
-
-g E
>-
"'C
2.0
.8
.0
as
E as
1.5
as
~
~
~
1.0
g>
0.5
E as
1.5 1.0
0) T""' I
~
0.5
C)
o
....J
...J
30
9
Figure 1. Antibody levels and delayedtype hypersensitivity (DTH) responses in hospital workers frequently exposed to patients with tuberculosis and in factory employees. A, Antibody titers to TB68 epitope of 14kDa antigen in bacille Calmette-Guerin (BCG)-vaccinated (BCG+) and nonvaccinated (BCG-) groups. B, Correlations ofantiLAM (lipoarabinomannan) in hospital staff and anti-19-kDa IgG antibody titers in factory employees with DTH responses to tuberculin.
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
A
JID 1992; 166 (July)
JlD 1992; 166 (July)
Concise Communications
of the immunomodulatory activity ofLAM permitted greater T cell reactivity and hence larger tuberculin responses. This suggestion agrees with the finding of a reduced anti-LAM response in children with disseminated rather than localized TB [15], since disseminated TB has generally been associated with diminished DTH reactions. Alternatively, the positive correlation between anti-LAM titers and DTH to PPD might merely indicate the overall strength of the immune response in individuals exposed to tuberculosis. Acknowledgements
We thank J. D. A. van Embden (National Institute of Public Health and Environmental Hygiene, Bilthoven, Netherlands) for the purified 65-kDa antigen, produced with support from the UNDP/World Bank/WHO-TDR special program, and P. J. Brennan (Colorado State University, Fort Collins) for purified LAM. References I. Ivanyi J, Sharp K, Jackett PS, Bothamley G. Immunological study of
2.
3.
4.
5.
6.
7.
8.
9.
10.
II. 12.
13.
the defined constituents of mycobacteria. Springer Semin ImmunopathoI1988;10:279-300. Sutherland J. Recent studies in the epidemiology of tuberculosis, based on the risk of being infected with tubercle bacilli. Adv Tuberc Res 1976;19:1-73. de March-Ayuela P. Choosing an appropriate criterion for true or false conversion in tuberculin testing. Am Rev Respir Dis 1990; 141:81520. Gibbs JH, Grange JM, Swanson Beck J, et al. Early delayed-type responses in tuberculin skin tests after heavy occupational exposure to tuberculosis. J C1in PathoI1991;44:919-23. Pitchappan M, Brahmajothi V, Rajaram K, Subramanyam PT, Balakrishnan K, Muthuveeralakshmi P. Immunological spectrum for BCG in hospital contacts from South India. Tubercle 1991;72: 133-9. Jackett PS, Bothamley GH, Batra HV, Mistry A, Young DB, Ivanyi J. Specificity of antibodies to immunodominant mycobacterial antigens in pulmonary tuberculosis. J C1in Microbiol 1988;26:2313-8. Bothamley G, Udani P, Rudd R, Festenstein F, Ivanyi J. Humoral responses to defined epitopes of tubercle bacilli in adult pulmonary and child tuberculosis. Eur J C1in Microbiol Infect Dis 1988;7:63945. Hoeppner VH, Jackett PS, Swanson Beck J, Kardjito T, Grange JM, Ivanyi J. Appraisal of the monoclonal antibody-based competition test for the serology of tuberculosis in Indonesia. Serodiagn Immunother 1987;1:69-77. Chandramuki A, Bothamley GH. Brennan PJ, Ivanyi J. Levels of antibody to defined antigens of Mycobacterium tuberculosis in tuberculous meningitis. J Clin Microbiol 1989; I 7:821-5. Bothamley GH, Rudd R, Festenstein F, Ivanyi J. Clinical value of the measurement ofMycobacteriullltuberculosis-specific antibody in pulmonary tuberculosis. Thorax 1992;47 (in press). Young DB, Garbe TR. Lipoprotein antigens of Mycobacterium tuberculosis. Res Microbiol 1991;142:55-66. Verbon A, Harterskeerl RA, Schuitema A, Kolk AHJ, Young DB, Lathigra R. The 14K antigen of Mycobacterium tuberculosis is related to the alpha-crystallin family of low molecular weight heat shock proteins. J Bacteriol 1992; 174: 1352-9. Havlir DV, Wallis RS, Boom WH, Daniel TM, Chervenak K, Ellner JJ. Human immune response to Mycobacterium tuberculosis antigens. Infect Immun 1991 ;59:665-70.
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
with tuberculous mediastinal lymphadenitis [7], and in the cerebrospinal fluid of patients with tuberculous meningitis [9]. In contrast, low anti-14-kDa antibody levels have been associated with extensive disease and a poor prognosis [10]. All of these data agree with the hypothesis that the 14-kDa antigen is immunogenic in the early stages of infection with M. tuberculosis. The fact that antibody levels to other mycobacterial antigens have not been found in primary or selfhealing infections [6, 7] further emphasizes the importance of the 14-kDa antigen in the early humoral immune response to M. tuberculosis. The 14-kDa antigen can be recovered from disrupted tubercle bacilli but is not readily found secreted into culture supernatants [11]. Perhaps intracellular antigens are the main targets for the immune response when the host is successfully killing mycobacteria, but once a parasitic relationship has been established, secreted antigens, associated with bacterial growth, assume a greater importance [6, 11). The 14-kDa antigen has been sequenced and found to be structurally homologous with proteins belonging to the family of low-molecular-weight heat-shock proteins [12]; heat-shock proteins can be powerful immunogens [1] and may be induced early in the bacterial colonization of host macrophages to facilitate the intracellular survival of M. tuberculosis. MAb TB68 binds to one of the three B cell epitopes, defined by murine MAbs, of the 14-kDa antigen [12]; there appears to be no human antibody response to one (IT-lor F23-49, unpublished data) of the remaining two epitopes, and the humoral response to the third epitope (defined by MAbs F67-8 and F67-16 [12]) has not been examined. Antibody to a recombinant preparation of the 14-kDa antigen, but not to its TB68 epitope, has been found in tuberculinnegative controls [13] (unpublished thesis). Differences in antibody responses to the TB68 and F67-8/16 epitopes between individuals exposed to tubercle bacilli who remain healthy or who develop tuberculosis might therefore warrant further investigation. Anti-19-kDa antibody levels were inversely related to the size of the tuberculin DTH response. A reciprocal relationship between antibodies to whole BCG and DTH reactions to PPD has been recently described [5] and may correspond to priming ofdifferent CD4 + T lymphocyte subsets [14]. The tuberculin responses in factory workers may have been due to exposure to environmental mycobacteria; the known cross-reactivity of the anti-19-kDa MAb, TB23, that binds to extracts from both M. tuberculosis and M. kansasii [1] would be in keeping with this suggestion. In contrast, anti-LAM antibody and DTH were positively correlated. In vitro, LAM has been shown to inhibit antigen processing. trigger the release of tumor necrosis factor from macrophages, inhibit the interferon-v-rnediated activation of macrophages, and cause local vascular damage [15]. Thus, the positive correlation between antibody levels to LAM and tuberculin responsiveness might suggest that neutralization
185
186
Concise Communications
JIO 1992: 166 (July)
14. Leclerq SA, Bretscher PA. T cells expressing delayed-type hypersensitiv-
15. Costello AM. Kumar A. Narayan V. et al. Does antibody to mycobacte-
ity can be derived from a humorally immune lymphocyte population. Eur J Immunol 1987; 17:949-54.
rial antigens. including lipoarabinornannan, limit dissemination in childhood tuberculosis? Trans R Soc Trop Med Hyg 1992 (in press).
Secreted Antigens of Mycobacterium tuberculosis: Characterization with T Lymphocytes from Patients and Contacts after Two-Dimensional Separation Sabine Daugelat, Heinz Guile,* Bernd Schoel, and Stefan H. E. Kaufmann
Department of Immunology. University of Ulm, Germany
Worldwide, >60 million people suffer from tuberculosis, and it is estimated that the disease causes 3 million deaths annually. About 30%-50% of the total world population have primary infections with the etiologic agent, Mycobacterium tuberculosis. Considering the increasing prevalence of tuberculosis in AIDS patients, its international significance is becoming even more acute. Tuberculosis can be treated chemotherapeutically; yet recently, several outbreaks of drug-resistant strains occurred in the United States. Effective control will be achieved best with appropriate vaccines. Because vaccination trials with Mycobacterium bovis BCG have given contradictory results, currently various attempts are made to develop a subunit vaccine composed of defined mycobacterial antigens in a suitable vector system (reviewed in [1D· M. tuberculosis lives and replicates inside mononuclear phagocytes, and immunity crucially depends on T lymphocytes. Thus, potential vaccine antigens must be characterReceived 29 July 1991; revised 18 February 1992. The appropriate guidelines for human experimentation of the Federal Republic of Germany were followed in the conduct of this research. Grant support: German Leprosy Relief Association (to S.H.E.K.); Graduate College of Biomolecular Medicine (to S.D.). Reprints or correspondence: Dr. S. H. E. Kaufmann. Department of Immunology. University of'Ulm, Albert-Einstein-Allee 11.0-7900 Ulm, Germany. * Present affiliation: Department of Veterinary Pathology and Public Health. Massey University. Palmerston North. New Zealand. The Journal of Infectious Diseases 1992;166:186-90 © 1992 by The University of Chicago. All rights reserved. 0022-1899/92/6601-0031 $0 1.00
ized at the T cell level. So far, the potential of bacterial cell wall and cytoplasmic antigens to stimulate T lymphocytes has been studied [I]. Processing and presentation to T cells of these antigens should depend on prior death and degradation of the microorganisms. However, at least at early stages of infection, such antigens would be available only to a limited degree. In contrast, proteins secreted by bacteria persisting inside macrophages should be readily accessible to the antigen processing machinery. Hence, T cell reactivity to proteins secreted by actively growing bacteria could serve as a model for an early stage of infection when pathogens still live within phagosomes. Recently a method has been developed in our laboratory that allows direct T cell screening of hundreds ofdifferent protein fractions after two-dimensional separation [2]. This technique was used to investigate the repertoire of T cells from tuberculosis patients and healthy contacts toward secreted antigens of M. tuberculosis.
Materials and Methods Preparation of antigens. M. tuberculosis H37Rv organisms were grown in Dubos broth base (Difco, Detroit) supplemented with 10% Dubos medium albumin (Difco) until mid-log phase. Aliquots of 5 ml containing 4.8 X 108 cfu/ml were stored at -70°C. For antigen preparations, aliquots were thawed. washed, inoculated in 500 ml of zinc-free Sauton medium [3]. and incubated at 3rC at 100 rpm for 3 weeks. Cultivation in this medium is known to give a particularly high yield of secreted proteins [4]. Bacteria were sedimented, and supernatant was filtered twice through 0.22-JLm filter units (Schleicher &
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
Little is known about T cell antigens involved in immunity against Mycobacterium tuberculosis. Most model systems use in vitro culture of human T lymphocytes with bacterial Iysates or secreted proteins as antigens. In this study, proteins from 3-week-old M. tuberculosis culture filtrates were separated by two-dimensional PAGE and subsequently transferred into soluble phase. The resulting 480 fractions were screened with T lymphocytes from tuberculosis patients and healthy contacts. T cells from all 9 patients and from 8 of 10 tuberculin-positive contacts preferentially responded to a cluster of acidic proteins (pl 4-5) with molecular masses of 30-1 00 kDa, although they also recognized a number of other fractions. In contrast, of 7 tuberculin-negative contacts, 4 were not and 3 were only weakly stimulated by this cluster region. Therefore, this distinct cluster of secreted proteins seems to comprise dominant T cell antigens of M. tuberculosis.
Concise Communications
4. Jonsson S. Musher OM. Chapman A. Goree G. Lawrence EC. Phagocytosis and killing of common bacterial pathogens of the lung by human alveolar macrophages. 1 Infect Dis 1985; 152:4-13. 5. Sutton A. Schneerson R, Kendall-Morris S. Robbins lB. Differential complement resistance mediates virulence of Haemophilus influenzae type b. Infect Immun 1982;35:95-104. 6. Musser 1M. Barenkamp Sl, Granoff OM. Selander RK. Genetic relationships of serologically nontypable and serotype b strains of Haemophilus influenzae. Infect Immun 1986;52: 183-91. 7. Kimura A. Hansen El. Antigenic and phenotypic variations of Haemophilus influenzae type b lipopolysaccharide and their relationship to virulence. Infect Immun 1986;51 :69-79. 8. Weiser IN. Williams A. Moxon ER. Phase-variable lipooligosaccharide structures enhance the invasive capacity of Haemophilus influenzae. Infect Immun 1990;58:3455-7. 9. Hoiseth SK. Moxon ER. Silver RP. Genes involved in Haemophilus influenzae type b capsule expression are part of an 18 kilobase tandem duplication. Proc Natl Acad Sci USA 1986;83: 1106-10.
110 1992; 166 (July)
10. Hoiseth SK. Connelly CJ. Moxon ER. Genetics of spontaneous. high frequency loss of b capsule expression in Haemophilus infiuenzae. Infect Immun 1985;49:389-95. II. Moxon ER. Deich RA. Connelly C. Cloning of chromosomal DNA from Haemophilus infiuenzae. Its use for studying expression of type b capsule and virulence. 1 Clin Invest 1984;73:298-306. 12. Williams AE. Maskell 01. Moxon ER. Relationship between intracellular survival in macrophages and virulence of Haemophilus influenzaetype b. 1 Infect Dis 1991;63:1366-9. 13. Gnehm HE. Pelton SI, Gulati S. Rice PA. Characterization of antigens from nontypable Haemophilus infiuenzae recognized by human bactericidal antibodies. 1 Clin Invest 1985;75: 1645-58. 14. Noel Gl. Mosser OM. Edelson Pl. Role of complement in mouse macrophage binding of Haemophilus infiuenzae type b. 1 Clin Invest 1990;85:208-18. 15. Noel Gl, Katz S. Edelson Pl. Complement-mediated early clearance of Haemophilus influenzae type b from blood is independent of serum lytic activity. 1 Infect Dis 1988; 157:85-90.
Graham H. Bothamley, John Swanson Beck, Robert C. Potts, John M. Grange, Thomas Kardjito, and J uraj Ivanyi
Medical Research Council Tuberculosis and Related Infections Unit. Royal Postgraduate Medical School. and Department ofMicrobiology. Royal Brompton Hospital. London. and Department of Pathology. Ninewells Hospital and Medical School. Dundee. United Kingdom; Department of Pulmonology. Airlangga University. Surabaya. Indonesia
Specific antibody levels and delayed-type hypersensitivity skin responses to antigens of Mycobacterium tuberculosis in 39 hospital staff who were heavily exposed to tuberculosis (TB) were compared with those in 36 factory employees from Indonesia. Antibody levels to the TB68 epitope of the 14-kDa antigen were significantly greater, while titers to the TB23 (19-kDa) and TB72 (38-kDa) epitopes and lipoarabinomannan (LAM) were lower in exposed than in nonexposed subjects (all P < .02). The intensity of tuberculin responses correlated positively with anti-LAM and negatively with anti-19-kDa antibody levels. Possible reasons for the selective humoral response of chronically exposed healthy subjects to the 14-kDa antigen, but not to other antigens immunogenic in patients with tuberculosis, are discussed. Despite recent advances in the immunologic study of mycobacterial antigens, little progress has been made in the allocation of either protective or pathogenic function to any antigenic constituent [I]. Patients with tuberculosis (TB) have abundant T cell-mediated and humoral immunity, but it is not known whether disease occurs because the immune repertoire fails in its effector properties or whether the speciReceived 14 November 1991; revised 27 January 1992. Informed consent was obtained from all patients and controls; ethical approval was obtained from the National Heart and Chest Hospitals and Surabaya University. Financial support: the Wellcome Trust. Reprints or correspondence (present address): Dr. Graham Botharnley, Department of Chest Medicine. University College Hospital. Gower St., London WCIE 6AU. UK. The Journal of Infectious Diseases 1992;166:182-6 © 1992 by The University of Chicago. All rights reserved. 0022-1899/92/6601-0030$01.00
ficity of its repertoire is associated with the pathogenic outcome. Definition of the antigens recognized by the immune system in healthy individuals would be of particular interest but is hampered by the complexity of immunogenic stimuli due to environmental or commensal mycobacteria, selfhealing exposure to tubercle bacilli in childhood, and Mycobacterium bovis-bacille Calmette-Guerin (BCG) vaccination. A special opportunity for study is represented by occupational exposure to TB, where adults remain healthy despite prolonged contact with infectious patients. Such exposure to TB has been associated with delayed-type hypersensitivity (DTH) to tuberculin and a greater risk of developing TB [2]. However, a high rate of tuberculin conversion overestimates the incidence ofTB [3]. An early (6-8 h) erythematous reaction to intradermal tuberculin in hospital staff exposed to patients with active pulmonary TB has been interpreted as
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
Specificity of Antibodies and Tuberculin Response after Occupational Exposure to Tuberculosis
1ID 1992; 166 (July)
Concise Communications
the expression of particularly strong DTH [4]. Moreover, a reciprocal relationship between the intensity of skin tuberculin reactions and IgM, IgG, and IgA antibodies to M. bovisBeG in hospital contacts has been reported [5]. This report describes differences between healthy exposed and nonexposed individuals in terms of antigenic specificity or levels of antibodies and their relationships to DTH.
Materials and Methods
Solid-phase antibody competition test. The method used was essentially that of Hewitt et al. [7]. Briefly, wells of polyvinyl flexible microtiter plates (M24; Dynatech) were coated with 50 ~l of a soluble extract of M. tuberculosis H37Rv at 30 JoLg/ml overnight at 4°e. Nonspecific binding was blocked with 3%bovine serum albumin (BSA) for 1 h at 20°e. Four dilutions of serum (1:5,1:25,1:125, and 1:625) were prepared in 3% BSA and 25 JoLI incubated in duplicate wells at 20°C for 4 h. Without washing, 25 JoLI of 125I-labeled MAb, diluted to give 1000-2000 cpm when incubated in wells without competing serum (the high control), was added to each well and plates incubated overnight at 4°e. After being washed repeatedly, wells were counted in a 1260 Multigamma counter (LKB Instruments, Rockville, MD). Antibody (inhibition) titers were calculated as the dilution of serum that gave 50% inhibition of binding of the labeled MAb. Statistical analysis. Antibody levels were compared using the Mann-Whitney U test. The diameters of tuberculin responses in the two groups were compared by Student's t test. Correlations were made using Spearman's rank correlation test.
Results Antibody levels to the TB68 epitope were significantly greater in hospital staffwhile antibody titers to the TB23 and TB72 epitopes and IgO binding to LAM were lower in hospital staff than in factory employees (table I). The difference Table 1. Serum antibody and delayed-type hypersensitivity (DTH) responses to tuberculin in hospital and factory workers.
Epitope. antigen
Hospital workers (n = 39)
Factory workers (n = 36)
p*
Inhibition antibody titer. median (geometric mean). range TB23 (19 kDa) TB68 (14 kDa) TB72 (38 kDa)
19.5 (19.6). 6-68 5 (5.3). 0-57 t 2 (1.7).0-7
26 (27.4). 11-76 2 (2.4). 0-15 4.5 (4.5). 0-13
.01 .02 .001
Antibody binding titer. median (geometric mean). range 19 kDa (IgG) 65 kDa (IgG) LAM (IgG)
0(2.3).0-30 20 (18.3). 0-237 9.5 (12.0). 0-448
0(3.4).0-125 41 (31.7).0-161 23 (30.8). 0-256
NS NS .015
No. of individuals Tuberculin DTH 20mm NOTE.
o
12
3
9
\I 10
27
o
NS. not significant; LAM. lipoarabinomannan.
* Mann-Whitney U test. t Greater proportion of patients with tuberculosis had antibody titers of >5 to TB68 epitope (79% vs. 43%). but only 8 of 118 patients had titers of >57 (data not shown).
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
Subjects. Thirty-nine staff of the William Booth Hospital (Surabaya, Indonesia) with occupational exposure to TB and 36 factory employees from the same socioeconomic group in that town with no such exposure were examined. Thirty-two hospital staff (82%) and 14 factory workers (39%) had received BCG vaccination as children. All subjects were residents of East Java, had normal chest radiographs, and were clinically well. Serum samples were obtained at the time of skin testing and were stored at - 20°e, transported to London on solid carbon dioxide (-40°C), and stored at -20°C until analyzed. DTH. Tuberculin PPD (purified protein derivative)-RT23 (10 IU) was injected intradermally on the volar aspect of the forearm. A ballpoint pen was used to define the edge of induration at 48 h; a ruler measured the diameter of induration, to the nearest millimeter, in the long axis of the forearm and at right angles, and the geometric mean of these two measurements was used as the tuberculin response. Monoclonal antibodies (MAbs). Three MAbs, TB23 (19-kDa antigen), TB68 (14-kDa antigen), and TB72 (38-kDa antigen), were used to prepare purified antigens and as reagents in the antibody competition assay (see below). These MAbs had been evaluated for their binding to extracts from different mycobacterial species and relative molecular mass of the respective antigens by independent laboratories in a World Health Organization workshop: TB68 and TB72 showed binding to Mycobacterium tuberculosis complex alone, while TB23 showed binding to M. tuberculosis. Mycobacterium kansasii, Mycobacterium marinum, and Mycobacterium duvalii [I]. Antigens. The I 9-kDa antigen was purified by affinity chromatography using MAb TB23 as previously described [6]. The 65-kDa antigen is a recombinant protein and lipoarabinomannan (LAM) is a glycolipid common to mycobacteria. ELISA. Microtiter plates (M 129B; Dynatech, Billingshurst, UK) were coated with 0.2 ~g/ml 19-kDa antigen, 1 ~g/ml 65kDa antigen, 0.1 ~g/ml LAM, or PBS overnight at 4°e. Nonspecific binding was blocked with PBS containing 0.05% Tween 20 (Sigma, Poole, UK) and 1% skimmed milk (Sainsbury, London; PBSTM). Four serum dilutions (1:5, 1:25, 1:125, and I:625) in PBSTM were incubated for I h at 37°C in duplicate wells and, after washing, with goat anti-human IgG peroxidase conjugate. Bound peroxidase activity was measured using O. I mg/ml tetramethylbenzidine in 0.1 M citrate buffer, pH 5.2, containing 0.01% hydrogen peroxide for 10 min. The reaction was stopped by the addition of0.5 M sulfuric acid, and optical densities were measured at 450 nm. The antibody binding titer was calculated as the dilution of serum giving 30% of the plateau value of a standard hyperimmune serum.
183
184
Concise Communications
between hospital and factory staff in TB68 titers was also clearly demonstrated when groups with and without BCG vaccination were compared separately (figure IA). The decrease in TB23, TB72, and LAM antibodies remained significant when only BCG-vaccinated hospital staff and factory employees were compared. However, there were no significant differences in any of the antimycobacterial antibody levels between BCG-vaccinated and non vaccinated factory employees. In BCG-vaccinated factory workers, IgG antibody levels to the 19-kDa antigen showed a significant correlation with anti-LAM IgG (r = .70, P < .05) but not with titers to the TB23 epitope. The diameters of the tuberculin skin reaction were much greater in hospital staff than in factory workers (table I). A tuberculin reaction of induration ~ 15 mm in diameter was found in 3 hospital staff and in 23 factory employees (70%; X2 = 8.7, P < .0 I). However, no significant difference in tuberculin reactions between BCG-vaccinated and nonvaccinated factory employees was observed. Three vaccinated factory employees did not respond to tuberculin.
BCG+
Discussion This study demonstrated significant associations of antibody levels to distinct antigens of M. tuberculosis and skin DTH responses with occupational exposure to TB. Antibody titers to the TB68 (14-kDa) epitope were higher in hospital staff than in factory workers. Selectively raised antibody levels to either the 14-kDa antigen or its TB68 epitope have previously been described in sera of healthy subjects after occupational exposure to TB [8], in family contacts of patients with smear-positive pulmonary TB [6], in children
BCG15
20
o ~
fI)
o
Factory (n=14) Hospital (n=30)
f:J
t5 Q) E ~
The intensity of the DTH response to tuberculin was positively correlated only with antibody levels to LAM in BCGvaccinated hospital staff (r = .43, P = .00 I; figure I B), but there was no relationship between anti- TB68 antibody levels and the diameter ofthe tuberculin response (r = -.05). Moreover, a significant negative correlation was found between DTH reactions and anti-19-kDa antibody levels in factory employees (r = -.51, P = .02; figure IB).
Factory (n=22) Hospital (n=7)
10 10
'0 c:i
z
-
-g E
>-
"'C
2.0
.8
.0
as
E as
1.5
as
~
~
~
1.0
g>
0.5
E as
1.5 1.0
0) T""' I
~
0.5
C)
o
....J
...J
30
9
Figure 1. Antibody levels and delayedtype hypersensitivity (DTH) responses in hospital workers frequently exposed to patients with tuberculosis and in factory employees. A, Antibody titers to TB68 epitope of 14kDa antigen in bacille Calmette-Guerin (BCG)-vaccinated (BCG+) and nonvaccinated (BCG-) groups. B, Correlations ofantiLAM (lipoarabinomannan) in hospital staff and anti-19-kDa IgG antibody titers in factory employees with DTH responses to tuberculin.
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
A
JID 1992; 166 (July)
JlD 1992; 166 (July)
Concise Communications
of the immunomodulatory activity ofLAM permitted greater T cell reactivity and hence larger tuberculin responses. This suggestion agrees with the finding of a reduced anti-LAM response in children with disseminated rather than localized TB [15], since disseminated TB has generally been associated with diminished DTH reactions. Alternatively, the positive correlation between anti-LAM titers and DTH to PPD might merely indicate the overall strength of the immune response in individuals exposed to tuberculosis. Acknowledgements
We thank J. D. A. van Embden (National Institute of Public Health and Environmental Hygiene, Bilthoven, Netherlands) for the purified 65-kDa antigen, produced with support from the UNDP/World Bank/WHO-TDR special program, and P. J. Brennan (Colorado State University, Fort Collins) for purified LAM. References I. Ivanyi J, Sharp K, Jackett PS, Bothamley G. Immunological study of
2.
3.
4.
5.
6.
7.
8.
9.
10.
II. 12.
13.
the defined constituents of mycobacteria. Springer Semin ImmunopathoI1988;10:279-300. Sutherland J. Recent studies in the epidemiology of tuberculosis, based on the risk of being infected with tubercle bacilli. Adv Tuberc Res 1976;19:1-73. de March-Ayuela P. Choosing an appropriate criterion for true or false conversion in tuberculin testing. Am Rev Respir Dis 1990; 141:81520. Gibbs JH, Grange JM, Swanson Beck J, et al. Early delayed-type responses in tuberculin skin tests after heavy occupational exposure to tuberculosis. J C1in PathoI1991;44:919-23. Pitchappan M, Brahmajothi V, Rajaram K, Subramanyam PT, Balakrishnan K, Muthuveeralakshmi P. Immunological spectrum for BCG in hospital contacts from South India. Tubercle 1991;72: 133-9. Jackett PS, Bothamley GH, Batra HV, Mistry A, Young DB, Ivanyi J. Specificity of antibodies to immunodominant mycobacterial antigens in pulmonary tuberculosis. J C1in Microbiol 1988;26:2313-8. Bothamley G, Udani P, Rudd R, Festenstein F, Ivanyi J. Humoral responses to defined epitopes of tubercle bacilli in adult pulmonary and child tuberculosis. Eur J C1in Microbiol Infect Dis 1988;7:63945. Hoeppner VH, Jackett PS, Swanson Beck J, Kardjito T, Grange JM, Ivanyi J. Appraisal of the monoclonal antibody-based competition test for the serology of tuberculosis in Indonesia. Serodiagn Immunother 1987;1:69-77. Chandramuki A, Bothamley GH. Brennan PJ, Ivanyi J. Levels of antibody to defined antigens of Mycobacterium tuberculosis in tuberculous meningitis. J Clin Microbiol 1989; I 7:821-5. Bothamley GH, Rudd R, Festenstein F, Ivanyi J. Clinical value of the measurement ofMycobacteriullltuberculosis-specific antibody in pulmonary tuberculosis. Thorax 1992;47 (in press). Young DB, Garbe TR. Lipoprotein antigens of Mycobacterium tuberculosis. Res Microbiol 1991;142:55-66. Verbon A, Harterskeerl RA, Schuitema A, Kolk AHJ, Young DB, Lathigra R. The 14K antigen of Mycobacterium tuberculosis is related to the alpha-crystallin family of low molecular weight heat shock proteins. J Bacteriol 1992; 174: 1352-9. Havlir DV, Wallis RS, Boom WH, Daniel TM, Chervenak K, Ellner JJ. Human immune response to Mycobacterium tuberculosis antigens. Infect Immun 1991 ;59:665-70.
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
with tuberculous mediastinal lymphadenitis [7], and in the cerebrospinal fluid of patients with tuberculous meningitis [9]. In contrast, low anti-14-kDa antibody levels have been associated with extensive disease and a poor prognosis [10]. All of these data agree with the hypothesis that the 14-kDa antigen is immunogenic in the early stages of infection with M. tuberculosis. The fact that antibody levels to other mycobacterial antigens have not been found in primary or selfhealing infections [6, 7] further emphasizes the importance of the 14-kDa antigen in the early humoral immune response to M. tuberculosis. The 14-kDa antigen can be recovered from disrupted tubercle bacilli but is not readily found secreted into culture supernatants [11]. Perhaps intracellular antigens are the main targets for the immune response when the host is successfully killing mycobacteria, but once a parasitic relationship has been established, secreted antigens, associated with bacterial growth, assume a greater importance [6, 11). The 14-kDa antigen has been sequenced and found to be structurally homologous with proteins belonging to the family of low-molecular-weight heat-shock proteins [12]; heat-shock proteins can be powerful immunogens [1] and may be induced early in the bacterial colonization of host macrophages to facilitate the intracellular survival of M. tuberculosis. MAb TB68 binds to one of the three B cell epitopes, defined by murine MAbs, of the 14-kDa antigen [12]; there appears to be no human antibody response to one (IT-lor F23-49, unpublished data) of the remaining two epitopes, and the humoral response to the third epitope (defined by MAbs F67-8 and F67-16 [12]) has not been examined. Antibody to a recombinant preparation of the 14-kDa antigen, but not to its TB68 epitope, has been found in tuberculinnegative controls [13] (unpublished thesis). Differences in antibody responses to the TB68 and F67-8/16 epitopes between individuals exposed to tubercle bacilli who remain healthy or who develop tuberculosis might therefore warrant further investigation. Anti-19-kDa antibody levels were inversely related to the size of the tuberculin DTH response. A reciprocal relationship between antibodies to whole BCG and DTH reactions to PPD has been recently described [5] and may correspond to priming ofdifferent CD4 + T lymphocyte subsets [14]. The tuberculin responses in factory workers may have been due to exposure to environmental mycobacteria; the known cross-reactivity of the anti-19-kDa MAb, TB23, that binds to extracts from both M. tuberculosis and M. kansasii [1] would be in keeping with this suggestion. In contrast, anti-LAM antibody and DTH were positively correlated. In vitro, LAM has been shown to inhibit antigen processing. trigger the release of tumor necrosis factor from macrophages, inhibit the interferon-v-rnediated activation of macrophages, and cause local vascular damage [15]. Thus, the positive correlation between antibody levels to LAM and tuberculin responsiveness might suggest that neutralization
185
186
Concise Communications
JIO 1992: 166 (July)
14. Leclerq SA, Bretscher PA. T cells expressing delayed-type hypersensitiv-
15. Costello AM. Kumar A. Narayan V. et al. Does antibody to mycobacte-
ity can be derived from a humorally immune lymphocyte population. Eur J Immunol 1987; 17:949-54.
rial antigens. including lipoarabinornannan, limit dissemination in childhood tuberculosis? Trans R Soc Trop Med Hyg 1992 (in press).
Secreted Antigens of Mycobacterium tuberculosis: Characterization with T Lymphocytes from Patients and Contacts after Two-Dimensional Separation Sabine Daugelat, Heinz Guile,* Bernd Schoel, and Stefan H. E. Kaufmann
Department of Immunology. University of Ulm, Germany
Worldwide, >60 million people suffer from tuberculosis, and it is estimated that the disease causes 3 million deaths annually. About 30%-50% of the total world population have primary infections with the etiologic agent, Mycobacterium tuberculosis. Considering the increasing prevalence of tuberculosis in AIDS patients, its international significance is becoming even more acute. Tuberculosis can be treated chemotherapeutically; yet recently, several outbreaks of drug-resistant strains occurred in the United States. Effective control will be achieved best with appropriate vaccines. Because vaccination trials with Mycobacterium bovis BCG have given contradictory results, currently various attempts are made to develop a subunit vaccine composed of defined mycobacterial antigens in a suitable vector system (reviewed in [1D· M. tuberculosis lives and replicates inside mononuclear phagocytes, and immunity crucially depends on T lymphocytes. Thus, potential vaccine antigens must be characterReceived 29 July 1991; revised 18 February 1992. The appropriate guidelines for human experimentation of the Federal Republic of Germany were followed in the conduct of this research. Grant support: German Leprosy Relief Association (to S.H.E.K.); Graduate College of Biomolecular Medicine (to S.D.). Reprints or correspondence: Dr. S. H. E. Kaufmann. Department of Immunology. University of'Ulm, Albert-Einstein-Allee 11.0-7900 Ulm, Germany. * Present affiliation: Department of Veterinary Pathology and Public Health. Massey University. Palmerston North. New Zealand. The Journal of Infectious Diseases 1992;166:186-90 © 1992 by The University of Chicago. All rights reserved. 0022-1899/92/6601-0031 $0 1.00
ized at the T cell level. So far, the potential of bacterial cell wall and cytoplasmic antigens to stimulate T lymphocytes has been studied [I]. Processing and presentation to T cells of these antigens should depend on prior death and degradation of the microorganisms. However, at least at early stages of infection, such antigens would be available only to a limited degree. In contrast, proteins secreted by bacteria persisting inside macrophages should be readily accessible to the antigen processing machinery. Hence, T cell reactivity to proteins secreted by actively growing bacteria could serve as a model for an early stage of infection when pathogens still live within phagosomes. Recently a method has been developed in our laboratory that allows direct T cell screening of hundreds ofdifferent protein fractions after two-dimensional separation [2]. This technique was used to investigate the repertoire of T cells from tuberculosis patients and healthy contacts toward secreted antigens of M. tuberculosis.
Materials and Methods Preparation of antigens. M. tuberculosis H37Rv organisms were grown in Dubos broth base (Difco, Detroit) supplemented with 10% Dubos medium albumin (Difco) until mid-log phase. Aliquots of 5 ml containing 4.8 X 108 cfu/ml were stored at -70°C. For antigen preparations, aliquots were thawed. washed, inoculated in 500 ml of zinc-free Sauton medium [3]. and incubated at 3rC at 100 rpm for 3 weeks. Cultivation in this medium is known to give a particularly high yield of secreted proteins [4]. Bacteria were sedimented, and supernatant was filtered twice through 0.22-JLm filter units (Schleicher &
Downloaded from http://jid.oxfordjournals.org/ by guest on August 11, 2015
Little is known about T cell antigens involved in immunity against Mycobacterium tuberculosis. Most model systems use in vitro culture of human T lymphocytes with bacterial Iysates or secreted proteins as antigens. In this study, proteins from 3-week-old M. tuberculosis culture filtrates were separated by two-dimensional PAGE and subsequently transferred into soluble phase. The resulting 480 fractions were screened with T lymphocytes from tuberculosis patients and healthy contacts. T cells from all 9 patients and from 8 of 10 tuberculin-positive contacts preferentially responded to a cluster of acidic proteins (pl 4-5) with molecular masses of 30-1 00 kDa, although they also recognized a number of other fractions. In contrast, of 7 tuberculin-negative contacts, 4 were not and 3 were only weakly stimulated by this cluster region. Therefore, this distinct cluster of secreted proteins seems to comprise dominant T cell antigens of M. tuberculosis.
