Other members of the VACSEL Study Group are as follows. University of Zambia School of Medicine, Lusaka, Zambia: Professor Chifumbe Chintu, Ms Gina Mulundu and Dr Peter Mwaba. Statens Serum Institute, Denmark: Dr Peter Andersen. Armauer Hansen Research Institute, Addis Ababa, Ethiopia: Ms Abebech Demissie. MRC, Gambia: Professor K. P. W. J. McAdam (until March 2003), Dr David Warndorff (until July 2001) and Dr Christian Lienhardt (until end of 2000).
Graham Rook, The Centre for Infectious Diseases and International Health, Windeyer Institute of Medical Sciences, Royal Free and University College Medical School, Cleveland Street, London W1P 6DB, UK. E-mail: firstname.lastname@example.org
Expression of interleukin (IL)-4 is increased in tuberculosis and thought to be detrimental. We show here that in healthy contacts there is increased expression of its naturally occurring antagonist, IL-4delta2 (IL-4δ2). We identified contacts by showing that their peripheral blood mononuclear cells (PBMC) released interferon (IFN)-γ in response to the Mycobacterium tuberculosis-specific antigen 6 kDa early secretory antigenic target (ESAT-6). Fresh unstimulated PBMC from these contacts contained higher levels of mRNA encoding IL-4δ2 (P=0·002) than did cells from ESAT-6 negative donors (noncontacts). These data indicate that contact with M. tuberculosis induces unusual, previously unrecognized, immunological events. We tentatively hypothesize that progression to active disease might depend upon the underlying ratio of IL-4 to IL-4δ2.
Although there is strong evidence that type 1 T helper (Th1) lymphocytes and interferon (IFN)-γ production are needed for protection against mycobacterial infection, the role of type 2 T helper (Th2) cells secreting interleukin (IL)-4 is not clear. IL-4 functions at very low concentrations, and has a low mRNA copy number, a very short half-life and a splice variant (IL-4δ2) that acts as a competitive antagonist.1,2 Appropriate methods have shown increased expression of both IL-4 and IL-4δ2 in fresh unstimulated peripheral blood mononuclear cells (PBMC) from tuberculosis (TB) patients, and the level of IL-4 mRNA correlates with disease severity.3 The secretion of IL-4 by CD8 lymphocytes from TB patients has also been described.4,5 However, so far there have been no studies focused on production of IL-4 and its splice variant in latently infected healthy TB contacts. This is important because in mice a pre-existing IL-4 response exacerbates susceptibility to subsequent infection.6,7 IL-4 is a potent down-regulator of inducible nitric oxide synthase (iNOS),8 which is implicated in the induction of the latent or ‘dormant’ state in Mycobacterium tuberculosis9 and so may contribute to determining whether infected contacts develop progressive disease or not.
To identify the recently infected contacts we have used the secretion of IFN-γ in response to 6 kDa early secretory antigenic target (ESAT-6). This is currently the most sensitive and specific test available,10 and a second objective of this study was therefore to compare the use of this test with that of more traditional methods, such as responses to purified protein derivative (PPD) in vitro or in vivo, or cohabitation with a known index case (i.e. household contact).11,12 In this study we have therefore used responsiveness to ESAT-6 to ask whether the increased expression of IFN-γ, IL-4 and IL-4δ2 can be seen as a correlate of infection with M. tuberculosis even in individuals in whom no disease is apparent.
Materials and methods
Samples were collected after appropriate Ethics Committee approval. TB index cases (IC), their household contacts (HC) and matched community controls (CC), were selected consecutively from within a larger cohort in The Gambia. Cases were defined as newly diagnosed smear-positive pulmonary TB cases older than 15 years. Pulmonary TB was confirmed by the presence of acid-fast bacilli in two consecutive sputum samples, or one positive sputum sample and a positive culture. No prophylaxis was given to contacts, and after more than 1 year none of them is known to have developed tuberculosis. For each case enrolled, a matched community control was recruited at random from a nearby household in the same community and recent exposure to an identified TB case excluded by questionnaire. A blood sample for a functional T-cell immunoassay and RNA extraction was obtained from cases, controls and contacts. Chest X-rays and Mantoux tuberculin skin tests (TSTs) were performed at enrolment.
Ex vivo IFN-γ ELISPOT assay
The assay was performed as previously described.13 For this study, PPD (20 μg/ml), ESAT-6 (10 μg/ml) (Statens Serum Institut, Copenhagen, Denmark), phytohaemagglutinin (PHA) (5 μg/ml) as positive control and media as negative control were added to duplicate wells of an ELISPOT plate. PBMC were analysed immediately or after recovery from liquid nitrogen. ELISPOT plates were counted using an AID plate reader (Autoimmun Diagnostika, Strasburg, Germany). A positive response to antigen was taken as twice the background with more than 10 spot forming units (SFU) and more than 20 SFU above background.
RNA was extracted from 1 × 106 PBMC using RNAzol B™ (Biogenesis, Poole, UK) and purified according to the manufacturer's instructions. Amendments to the protocol included isopropanol precipitation of RNA for 1 hr at −20° and the drying of the RNA pellet for 5 min at 60°. Reverse transcription of mRNA was performed using oligo-dt and the Omniscript Kit (Qiagen, Crawley, UK). A volume of 20 µl of RNA extract was reverse-transcribed in a total reaction volume of 60 µl. cDNA was stored at −20° until use. Real-time PCR was performed using the ABI SDS 7700 and Quantitect Mastermix kit (Qiagen). Standard curves of quantified and diluted PCR product, and at least eight negative controls per 96-well plate, were included in each PCR run. Primers were designed for the following targets: GAPDH (forward ATCATCCCTGCCTCTACTGG, reverse TGCTGTAGCCAAATTCGTTG), IFN-γ (forward ATTCGGTAACTGACTTGAATGTCC, reverse CTCTTCGACCTCGAAACAGC), IL-4 (forward CGAGTTGACCGTAACAGACAT, reverse CGTCTTTAGCCTTTCCAAGAAG) and IL-4δ2 (forward CAGAGCAGAAGAACACAACTG, reverse GTCTTTAGCCTTTCCAAGAAG).14 Cycling conditions of an initial activation step of 15 min at 95° followed by 45 cycles of 30 s at 94°, 30 s at 60° and 1 min at 72° were used for each primer pair.
HC and CC were subdivided on the basis of ESAT-6 positivity by ELISPOT, to give a factorial structure with four groups (HC–, HC+, CC– and CC+ with 13, 14, 17 and 5 individuals, respectively). The data for expression of mRNA encoding IL-4δ2 and IL-4 were analysed using rank-transformed analysis of variance, with randomized F-tests. The IFN-γ data were log-transformed and analysed using a standard parametric analysis of variance. Proportions were compared using Fisher's exact test and conditional logistic regression.
When the epidemiological and ELISPOT data were analysed according to the conventional grouping of ‘index cases’, ‘household contacts’ and ‘community controls’, a significant difference was found for the Mantoux skin test, which decreased progressively from cases to contacts to controls (P>0·001) (Table 1). When lymphocytes were stimulated in ex vivo IFN-γ ELISPOT assays with ESAT-6, the frequency of response showed a decreasing progression across groups with a significant proportional difference (P=0·036, Fisher's exact test). The data are consistent with other household-based TB case–contact studies indicating higher M. tuberculosis infection amongst contacts.10–12
Table 1. Summary immunoepidemiology data for index cases, household contacts and community controls
Index cases (IC)
Household contacts (HC)
Community controls (CC)
The Mantoux skin test; decreases progressively from IC to HC to CC (P>0·001).
The frequency response to ESAT-6 antigen in IFN-
ex vivo ELISPOT assay for HC and CC showed a decreasing progression with a significant proportional difference (P=0·036, Fisher's exact test).
There were no significant differences for eosinophil count, BCG scar (data not shown), or PPD ELISPOT frequency.
Nevertheless, in an endemic setting, some community controls are likely to be contacts of unknown random cases, and not all contacts are necessarily infected. Therefore, in an attempt to focus analysis only on individuals recently infected with M. tuberculosis, contacts and controls were pooled and re-segregated according to their response to ESAT-6 in the ELISPOT assay. Segregated in this way, the ESAT-6 positive group had significantly higher Mantoux scores (P=0·04) than ESAT-6 negatives [95% confidence interval (CI) 0·2–7·2 mm; data not shown].
For PPD ELISPOT there were no significant differences between contacts and controls. However, when pooled and re-segregated according to ESAT-6 ELISPOT response, the ESAT-6 positive group had a significantly higher proportion of PPD ELISPOT positive subjects (P=0·013, conditional logistic regression, data not shown).
Gene expression profiles
When gene expression profiles were analysed, the ESAT-6 positive group showed significantly increased levels of mRNA for both IL-4 (P=0·04) and IL-4δ2 (P=0·002), compared to the ESAT-6 negative group. No such differences in mRNA level were seen when the HC and CC groups were compared without segregation by ESAT-6 response (HC and CC, Fig. 1). The differences between ESAT-6 positive and ESAT-6 negative groups were consistent, allowing for gender, age, eosinophil count, TST and bacille Calmette Guérin (BCG) scar. Consequently at the 5% level there were no interactions between the ESAT-6 positive and ESAT-6 negative groups and these covariates. Therefore the main effects as illustrated in Fig. 1 provide a valid inference. Levels of mRNA encoding IL-4 and IL-4δ2 were strongly correlated in the ESAT-6 positive individuals (0·00016, Pearson's rank order correlation), and in those ESAT-6 negative donors in whose cells both cytokines were detectable (Fig. 2).
The levels of mRNA encoding IFN-γ were similar regardless of the response to ESAT-6 (data not shown). The increased expression of mRNA encoding IL-4 and IL-4δ2 appeared to be specific for the ESAT-6 positive group, as no such differences were found when the pooled HC and CC were segregated by PPD ELISPOT or Mantoux status rather than by ESAT-6 response.
It has been shown that there is greatly increased expression of both IL-4 and IL-4δ2 in fresh unstimulated PBMC from TB patients.3 (This was confirmed in the small group of 11 index cases associated with the current study, though here we analysed only the larger groups of contacts and community controls.) The level of IL-4 mRNA correlates with disease severity.3 Furthermore, recent work in Balb/c mice, in which progressive TB is accompanied (as in humans) by increased IL-4 expression, has shown that IL-4-knockout markedly reduces both pulmonary fibrosis and the toxicity of tumour necrosis factor (TNF)-α, which are neglected aspects of progressive human disease.15 These points, together with the possibility that the ability of IL-4 to down-regulate iNOS8 might promote a switch from latency to progressive disease,9 led us to investigate IL-4 and its splice variant IL-4δ2 in a TB case–contact study using sensitive RT-PCR assays. It is estimated that 59% of human genes are alternatively spliced,16 and splice variants along with nonprotein coding RNA and antisense RNA play an important role in the control of gene expression. IL-4δ2 is a naturally occurring splice variant of IL-4 that is preferentially expressed in the thymus and airways and thought to inhibit the function of IL-4 by competing for binding sites.1,2 Recently, IL-4δ2 has been characterized in nonhuman primates but was absent from mouse and rat.17 It is therefore possible that any detrimental effect of IL-4 on immunity to TB can be counteracted in humans and primates (but not rodents) by release of a competitive antagonist.1,2 Thus, in humans, assay of the two mRNAs separately is essential, or the net outcome in terms of IL-4 activity cannot be assessed. Currently available enzyme-linked immunosorbent assay (ELISA) and ELISPOT assays do not distinguish between the two cytokines. This is the first study to differentiate between IL-4 and IL-4δ2 in healthy TB contacts.
In addition, we chose to further refine the case–contact model by separating recently infected from noninfected individuals using the ESAT-6 ELISPOT assay.13 Fourteen of 27 contacts and five of 22 controls were ESAT-6 positive. In this endemic setting, the five ESAT-6 positive controls are most likely to be positive as a result of prior exposure to TB, though cross-reactivity with environmental mycobacteria has also been reported.18 Analysing the data according to an ESAT-6 positive or ESAT-6 negative response reveals correlates that would have been overlooked if the data had been analysed only by the conventional grouping contacts (HC) vs. controls (CC).
Interestingly, the IL-4 response in TB patients differs from that seen in asthma, where IL-4 itself is expressed about 1000 times more than IL-4δ2.19 In TB patients both cytokines are increased.19 However, our data indicate that healthy ESAT-6 positive individuals have a higher ratio of IL-4δ2 to IL-4 compared to ESAT-6 negatives (P=0·04, Mann–Whitney). Our sample size is small, but if this is confirmed in larger studies, it will be rational to hypothesize that the increased IL-4δ2 is a physiological attempt to block the activity of IL-4 induced by M. tuberculosis.1,2 Similarly, it will be of interest to discover whether progression to active TB is accompanied by a relative increase in the ratio of IL-4 to IL-4δ2, allowing IL-4 agonist activity, macrophage deactivation, reactivation of latent disease, fibrosis and immunopathology.
In conclusion, we have demonstrated that separating infected and noninfected contacts on the basis of a cellular immune response to ESAT-6 reveals immunological markers that do not emerge from conventional contact–control comparisons. In addition, we have found that increased expression of IL-4δ2, and possibly of IL-4, in fresh unstimulated PBMC is associated with recent TB infection. We recently learnt that a similar significant association has been found in Singapore, using RT-PCR for IL-4 mRNA in unstimulated cells, and an ELISA assay for IFN-γ release in response to overlapping peptides representing ESAT-6 and culture filtrate protein 10 (G. T. Seah, National University of Singapore, Singapore, personal communication). Similarly, studies undertaken in Ethiopia as another arm of VACSEL are revealing the same phenomenon in a third ethnically different population.20 It is clear that further investigations of the role of IL-4 in latent TB, and during progressive infection, are urgently needed.
We are grateful to the European Union INCO-DEV Framework 5 Programme for supporting this work (VACSEL; Grant Contract Number: ICA-CT-1999-1005).