Tuberculin Skin Testing and In Vitro T Cell Responses to ESAT-6 and Culture Filtrate Protein 10 after Infection with Mycobacterium marinum

Transcription

1 1797 Tuberculin Skin Testing and In Vitro T Cell Responses to ESAT-6 and Culture Filtrate Protein 10 after Infection with Mycobacterium marinum or M. kansasii Sandra M. Arend, 1 Krista E. van Meijgaarden, 2 Kirsten de Boer, 1 Elisabeth Cerdá de Palou, 3 Dick van Soolingen, 4 Tom H. M. Ottenhoff, 2 and Jaap T. van Dissel 1 Departments of 1 Infectious Diseases and 2 Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, and 3 Regional Health Service (Gemeentelijke Gezondheids Dienst), Geleen, and 4 Department of Mycobacteria, National Institute of Public Health and the Environment (Rijks Instituut voor Volksgezondheid en Milieu), Bilthoven, The Netherlands T cell responses to ESAT-6 and culture filtrate protein 10 (CFP-10), antigens expressed by Mycobacterium tuberculosis but not by M. bovis bacille Calmette-Guérin (BCG), were found to discriminate reliably between infection with M. tuberculosis and BCG vaccination. Because the esat-6 and cfp-10 genes occur in M. kansasii and M. marinum, T cell responses to ESAT- 6 and CFP-10 were investigated in patients infected with M. kansasii or M. marinum, persons intensively exposed to environmental mycobacteria, and unexposed control subjects. Tuberculin skin tests were performed, and peripheral blood mononuclear cells were cocultured with ESAT-6, CFP-10, peptide mixtures of ESAT-6 and CFP-10, and control antigens. When enzyme-linked immunosorbent assay (ELISA) and enzyme-linked immunospot assay (ELI- SPOT) were used to measure interferon-g production, most M. kansasii orm. marinum infected patients and several persons exposed to environmental mycobacteria were found to respond to ESAT-6 and/or CFP-10. ELISA and ELISPOT yielded comparable results, as did whole antigen and peptides ( P!.0001). These results may be relevant for the development of novel assays for diagnosis of tuberculosis. The tuberculin skin test (TST) is characterized by a high frequency of false-positive test results among persons who have been vaccinated with Mycobacterium bovis bacille Calmette- Guérin (BCG) or exposed to environmental mycobacteria, because antigens are present in purified protein derivative (PPD) that are shared by M. tuberculosis, BCG, and (usually apathogenic) environmental mycobacteria [1, 2]. BCG vaccination of children is common practice in most of the world, and currently more than one-half of all newly detected cases of tuberculosis (TB) in many industrialized countries occur in immigrants originating from regions where TB is highly endemic and BCG vaccination is routinely used. With the aim of developing a more specific immunodiagnostic test for detection of active or latent TB infection, recent research has been focused on antigens that occur exclusively in M. tuberculosis and pathogenic Received 17 June 2002; revised 19 August 2002; electronically published 19 November All subjects gave informed consent for blood sampling and tuberculin skin testing after written information was provided. The study protocol was approved by the Medical Ethics Committee of the Leiden UniversityMedical Center (protocol P136/97). Financial support: Netherlands Leprosy Foundation, Netherlands Organization for Scientific Research (Nederlandse Organisatie voor Wetenschappelijk Onderzoek); European Commission. Reprints or correspondence: Dr. Sandra M. Arend, Dept. of Infectious Diseases, C5P, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands (s.m.arend@lumc.nl). The Journal of Infectious Diseases 2002;186: by the Infectious Diseases Society of America. All rights reserved /2002/ $15.00 M. bovis strains. Use of comparative genomic methods led to the identification of a region of difference, named RD1, that was found to be present in all M. tuberculosis and M. bovis strains but was lacking in all BCG strains and most environmental mycobacteria [3 5]. ESAT-6 [6, 7] and culture filtrate protein 10 (CFP-10) [8] were identified as potent RD1-encoded T cell antigens that induce interferon (IFN) g production in mice [9, 10], cattle [11], and humans [12 17] infected with M. tuberculosis or pathogenic M. bovis but not those vaccinated with BCG. Such responses were observed both in patients with clinically active TB and in healthy but infected TST-positive individuals [18, 19]. RD1-encoded antigens are, therefore, currently considered to be promising candidate antigens for inclusion in a novel immunodiagnostic assay [20]. Although RD1 is absent from most environmental mycobacteria, the genes coding for ESAT-6 and CFP-10 were demonstrated, by polymerase chain reaction (PCR) and/or Southern blot hybridization using DNA from single isolates per species (mainly American Type Culture Collection strains), to be present in M. kansasii, M. marinum, M. szulgai, M. flavescens, and M. gastrii [6, 11, 21 23]. Moreover, an ESAT-6 homologue has recently been identified in the genome of M. leprae [23]. M. kansasii and M. marinum are known to cause characteristic diseases in immunocompetent persons, although this occurs infrequently. This raises the question of whether infection with or even exposure to M. kansasii or M. marinum can induce T cell responses to ESAT-6 and CFP-10; this could have important implications for the ongoing development of novel

2 1798 Arend et al. JID 2002;186 (15 December) diagnostic tests with improved specificity for detection of infection with M. tuberculosis. Thus far, no studies have been published that address T cell responses to RD1-encoded antigens in persons infected with M. kansasii or M. marinum. Apart from the intrinsic specificity of novel test antigens such as ESAT-6 and CFP-10, it is to be expected that technical characteristics of an assay will affect test performance. In initial studies of in vitro T cell responses to M. tuberculosis specific antigens, the concentration of IFN-g in supernatants of cell cultures was used to measure T cell activation, usually after 5 or 6 days of culture. More recent studies used the ELISPOT technique, by which IFN-g producing cells in stimulated cultures were enumerated after a period of h of antigenic stimulation. The results of several studies suggested that ELI- SPOT may be more sensitive than IFN-g detection by ELISA [24] or proliferative responses [25]. Enumeration of ESAT-6 responsive T cells from recently exposed contacts of contagious TB patients was highly accurate for detection of latent infection, and the findings even suggested that this measure may be superior to the classic TST [26]. In a direct comparison, ELISPOT was times more sensitive for detection of interleukin (IL) 2, IL-4, IL-5, and IL-6 production by murine T cells than was ELISA performed on culture supernatants [27], but a high correlation between both techniques was found for detection of IFN-g production by cultured mouse splenocytes after a nonspecific stimulus [28]. To our knowledge, no direct comparison of the ELISA and ELISPOT techniques for detection of IFN-g production by human blood cells in response to antigens of M. tuberculosis has been published as yet. Recombinant protein antigens were used in some studies, and single peptides or mixtures of overlapping peptides were used in others. We previously demonstrated, in a direct comparison using the ELISA technique, that ESAT-6 and CFP-10 protein antigens and mixtures of overlapping peptides are antigenically equivalent [29], but this finding remains to be evaluated for ELISPOT. The primary aim of the present study was to investigate T cell responses to ESAT-6 and CFP-10 in individuals with culture-positive infection with M. marinum or M. kansasii and in persons who experience intensive exposure to environmental mycobacteria as a result of a profession or hobby. The secondary aim was to directly compare IFN-g production measured by ELISA at day 6 of cell cultures with the number of IFN-g producing cells detected with ELISPOT after 20 h of culture, using recombinant whole-protein antigens and peptide mixtures of ESAT-6 and CFP-10 for both techniques. Subjects and Methods Study subjects. Study subjects were recruited at the Leiden University Medical Center in Leiden and at the Regional Health Service in Geleen, The Netherlands. We included patients for whom M. marinum had been cultured from a specimen from a biopsy of a typical lesion on the hand or forearm. To discriminate between infection and colonization, M. kansasii infection was defined according to the criteria used by the American Thoracic Society (ATS) [30]. In brief, these consist of characteristic radiographic abnormalities in combination with a positive culture from a bronchoscopic washing or lung biopsy specimen or at least 2 positive sputum cultures, in the absence of an alternative diagnosis. No reference standard or parameter exists by which exposure to environmental mycobacteria can be quantified. However, we reasoned that such exposure would be most pronounced in individuals who had repeated and intensive contact with the natural world (e.g., soil, plants, animals, and fish tanks) as a result of a profession or hobby. We chose owners of tropical fish tanks (recruited via the local fish-tank society), veterinarians (from a rural practice), and professional flower growers (from flower farms located within a few miles distance of each other) to represent these groups. Control subjects were TB patients who had high levels of response to the test antigens in a previous study (to ensure that a full range of responses was obtained for the analysis of the correlation between responses on ELISA and ELISPOT), patients with documented infection with nontuberculous mycobacteria other than M. marinum and M. kansasii, and healthy TST-negative persons who did not belong to any of the groups mentioned above and had never been exposed to a patient with active pulmonary TB or been included in a contact investigation. All study subjects answered a questionnaire about BCG vaccination, travel history, contact with patients known to have TB or risk groups for such infection, and duration of possible environmental exposure to mycobacteria. None of the study subjects used or had used immunosuppressive treatment or belonged to a risk group for human immunodeficiency virus infection. Tuberculin skin testing. TSTs were performed according to the guidelines issued by the Royal Netherlands Tuberculosis Association, using 2 TU of tuberculin (type RT23, batch or 14211; Statens Serum Institute); as is the convention, the induration of the skin in millimeters was measured after h. An induration of 10 mm was regarded as a positive result. All skin testing and test reading was performed by experienced personnel, except in the cases of 2 patients infected with M. marinum, whose TSTs were read by a general practitioner, and 1 fish-tank owner, who read the TST himself after instruction. Three tubes of heparinized venous blood (27 ml total) were drawn at the time of the TST. Antigens. As complex antigens (i.e., preparations containing multiple antigens, including those present in mycobacteria other than M. tuberculosis, and thus expected to give cross-reactive responses), we used PPD (RT49; Statens Serum Institute); M. tuberculosis H37Rv sonicate (provided by the National Institute of Public Health and the Environment, Bilthoven, The Netherlands); short-term culture filtrate (ST-CF), the production of which has been described elsewhere [31]; and M. avium sonicate, which was produced by Dr. A. Kolk (Royal Tropical Institute, Amsterdam). Recombinant ESAT-6 (batch F800) and CFP-10 (batch 00-02) were expressed in Escherichia coli, as described elsewhere [6], and made available by the Statens Serum Institute, Copenhagen, Denmark. ESAT-6 and CFP-10 were also tested as mixtures of 9 overlapping 20-mer peptides per antigen. TB10.4 and Ag85B were included as control antigens, because they are immunodominant antigens that are present in most pathogenic and environmental mycobacterial

3 JID 2002;186 (15 December) RD1 Responses after Mycobacterial Infection 1799 species, including M. marinum and M. kansasii [21, 32]. We hypothesized that T cell responses to these antigens would not differentiate between the different study groups. TB10.4 (Rv0288) was tested as a mixture of 9 overlapping 18-mer peptides; recombinant Ag85B was produced at the Leiden University Medical Center. The amino acid sequences of the peptides used in this study have been described elsewhere [18, 29]. Lymphocyte stimulation assay. Peripheral blood mononuclear cells (PBMC) were isolated from heparinized venous blood by ficoll-hypaque density-gradient centrifugation. Cells were frozen in RPMI 1640 medium (Gibco Life Technologies) supplemented with 2mM glutamine, 20% fetal calf serum, and 10% dimethylsulfoxide. Cells were stored at 70 C until thawing. For experiments, PBMC 5 ( cells/well) were incubated in round-bottom microtiter wells in the presence or absence of antigen in 200 ml of Iscove s modified Dulbecco s medium (IMDM; Gibco), supplemented with 10% pooled human AB serum, 40 U/mL penicillin, and 40 mg/ml streptomycin (referred to as complete medium ) in triplicate at 37 C in humidified air containing 5% CO 2. The final concentrations of the antigens were 1 mg/ml for PPD, M. tuberculosis sonicate, ST-CF, and M. avium sonicate; 1 and 10 mg/ml for ESAT-6, ESAT- 6 peptide mixture, CFP-10, CFP-10 peptide mixture, and Ag85B; and 2 and 20 mg/ml for TB10.4 peptide mixture. Supernatants for IFN-g determinations were collected at day 6 (50 ml/well), pooled for each triplicate experiment, and tested on the same day. IFN-g was measured with a standard ELISA technique (U-CyTech). The detection limit of the assay was 20 pg/ml IFN-g. When 2 antigen concentrations were tested, the highest IFN-g response was used for analysis. As a rule, the highest of 2 antigen concentrations resulted in the highest response; however, in 2 patients with TB, responses to ESAT-6 at 1 mg/ml exceeded those at 10 mg/ml. IFNg values in unstimulated cultures typically were undetectable. Ex vivo ELISPOT assay for single-cell IFN-g release. Cultures were performed in 96-well polyvinylidene difluoride (PVDF) backed plates (MAIPS45; Millipore) precoated with 5 mg/ml anti IFN-g monoclonal antibody (MAb) 1-D1K (Mabtech), washed 6 times with IMDM, and blocked (for 2 h) with IMDM 10% fetal 5 calf serum. PBMC ( 3 10 cells/well) were incubated in complete medium in a final volume of 200 ml/well in the PVDF-backed 96- well plates at 37 C in the presence of 5% CO 2. Final antigen concentrations were 5 mg/ml for PPD and 10 mg/ml for ESAT-6 and CFP-10. A mixture of all 9 overlapping peptides of ESAT-6 and all 9 peptides of CFP-10 was used at a total concentration of 20 mg/ml. Phytohemagglutinin at 5 mg/ml was used as a positive control, and complete medium without antigen was added to the negative-control wells. All antigens were tested in quadruplicate. After 20 h, plates were washed (PBS containing 0.05% Tween 20), incubated with 100 ml of biotinylated anti IFN-g MAb (0.3 mg/ ml) for 3 h at room temperature, washed, incubated with streptavidin alkaline phosphatase conjugate (1:1000; Mabtech) for 2 h, washed again, and incubated with 100 ml of nitroblue tetrazolium 5-bromo-4-chloro-3-indolyl phosphate substrate (Sigma). The reaction was stopped by addition of water. Plates were dried and analyzed on a Zeiss Axioplan 2 microscope, using KS ELISPOT software (CarlZeiss Vision). Positive responses were always obvious to the naked eye as well. For analysis, the mean number of spot-forming units (sfu) per well from quadruplicate values for each antigen was calculated and transformed to number of sfu per 10 6 PBMC. To relate the number of sfu in the presence of antigen to the background value in unstimulated wells, which may vary between individuals, the ELI- SPOT result was expressed as mean sfu antigen (mean sfumedium [3 SEM medium]). A positive result was predefined as a result 10, cor- responding to a mean sfu antigen that exceeded the mean sfu medium (3 SEM medium). Statistical analysis. The analysis of correlations between results of different tests was performed nonparametrically with Spearman s test. Comparisons between groups were analyzed with the Kruskal-Wallis or Mann-Whitney U test, as appropriate. All statistical analyses were 2-sided; P!.05 was considered to be statistically significant. Results Characteristics of the study subjects. In all, 42 study subject were included: 7 patients infected with M. marinum, 7 owners of tropical fish tanks, 6 veterinarians, 5 professional flower growers, 5 patients infected with M. kansasii (recruited at the Regional Health Service, Geleen), 4 patients with TB, 2 patients infected with nontuberculous mycobacteria (1 patient infected with M. avium and 1 infected with M. abscessus), and 6 healthy TST-negative individuals. Characteristics of the study subjects, including BCG vaccination state, are listed in table 1. The median duration of contact with tropical fish tanks was 10.8 years (range, years) among patients infected with M. marinum, compared with 18.7 years (range, 2 40 years) among fish-tank owners without such infection (control). Interestingly, 1 patient infected with M. marinum had never had contact with tropical fish tanks. He was a professional vegetable grower whose greenhouse was sprinkler-irrigated from a water supply system fed by a rainwater basin, and this was assumed to be the risk factor for the infection. None of the owners of tropical fish tanks, the veterinarians, or the professional flower growers reported the use of gloves during their work or cleaning of the fish tanks. None of them had been vaccinated with BCG or had ever traveled to regions in which TB is highly endemic, and none had ever received a diagnosis of active TB or been included in a contact investigation. They were from small-town or rural residences and did not have contact with patients with TB or persons belonging to risk groups for TB, such as prison inmates or homeless persons. TST results. TST results are included in table 2. A positive TST result was observed in most patients infected with M. marinum and in one-half of those infected with M. kansasii, whereas results were negative in most persons with intensive exposure to environmental mycobacteria. The wide range of observed TST results allowed us to analyze the correlation between individual TST results and in vitro IFN-g production in response to various antigens. The highest correlation existed between TST results and IFN-g production in response to PPD ( r p 0.64; P!.0001), but the correlations between TST results and IFN-g responses to several other antigens were significant

4 1800 Arend et al. JID 2002;186 (15 December) Table 1. Characteristics of subjects included in a study of in vitro T cell responses to ESAT-6 and culture filtrate protein 10 after Mycobacterium infection. Subject group No. of subjects Sex, no. of men/ no. of women Mean age, years (range) Vaccinated with BCG Exposed to TB a Years of diagnosis M. marinum infected patients 7 6/ ( ) 1/ , 1998, 2000, 2000, 2000, 2001, 2001 M. kansasii infected patients 5 4/ ( ) 0/ , 1993, 1995, 1997, 2000 NTM-infected patients b 2 0/ ( ) 0/ , 2000 Patients with TB 4 2/ ( ) 1/ , 1993, 1998, 1999 Median duration of exposure, years (range) Aquarium owners 7 5/ ( ) 0/ (2 40) Veterinarians 6 5/ ( ) 0/6 1 c 9.5 (3 45) Flower growers 5 5/ ( ) 0/ (15 44) TST-negative control subjects 6 0/ ( ) 0/6 0 NA NOTE. BCG, bacille Calmette-Guérin; NA, not applicable; NTM, nontuberculous mycobacteria; TB, tuberculosis; TST, tuberculin skin test. a Received a diagnosis of active TB, was exposed to a patient with pulmonary TB, or was included in a contact investigation. b This group includes 1 patient with disseminated M. avium osteomyelitis (a partial interferon-g receptor 1 defect was found, as described elsewhere [33]), and 1 patient with M. abscessus lung infection (no immune disorder was identified). c The oldest person in this group had been infected in his youth, with documented TST positivity but without clinical signs or symptoms. as well (for ST-CF, r p 0.47 and P p.005; for M. tuberculosis sonicate, r p 0.4 and P p.02; for ESAT-6, r p 0.49 and P p.003; and for CFP-10, r p 0.48 and P p.004). Individual test results are summarized in table 2. IFN-g responses in 6-day cell cultures (ELISA). Next, individual IFN-g responses to PPD were compared with IFN-g responses to the other antigens. Individual IFN-g responses to PPD significantly correlated with responses to each of the other antigens tested ( P p.0005 for TB10.4 peptide mixture; P!.0001 for all other comparisons). Responses to M. tuberculosis sonicate, ST-CF, M. avium sonicate, and Ag85B were generally higher than the response to PPD, whereas responses to ESAT- 6, CFP-10, and TB10.4 peptide mixture were generally lower (figure 1). IFN-g responses to recombinant ESAT-6 and CFP- 10 were highly correlated with responses to the corresponding peptide mixtures ( P!.0001). However, the responses of several patients with TB to whole CFP-10 antigen exceeded responses to the peptide mixture in that group of subjects (table 2), indicating that differences occurred at the individual level. A comparison of IFN-g responses to PPD, ESAT-6, CFP- 10, and Ag85B by subject group is shown in figure 2. Detectable IFN-g responses (i.e., IFN-g levels 120 pg/ml) to ESAT-6 and CFP-10 were observed in 5 and 6, respectively, of the 7 subjects who were infected with M. marinum and in 4 and 2, respectively, of the 5 subjects who were infected with M. kansasii. Among the 7 long-term owners of tropical fish tanks, detectable responses to ESAT-6 and CFP-10 were found in 5 and 2 subjects, respectively. Somewhat unexpectedly, such responses were also observed in 7 and 5, respectively, of 11 veterinarians and flower growers, whereas TST results were mostly negative in these groups; in roughly one-half of those with detectable responses, the response exceeded 60 pg/ml IFN-g. T cell responses to Ag85B were found in most study subjects (table 2 and figure 2) and did not differ between groups, which is in keeping with the widespread distribution of Ag85B among mycobacterial species. Responses to TB10.4 were observed less frequently and were generally lower than responses to Ag85B but, similarly, did not differ between study groups (table 2). Enumeration of IFN-g producing cells by ELISPOT. The median number of sfu in unstimulated cultures was 3 sfu/well (range, 0 24 sfu; 25th and 75th percentile, 1 and 8 sfu, respectively), corresponding to 10 sfu/10 6 PBMC. The results of the ELISPOT assay were analyzed according to a predefined individual cutoff value for a positive response, as described in Subjects and Methods. Results obtained using the formula mean sfuantigen sfu medium were not different from those obtained using mean areaantigen area medium for the analysis (data not shown). The ELISPOT results correlated well with the corresponding IFN-g concentration obtained with ELISA (for PPD, r p 0.61 and P!.0001; for ESAT-6, r p 0.40 and P p.01; and for CFP-10, r p 0.59 and P!.0001). A mixture of peptides of both ESAT-6 and CFP-10 was used in the ELISPOT assay, and mixtures of overlapping peptides of either ESAT-6 or CFP-10 were tested separately in the 6-day cell culture. ELISPOT results with the ESAT-6/CFP-10 peptide cocktail correlated well with the IFN-g responses to the individual peptide mixtures (for the ESAT-6 peptide mixture, r p 0.49 and P p.002, and for the CFP-10 peptide mixture, r p 0.52 and P p.0006 ). ELISPOT results are depicted, by study group, in figure 3; the overall impression of the distribution of responses is similar to that observed for the IFN-g results obtained by ELISA shown in figure 2. Among the 18 subjects with intensive exposure to environmental mycobacteria, positive ELISPOT responses to CFP-10 and the ESAT-6/CFP-10 peptide mixture were less frequent than responses to ESAT-6 (responses were seen in 3, 3, and 11 subjects, respectively). This discrepancy was found only in these groups and could be related to differences between the amino acid sequence of ESAT-6 homologues of environmental

5 Table 2. Tuberculin skin test (TST) results, interferon (IFN) g production, and enumeration of IFN-g producing cells in response to Mycobacterium antigens. Group, subject TST result, mm of induration PPD MTB ST-CF ELISA result, pg of IFN-g/mL a M. avium ESAT-6 sonicate ESAT-6 pmix CFP-10 CFP-10 pmix ELISPOT result b TB10.4 pmix Ag85B PPD ESAT-6 CFP-10 RD1 pmix M. marinum infected patients UD UD UD UD UD UD ND UD ND M. kansasii infected patients UD UD 30 UD , UD UD UD UD UD UD ND UD UD UD UD UD NTM-infected patients UD 466 UD 899 UD UD UD UD UD Patients with tuberculosis , ND c Aquarium owners 19 0 UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD ND UD UD UD UD ND UD UD UD UD Veterinarians UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD UD Flower growers UD UD UD UD UD UD UD UD UD UD TST-negative control subjects 37 0 ND ND ND ND ND ND ND ND ND ND UD UD UD UD 93 UD UD UD UD 53 UD 278 UD UD UD UD UD UD ND ND ND ND 41 0 UD 255 UD 115 UD UD UD UD UD UD UD UD NOTE. CFP-10, culture filtrate protein 10; MTB, M. tuberculosis sonicate; ND, not done; pmix, mixture of all 9 overlapping peptides (see Subjects and Methods); NTM, nontuberculous mycobacteria; PPD, purified protein derivative; sfu, spot-forming units; ST-CF, short-term culture filtrate; UD, undetectable. a After 6 days of culture. b After 20 h of culture. ELISPOT results are expressed as mean sfuantigen (mean sfumedium [3 SEM medium]) per 106 peripheral blood mononuclear cells, which was chosen as the predefined measure of T cell activation. Thus, values 10 represent positive responses. c This patient had a very high no. of sfu in unstimulated cultures done for the ELISPOT assay, which resulted in negative ELISPOT results. However, production of high levels of IFN-g was detected by ELISA. The ELISPOT data were excluded from the analysis.

6 1802 Arend et al. JID 2002;186 (15 December) interaction between these proteins occurs and suggests that they have closely linked functions. Discussion Figure 1. Comparison between individual interferon (IFN) g responses to purified protein derivative (PPD) and IFN-g responses to Mycobacterium tuberculosis sonicate (MTB), short-term culture filtrate (ST-CF), and M. avium sonicate (A); ESAT-6 and culture filtrate protein 10 (CFP-10; B); and TB10.4 peptide mixture (pmix) and Ag85B (C). Data points in A lie predominantly in the area above the slanted dotted line representing X p Y, whereas those in B lie predominantly below that line. The data for Ag85B are distributed according to the pattern in A, and those for TB10.4 pmix according to the pattern in B. Horizontal and vertical dotted lines represent the detection limit of the IFN-g ELISA (20 pg/ml). mycobacterial species and that of the M. tuberculosis ESAT-6 derived peptides that were used in the assay; this could become relevant during the much shorter incubation period of the ELISPOT assay. Alternatively, competition between peptides for HLA binding or molecular interaction could have played a role. A recent study found that ESAT-6 and CFP-10 form a tight, 1:1 complex in vivo [34], which indicates that molecular The ESAT-6 and CFP-10 antigens are currently regarded as the most promising candidates for highly specific immunodiagnosis of infection with M. tuberculosis. However, both M. marinum and M. kansasii yielded positive responses of PCR and Southern blot analysis done using primers and probes, respectively, for esat-6 and cfp-10 of M. tuberculosis [6, 11, 21, 22]. In the present study, we investigated T cell responses to ESAT-6 and CFP-10 in persons infected with M. marinum or M. kansasii and in individuals with intensive exposure to environmental mycobacteria. Our results indicate that infection with M. marinum and infection with M. kansasii can result in detectable T cell responses to ESAT-6 and/or CFP-10; however, the proportion of responsive individuals is greater and the response levels are higher among those infected with M. marinum. In addition, this study is the first to directly compare the ELISA and ELISPOT techniques using ESAT-6 and CFP-10 antigens both as recombinant antigens and as a mixture of synthetic overlapping peptides spanning the complete sequence of these antigens. Results were consistent, regardless of whether IFN-g production after 6 days of cell culture or the number of IFN-g producing cells after 20 h of culture was used as the measure of T cell activation. Similarly, results obtained with whole recombinant protein antigen or a mixture of overlapping synthetic peptides correlated strongly, which confirm previous findings obtained with the ELISA technique [29] and expands findings obtained with ELISPOT. However, although results obtained with both techniques were highly correlated in our study, this does not necessarily imply that the 2 techniques are interchangeable. A comparison of the sensitivity and specificity of ELISA and ELISPOT techniques requires larger studies in well-defined epidemiological settings. Both at the bacterial level and clinically, a number of similarities exist between M. marinum or M. kansasii on the one hand and M. tuberculosis on the other. By analysis of cell-wall lipids, DNA-DNA hybridization, and the 16S rrna gene sequence, M. marinum was surprisingly similar to M. tuberculosis [35]. In contrast to most other environmental mycobacterial species, M. marinum and M. kansasii can cause significant clinical illness in persons without recognized immune defects. Fishtank granuloma is the classical clinical manifestation of infection with M. marinum in humans; it is characterized by a nodule that may develop into an ulcer at the inoculation site, usually on a finger or the dorsum of the hand, and nodular lymphangitis on the affected extremity that may result in draining ulcers along the lymphatic tract. As a rule, the disease remains limited to the affected extremity, which is explained by the preference of M. marinum for lower temperatures that are found only at

7 JID 2002;186 (15 December) RD1 Responses after Mycobacterial Infection 1803 Figure 2. Interferon (IFN) g production in response to purified protein derivative (PPD), ESAT-6, culture filtrate protein 10 (CFP-10), and Ag85B, by subject group, including patients with tuberculosis (TB), patients infected with Mycobacterium kansasii or M. marinum, aquarium owners, veterinarians, flower growers, and control subjects. IFN-g concentrations were measured in supernatants after 6 days of culture of peripheral blood mononuclear cells. The horizontal dotted line represents the detection limit of the IFN-g ELISA (20 pg/ml). the surface of the human body [36]. Thus, M. marinum causes systemic illness in reptilian species, whereas disseminated infection in humans has been described only sporadically and predominantly in persons with a severe defect in cellular immunity [37] or in association with massive exposure in a healthy person [38]. TST results were positive in all but 1 of the patients infected with M. marinum in our study and in 8 of 10 persons in a study published elsewhere [39], illustrating the lack of specificity of the TST for detection of infection with M. tuberculosis. Broad TST cross-reactivity exists between tuberculins of various mycobacterial species. In a recent study, cross-reactive TST responses to tuberculin of M. marinum were highly correlated with positive TST responses to PPD RT23 and with previous BCG vaccination [40]. However, M. marinum infection in immunocompetent persons is a distinct clinical entity and should, therefore, be of no concern in using a novel diagnostic assay based on ESAT-6 and CFP-10 for detection of infection with M. tuberculosis. The situation is different for infection with M. kansasii, which may have more-diverse clinical manifestations [41, 42]. Localized lung disease that is clinically indistinguishable from pulmonary TB can be observed in immunocompromised persons with impaired T cell function, including those infected with human immunodeficiency virus [43, 44], but also in otherwise healthy individuals. Localized infection of soft tissues or bone with M. kansasii is rarely found in immunocompetent persons, and, when it is found, it is often related to an operation or trauma [45]. Disseminated infection with M. kansasii occurs exclusively in persons who have a defect in cellular immunity. The management of the condition of an immunocompetent person with a cavitating pneumonia and sputum containing acid-fast bacilli includes isolation of the patient and investigation of contacts by skin testing, both of which have an important impact on the patient and his or her contacts. In the case of infection caused by M. kansasii, isolation of the patient and contact investigation are unnecessary. A test that can discriminate between M. tuberculosis and M. kansasii could prevent the unnecessary use of such measures. Our data demonstrate that neither the TST nor T cell responses to ESAT-6 and CFP-10 contribute to discrimination between TB and infection

8 1804 Arend et al. JID 2002;186 (15 December) Figure 3. Enumeration of interferon-g producing cells in response to purified protein derivative (PPD), ESAT-6, culture filtrate protein 10 (CFP-10), and RD1 peptide mixture (pmix), by subject group. ELISPOT results are expressed as mean sfuantigen (mean sfumedium [3 SEM per 10 6 medium]) peripheral blood mononuclear cells. The horizontal dotted line indicates the cutoff value, defined as mean sfumedium (3 SEM per 10 6 medium) peripheral blood mononuclear cells (see Subjects and Methods); values above this line represent positive responses. with M. kansasii. In this respect, more-reliable information could be obtained by PCR using samples obtained from the patient (sputum or lung biopsy specimens) and primers specific for the M. tuberculosis complex or M. kansasii. The relative contribution of M. kansasii to mycobacterial pneumonia varies by geographical location. One study in an area of low TB endemicity even reported that the number of M. kansasii isolates exceeded that of M. tuberculosis isolates [46]. In our study, we recruited patients infected with M. kansasii from a distinct region in The Netherlands where an unusually high frequency of infection with M. kansasii had been described [47]. Thus, the geographical clustering of M. kansasii infections could imply that this factor could be taken into account when ESAT- 6 and CFP-10 responses are interpreted. Responses to ESAT-6 and CFP-10 observed in a number of long-term owners of tropical fish tanks may have resulted from repeated exposure to M. marinum, which is a common inhabitant of tropical aquaria. The frequency of responses to ESAT- 6 and CFP-10 among veterinarians and flower growers was unexpectedly high, but we think that the internal consistency between ELISA and ELISPOT underscores the validity of the observed responses. With the exception of 1 older veterinarian, none of these healthy study subjects had ever had recognized contact with a patient with TB or been included in a contact investigation. These persons were from groups with very low risk for exposure to M. tuberculosis (younger age, rural residence, and the absence of travel to regions in which TB is endemic), as confirmed by the responses to the questionnaire. With regard to these characteristics, they were identical to the healthy control subjects with no intensive exposure to environmental mycobacteria who were tested in this and previous studies and whose responses to ESAT-6 and CFP-10 generally were undetectable [12, 14, 15, 48, 49]. The combination of low TB risk profile and mostly negative TST results among persons with intensive exposure to environmental mycobacteria strongly argues against unrecognized latent infection with M. tuberculosis. More likely, the observed responses to ESAT-6 and CFP- 10 in these persons were induced by exposure to mycobacteria other than M. tuberculosis, but our study did not provide data that could identify the environmental mycobacterial species that produced those responses. M. kansasii and M. marinum lysates were not tested in our study, but results might not have been convincing, even if they had been included, because of the high level of cross-reactivity between mycobacterial lysates or tu-

9 JID 2002;186 (15 December) RD1 Responses after Mycobacterial Infection 1805 berculins of various species, including M. marinum and M. kansasii (as was recently demonstrated in TST responses) [40]. This poses a dilemma: there is currently no reference test that can be used to demonstrate infection with M. kansasii or M. marinum in persons without manifest illness. M. kansasii and M. marinum are known to be inhabitants of water, but their predilection differs in that M. kansasii prefers man-made water supply systems, whereas M. marinum can be found in a large variety of water reservoirs, both man-made and natural. Flower growers use both types of water sources during their work, and those in our study had been active for decades in this line of work and thus could be expected to have been repeatedly exposed to M. kansasii and M. marinum. However, this is not the case for veterinarians, who are more likely to be exposed to soil bacteria while examining and treating animals. The possibility exists that homologues of ESAT-6 and CFP-10 are expressed by other environmental mycobacterial species; a recent in silico analysis of 98 microbial, mainly mycobacterial, genomes showed the presence of genome sequences partly homologous to the ESAT-6 gene cluster of M. tuberculosis in several species [50]. Moreover, many environmental mycobacterial species have not been investigated in this regard, because they either are not culturable with currently used techniques or are considered completely nonpathogenic. In a recent study, a high rate of cross-reactivity was found between the T cell responses of patients with TB and those of patients with leprosy to ESAT-6 of M. tuberculosis and to the partly homologous protein of M. leprae [23], and similar findings were obtained with CFP-10 homologues (A. Geluk, K. E. van Meijgaarden, S. M. Arend, and T. H. M. Ottenhoff, unpublished results). Both ESAT-6 and CFP-10 contain multiple epitopes along the entire length of the proteins that can be recognized in the context of different HLA-DR types [29], which suggests that multiple opportunities exist for cross-reactive immune responses to partial homologues to be elicited [51]. The sequencing of the esat-6 and cfp-10 genes of M. marinum and M. kansasii may reveal differences between these and the M. tuberculosis homologues. Preliminary results indicate that several small differences are present along the length of the proteins (S. M. Arend, P. de Haas, and D. van Soolingen, unpublished results). The TST results obtained in our study must be interpreted with caution. According to the guidelines in The Netherlands, we used 2 TU of PPD, compared to the 5 TU used in the United States. Thus, TST responses may have been underestimated in our study. In addition, according to the local consensus, we used 10 mm of induration as the criterion for a positive TST result, although the most recent guidelines described in the official statement of the Centers for Disease Control and Prevention/ATS define a positive TST response as 5 mm, 10 mm, or 15 mm of induration, depending on an individual s level of risk of infection with M. tuberculosis and immune status [52]. However, in our study, selection of study subjects was based exclusively on clinical history, without knowledge of TST results, and different TST results would, therefore, not have influenced the overall results of the study. We think that future studies addressing the sensitivity and specificity of RD1-based diagnostic tests should take into account exposure to environmental mycobacteria, but here another dilemma arises: currently, there is no measure available to express or quantify such exposure, except, perhaps, the history of a person s work, hobbies, previous travel, and lifestyle. The persons with intensive exposure to environmental mycobacteria who were included in our study probably represent the extreme of such exposure, and further study is needed to determine the effect, if any, of less pronounced levels of exposure. There are still practical limitations to be overcome on the way toward a reliable, feasible, and affordable diagnostic test for detection of infection with M. tuberculosis. Both the ELISA and the ELISPOT technique that were used in our study require the isolation of PBMC, which is probably the most labor-intensive step of the assay and is likely to limit its implementation in general clinical practice, especially in resource-poor circumstances. There have been reports of detection of IFN-g production by ELISA in cultures of diluted or undiluted whole blood stimulated with PPD [53 55], which were characterized by cross-reactive immune responses, just as is the TST. The use of whole-blood assays based on ESAT-6 [56] or ESAT-6 and CFP-10 [49] resulted in a higher specificity but lower sensitivity, compared with PPD. To our knowledge, a whole blood based ELISPOT assay using mycobacterial antigens has not yet been described. The results of this study demonstrate that T cell responses to ESAT-6 and CFP-10 are not completely specific for infection with M. tuberculosis and pathogenic M. bovis but may result from infection with M. marinum or M. kansasii. This finding is relevant in light of current efforts to develop a TB-specific immunodiagnostic assay. Acknowledgments The cooperation of all patients and control subjects is gratefully acknowledged. We thank Peter Andersen from the Department of Tuberculosis Immunology at the Statens Serum Institute in Copenhagen, Denmark, for making available recombinant ESAT-6 and culture filtrate protein 10. References 1. Huebner RE, Schein MF, Bass JB Jr. The tuberculin skin test. Clin Infect Dis 1993;17: Menzies RI. Tuberculin skin testing. In: Reichman LB, Hershfield ES, eds. Tuberculosis, a comprehensive international approach. 2nd ed. New York: Marcel Dekker, 2000: Mahairas GG, Sabo PJ, Hickey MJ, Singh DC, Stover CK. Molecular analysis of genetic differences between Mycobacterium bovis BCG and virulent M. bovis. J Bacteriol 1996;178: Gordon SV, Brosch R, Billault A, Garnier T, Eiglmeier K, Cole ST. Identification of variable regions in the genomes of tubercle bacilli using bacterial artificial chromosome arrays. Mol Microbiol 1999;32:

10 1806 Arend et al. JID 2002;186 (15 December) 5. Behr MA, Wilson MA, Gill WP, et al. Comparative genomics of BCG vaccines by whole-genome DNA microarray. Science 1999;284: Harboe M, Oettinger T, Wiker HG, Rosenkrands I, Andersen P. Evidence for occurrence of the ESAT-6 protein in Mycobacterium tuberculosis and virulent Mycobacterium bovis and for its absence in Mycobacterium bovis BCG. Infect Immun 1996;64: Sørensen AL, Nagai S, Houen G, Andersen P, Andersen ÅB. Purification and characterization of a low-molecular-mass T-cell antigen secreted by Mycobacterium tuberculosis. Infect Immun 1995;63: Berthet FX, Rasmussen PB, Rosenkrands I, Andersen P, Gicquel B. A Mycobacterium tuberculosis operon encoding ESAT-6 and a novel low-molecular-mass culture filtrate protein (CPF-10). Microbiology 1998;144: Andersen P, Andersen ÅB, Sørensen AL, Nagai S. Recall of long-lived immunity to Mycobacterium tuberculosis infection in mice. J Immunol 1995; 154: Brandt L, Oettinger T, Holm A, Andersen AB, Andersen P. Key epitopes on the ESAT-6 antigen recognized in mice during the recall of protective immunity to Mycobacterium tuberculosis. J Immunol 1996;157: Pollock JM, Andersen P. The potential of the ESAT-6 antigen secreted by virulent mycobacteria for specific diagnosis of tuberculosis. J Infect Dis 1997;175: Ravn P, Demissie A, Eguale T, et al. Human T cell responses to the ESAT- 6 antigen from Mycobacterium tuberculosis. J Infect Dis 1999;179: Mustafa AS, Amoudy HA, Wiker HG, et al. Comparison of antigen-specific T-cell responses of tuberculosis patients using complex or single antigens of Mycobacterium tuberculosis. Scand J Immunol 1998;48: Ulrichs T, Munk ME, Mollenkopf H, et al. Differential T cell responses to Mycobacterium tuberculosis ESAT6 in tuberculosis patients and healthy donors. Eur J Immunol 1998;28: Arend SM, Andersen P, Van Meijgaarden KE, et al. Detection of active tuberculosis infection by T cell responses to early-secreted antigenic target 6-kDa protein and culture filtrate protein 10. J Infect Dis 2000;181: Arend SM, Ottenhoff TH, Andersen P, van Dissel JT. Uncommon presentations of tuberculosis: the potential value of a novel diagnostic assay based on the Mycobacterium tuberculosis specific antigens ESAT-6 and CFP-10. Int J Tuberc Lung Dis 2001;5: Munk ME, Arend SM, Brock I, Ottenhoff TH, Andersen P. Use of ESAT- 6 and CFP-10 antigens for diagnosis of extrapulmonary tuberculosis [letter]. J Infect Dis 2001;183: Arend SM, Engelhard AC, Groot G, et al. Tuberculin skin testing compared with T-cell responses to Mycobacterium tuberculosis specific and nonspecific antigens for detection of latent infection in persons with recent tuberculosis contact. Clin Diagn Lab Immunol 2001;8: Lalvani A, Nagvenkar P, Udwadia Z, et al. Enumeration of T cells specific for RD1-encoded antigens suggests a high prevalence of latent Mycobacterium tuberculosis infection in healthy urban Indians. J Infect Dis 2001; 183: Andersen P, Munk ME, Pollock JM, Doherty TM. Specific immune-based diagnosis of tuberculosis. Lancet 2000;356: Skjøt RLV, Oettinger T, Rosenkrands I, et al. Comparative evaluation of low-molecular-mass proteins from Mycobacterium tuberculosis identifies members of the ESAT-6 family as immunodominant T-cell antigens. Infect Immun 2000;68: Colangeli R, Spencer JS, Bifani P, et al. MTSA-10, the product of the Rv3874 gene of Mycobacterium tuberculosis, elicits tuberculosis-specific, delayedtype hypersensitivity in guinea pigs. Infect Immun 2000;68: Geluk A, Van Meijgaarden KE, Franken KL, et al. Identification and characterization of the ESAT-6 homologue of Mycobacterium leprae and T- cell cross-reactivity with Mycobacterium tuberculosis. Infect Immun 2002; 70: Lalvani A, Pathan AA, McShane H, et al. Rapid detection of Mycobacterium tuberculosis infection by enumeration of antigen-specific T cells. Am J Respir Crit Care Med 2001;163: Ulrichs T, Anding P, Porcelli S, Kaufmann SH, Munk ME. Increased numbers of ESAT-6 and purified protein derivative specific gamma interferon producing cells in subclinical and active tuberculosis infection. Infect Immun 2000;68: Lalvani A, Pathan AA, Durkan H, et al. Enhanced contact tracing and spatial tracking of Mycobacterium tuberculosis infection by enumeration of antigen-specific T cells. Lancet 2001;357: Tanguay S, Killion JJ. Direct comparison of ELISPOT and ELISA-based assays for detection of individual cytokine-secreting cells. Lymphokine Cytokine Res 1994;13: Favre N, Bordmann G, Rudin W. Comparison of cytokine measurements using ELISA, ELISPOT and semi-quantitative RT-PCR. J Immunol Methods 1997;204: Arend SM, Geluk A, Van Meijgaarden KE, et al. Antigenic equivalence of human T-cell responses to Mycobacterium tuberculosis specific RD1-encoded protein antigens ESAT-6 and culture filtrate protein 10, and those to mixtures of synthetic peptides. Infect Immun 2000;68: Diagnosis and treatment of disease caused by nontuberculous mycobacteria. Am Rev Respir Dis 1990;142: Andersen P, Askgaard D, Ljungqvist L, Bennedsen J, Heron I. Proteins released from Mycobacterium tuberculosis during growth. Infect Immun 1991;59: Harth G, Lee B-Y, Wang J, Clemens DL, Horwitz MA. Novel insights into the genetics, biochemistry, and immunocytochemistry of the 30-kilodalton major extracellular protein of Mycobacterium tuberculosis. Infect Immun 1996;64: Arend SM, Janssen R, Gosen JJ, et al. Multifocal osteomyelitis caused by nontuberculous mycobacteria in patients with a genetic defect of the interferon-g receptor. Neth J Med 2001;59: Renshaw PS, Panagiotidou P, Whelan A, et al. Conclusive evidence that the major T-cell antigens of the M. tuberculosis complex ESAT-6 and CFP- 10 form a tight, 1:1 complex and characterisation of the structural properties of ESAT-6, CFP-10 and the ESAT-6 CFP-10 complex: implications for pathogenesis and virulence. J Biol Chem 2002;277: Tønjum T, Welty DB, Jantzen E, Small PL. Differentiation of Mycobacterium ulcerans, M. marinum, and M. haemophilum: mapping of their relationships to M. tuberculosis by fatty acid profile analysis, DNA-DNA hybridization, and 16S rrna gene sequence analysis. J Clin Microbiol 1998;36: Ramakrishnan L, Falkow S. Mycobacterium marinum persists in cultured mammalian cells in a temperature-restricted fashion. Infect Immun 1994; 62: Tchornobay AM, Claudy AL, Perrot JL, Levigne V, Denis M. Fatal disseminated Mycobacterium marinum infection. Int J Dermatol 1992;31: King AJ, Fairley JA, Rasmussen JE. Disseminated cutaneous Mycobacterium marinum infection. Arch Dermatol 1983;119: Kullavanijaya P, Sirimachan S, Bhuddhavudhikrai P. Mycobacterium marinum cutaneous infections acquired from occupations and hobbies. Int J Dermatol 1993;32: Fine PE, Floyd S, Stanford JL, et al. Environmental mycobacteria in northern Malawi: implications for the epidemiology of tuberculosis and leprosy. Epidemiol Infect 2001;126: Lillo M, Orengo S, Cernoch P, Harris RL. Pulmonary and disseminated infection due to Mycobacterium kansasii: a decade of experience. Rev Infect Dis 1990;12: Evans SA, Colville A, Evans AJ, Crisp AJ, Johnston ID. Pulmonary Mycobacterium kansasii infection: comparison of the clinical features, treatment and outcome with pulmonary tuberculosis. Thorax 1996;51: Anzalone G, Cei M, Vizzaccaro A, Tramma B, Bisetti A. M. kansasii pulmonary disease in idiopathic CD4 T-lymphocytopenia. Eur Respir J 1996;9: Witzig RS, Fazal BA, Mera RM, et al. Clinical manifestations and implications of coinfection with Mycobacterium kansasii and human immunodeficiency virus type 1. Clin Infect Dis 1995;21:77 85.