a longitudinal study of in vitro ifnγ production and cytotoxic t cell responses of tuberculosis...

A longitudinal study of in vitro IFNgproduction and cytotoxic T cellresponses of tuberculosis patients inThe Gambia

J. Turner,* T. Corrah,† S. Sabbally,† H. Whittle,† H. M. Dockrell*Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, UK†MRC Laboratories, Fajara, The Gambia

Summary Setting: This study was carried out at the MRC Laboratories, The Gambia.Objectives: To characterize the antigen-specific IFNg production and cytotoxic T cell (CTL) responses of patientsduring active tuberculosis, treatment, and following recovery.

Design: PBMC were isolated from 37 patients with tuberculosis and incubated with either PPD, live M. bovis BCG,or no antigen and IFNg production measured after 7 days. CTL activity against these antigens was determined usingautologous antigen-pulsed monocyte-derived macrophages as target cells. A subset of these patients (7–18depending on antigen and assey used) were tested 2 months into drug treatment and 3 months after discharge. A group of blood bank donors (n = 21) were also tested to evaluate IFNg responses in endemic controls; a subset (n = 16) were also tested for CTL activity.

Results: The ability to produce IFNg in response to mycobacterial antigens correlated with the Mantoux skin teststatus of the patient. IFNg production to live M. bovis BCG was diminished at diagnosis but returned after 2 monthsof drug treatment, and was sustained after completion of drug therapy. The CTL responses to both PPD and live M. bovis BCG were reduced during the period of drug treatment compared to those at diagnosis, but returned to theoriginal levels after recovery.

Conclusions: Drug treatment induced marked alterations in the immune responses of tuberculosis patients withinduction of IFNg production in response to stimulation with live M. bovis BCG. This may indicate activation of bothCD4 and CD8 T cells. © 2000 Harcourt Publishers Ltd

Short communication

Tubercle and Lung Disease (2000) 80(3), 161–169© 2000 Harcourt Publishers Ltddoi:10.1054/tuld.2000.0241, available online athttp://www.idealibrary.com on

INTRODUCTION

Interferon gamma (IFNg) is a key cytokine involved in theprotective immune response against mycobacterial infec-tion. The primary function of this cytokine is to activatethe macrophage and enable this cell to carry out itsmycobacterial effector functions. A failure to generatethis cytokine in response to mycobacterial infectionproves fatal in the mouse model1,2 and in humans, an
inability to respond to IFNg by way of IFNg receptor defi-
Correspondence to: Dr H. M. Dockrell, Department of Infectious and TropicalDiseases, London School of Hygiene and Tropical Medicine, Keppel Street,London WC1E 7HT, UK. Tel.: +44 207 927 2466; Fax: +44 207 637 4314; E-mail: [email protected].

Accepted: 12 May 2000

ciency, has been shown to result in increased susceptibil-ity to mycobacterial infections.3 Both CD4+ and CD8+ Tcells have been shown to secrete this cytokine after acti-vation with mycobacterial antigen, and its production isan important indicator of T cell function.4–6 In this studywe assessed the ability of individuals infected with M. tuberculosis to produce IFNg in response to mycobac-terial antigens.

A cytotoxic T cell (CTL) response to mycobacterial anti-gens has been postulated whereby the CTL can lyse aninfected non-activated macrophage thus enabling themycobacteria to be taken up by an activated cell.7 Datafrom murine gene knockout mice suggest that this cellfunction may be utilized late in mycobacterial infection,
and a failure to generate CTL may be responsible for theregrowth of M. tuberculosis in this model.8,9 Several
161

162 Turner, Corrah, Sabbally et al.

human in vitro studies on cytotoxic T cells (CTL) havebeen carried out using autologous macrophages pulsedwith soluble mycobacterial antigens. CTL activity, medi-ated by effector cells restimulated in vitro with purifiedprotein derivative (PPD), has been demonstrated inhealthy controls and tuberculosis patients.10,11 However,the use of soluble antigen in these studies is not repre-sentative of a natural mycobacterial infection, as it isprocessed and presented via the MHC class II pathway. Ithas been shown that PPD induces CTL primarily of theCD4+ T cell phenotype, and most mycobacterial specificCTL activity has been attributed to this population.10,12,13

However, it has been demonstrated that the CD8+ T cellpopulation can also behave in a cytotoxic mannertowards macrophages which have been infected with thelive M. bovis BCG14 which might be more representativeof a natural infection. We therefore chose to compareboth PPD and live M. bovis BCG for their abilities toinduce CTL activity in patients with tuberculosis. Themeasurement of T cell immunity in patients with tuber-culosis is problematic due to the inability to analyse anindividual pre-infection. Often, particularly in developingcountries, the disease itself has to manifest severely witha productive cough and severe lung pathology before itcan be diagnosed. Thus an individual diagnosed withtuberculosis is at an extremely late phase of the infection.As an alternative approach to analysing individuals pre-exposure with M. tuberculosis, we requested the patientsto return to the clinic during their course of treatmentwhilst undergoing drug therapy, and then again 3 monthsafter the completion of drug treatment. This allowed us tocompare the immune response in individuals that hadbeen previously exposed to mycobacteria with that dur-ing active disease.

MATERIAL AND METHODS

Population

The study was carried out in Fajara, The Gambia. Patientswere recruited from individuals diagnosed with pul-monary tuberculosis at the MRC Outpatient Clinic or theSerrakunda Health Centre. The majority of patients (33out of 37) had not previously presented with tuberculosis.Diagnosis was carried out by the resident clinicians andconfirmed by the presence of acid fast bacilli (AFB) in thesputum. The patients in the study were HIV negative asdiagnosed by a rapid screen ELISA (HIVCHEK, OrthoDiagnostic Systems Inc, Bucks, UK). The tuberculosispatients were Mantoux skin tested with PPD (EvansMedical Ltd, Horsham, UK: 10U/0.1 ml) 1–2 weeks beforeinitiation of chemotherapy and 73% of the patients werePPD skin test positive. Skin testing with Candida (Hollister
Stair, Bayer Corporation, Washington, USA: 1/20 dilution,
Tubercle and Lung Disease (2000) 80(3), 161–169

0.1 ml) was simultaneously carried out as a control anti-gen. Evidence of a BCG scar was found in 62% of patientsin the study. Patient age ranged from 18–70 years, andthe male: female ratio was 3.5:1. Once the tuberculosisdiagnosis was confirmed, patients were initiated ontodirectly observed therapy (DOTS) as recommended byWHO, which consisted of alternative day, 4 drug therapy(isoniazid, rifampicin, pyrazinamide, and ethambutol) for2 months. Weight gain was used as a simple method fordetermining recovery and 24 of the returning 27 patientsgained weight after 2 months of treatment. A clinical andbacteriological examination was also carried out and thepatients were then given two drugs (isoniazid, rifampicin)for a further 4 months. At the completion of drug therapythe patients were recalled to the clinic and, if consideredto be free from infection with M. tuberculosis were dis-charged. In this study patients were asked to return to theclinic 3 months after discharge when they had not takenany anti-mycobacterial drugs for a period of 3 months.

Blood samples were also obtained from the GambianBlood Transfusion Service as a source of PBMC from indi-viduals who are non-infected or asymptomatic for tuber-culosis (n = 12). Because of the anonymity of the blooddonation it was not possible to skin test these individuals.However a recent study in The Gambia has shown that62% of all healthy individuals are Mantoux positive (aninduration of 5 mm or greater, mean age of 25 (age range12–83, Jepson et al., manuscript in preparation). This sug-gests that the majority of the population have had expo-sure to mycobacteria. All the blood donors were HIVscreened at the blood bank and found to be negative.

Antigens

M. bovis BCG (Glaxo strain) was obtained from EvansMedical Ltd, Liverpool, UK. Cultures were grown fromfreeze-dried stock in 7H9 medium (Difco, Michigan, USA)supplemented with 10% ADC (Difco) and 0.5% Tween 80for 2 weeks prior to their harvest. Aliquots were frozen inRPMI-1640 (Gibco, Paisley, UK) at -70°C and used at a MOIof 0.2–1.0 colony forming units (CFU) per monocyte asindicated, resulting in a final concentration of 2 3 104 – 1 3 105 CFU/ml. Purified protein derivative (PPD), batch47 for in vitro use, was purchased from Statens Serumin-stitut, Copenhagen, Denmark and used at a final con-centration of 10 mg/ml. Streptokinase/streptodornase(Varidase Topical, CV Laboratories, Hants, UK) was usedas an irrelevant control antigen, at a final concentration of250 mg/ml.

Peripheral blood mononuclear cell isolation

Peripheral blood mononuclear cell (PBMC) from tubercu-
losis patients or healthy blood bank controls were isolat-
© 2000 Harcourt Publishers Ltd

IFNg production and CTL responses in tuberculosis patients 163

ed over Ficoll-histopaque gradients (Sigma Chemical Co.,Poole, UK) and resuspended in growth medium (RPMI-1640, 2 mM-glutamine [Gibco], and 10% autologous plasma). Autologous plasma was used rather than pooledhuman AB serum in order to mimic the ability of cells torespond in vivo, as non-specific stimulatory or inhibitoryfactors may be present. The limited availability of PBMCdid not allow the comparison of responses in autologousplasma, with those in human AB serum. Ampicillin wasused in the growth medium at a concentration of 50 mg/ml; this b-lactam antibiotic does not have inhibito-ry effects on mycobacteria which possess b-lactamaseactivity.15 Cultures containing 2 3 106 cells in a volume of2 ml were stimulated with antigen and used for IFNg

determination and measurement of CTL activity asdescribed. In some cases where yields of PMBC were low,it was not possible to test PMBC with all the antigens.

IFNg ELISA

Supernatants were harvested on day 7 from antigen stim-ulated PBMC cultures (2 3 106 PBMC/well) and frozen at-20°C until use. Cultures were stimulated with PPD at afinal concentration of 10 mg/ml or M. bovis BCG at a MOIof 0.2 CFU/monocyte (2 3 104 CFU/ml). Immulon 4 ELISAplate strips (Dynatech, West Sussex, UK) were coated with2.5 mg/ml anti-human IFNg monoclonal antibody (HycultBiotechnology, Uden, Netherlands) overnight at 4°C, fol-lowed by a blocking step for one hour in 2.5% BSA in PBS.Samples were diluted in 5% human AB serum in RPMIprior to loading in 100 ml volumes, in duplicate, onto theplates. A standard curve (0–500 U/ml) was prepared usingrecombinant IFNg obtained from Hoffman La Roche Ltd,Basel, Switzerland. The samples were incubated for 2 h at37°C. The plates were then incubated with detection anti-body (polyclonal anti-human IFNg obtained by immuniz-ing a rabbit with recombinant human IFNg in Ribiadjuvant, kindly provided by Dr P. Kaye, LSHTM, London)for one hour, followed by a goat anti-rabbit IgG peroxi-dase (kpl, Maryland, USA), and then TMB substrate (kpl)to detect the presence of the secondary antibody.Between each step the excess antibody or sample waswashed away using PBS containing 0.05% Tween 20. TheELISA assay had a sensitivity of detection down to 7 U/mlIFNg.

CTL assay

Cytotoxic T cell activity was measured by chromiumrelease assay as optimized and described elsewhere previ-ously.14 Briefly, PMBC (effectors) were incubated for 7 days with either 0.2 colony forming units (CFU) of
BCG/monocyte, 10 mg/ml PPD, or no antigen prior to cellharvest. At day zero, a sample of the PMBC was also plat-

ed into 96-well tissue culture plates to be used as targetcells during the CTL assay. These wells were washed freeof non-adherent cells on day 1 and recultured. On day 6the remaining monocyte-derived macrophages wereloaded with Na2

51CrO4 (1 m Ci/well Amersham, Bucks,Hants) plus antigen (BCG 1 CFU/monocyte, PPD 20 mg/mlor streptokinase/streptodornase 250 mg/ml) 24 h beforethe CTL assay. On day 7 effector cells were harvested, andthen incubated with target cells at a 5:1 ratio, for 6 h. Theconcentration of antigens were optimized for maximalCTL induction using healthy donors in the UK prior tothe study. A positive CTL response was defined as beingabove 4.7% (mean % lysis of all non-stimulated effectorcells, plus twice the standard deviation).

Statistical analysis

The relationship between groups and timepoints was cal-culated using the Student t-test. Where appropriate, apaired Student t-test was used.

RESULTS

IFNg production in tuberculosis patients and healthydonors

Supernatants were collected from PBMC cultures whichhad been incubated with antigen for 7 days, and IFNg lev-els were measured by ELISA. As Figure 1 shows, bothhealthy donors (n = 21) (Fig. 1A) and symptomatic tuber-culosis patients (n = 33) (Fig. 1B) were capable of produc-ing IFNg after stimulation with mycobacterial antigens. Inboth groups PPD was the most potent stimulator of IFNg

production by PBMC although there were no significantdifferences between patients and healthy donors. How-ever, the production of IFNg in response to live M. bovisBCG was depressed in the tuberculosis patients in com-parison to responses in the healthy donors (Student t-test,P = 0.016).

Cytotoxic T cell activity in tuberculosis patients andhealthy donors

Newly diagnosed tuberculosis patients (n = 32) and bloodbank donors (n = 12–16 depending on antigen) wereassayed for CTL activity (Fig. 2). Neither healthy individu-als nor symptomatic tuberculosis patients had detectablecirculating CTL without prior antigenic stimulation (datanot shown). Seven days of in vitro antigen stimulationresulted in the activation of CTL with the extent of cyto-toxicity being dependent on the antigen used for stimula-tion. Stimulation of PBMC with live M. bovis BCG inducedmycobacteria specific CTL which could lyse target cells
pulsed with either live M. bovis BCG or PPD. This was


Fig. 1 PBMC from healthy donors (n = 21) (A) or tuberculosis patients (n = 33) (B) were incubated for 7 days without antigen, or in thepresence of live M. bovis BCG, or PPD. Supernatants were collected and IFNg measured by ELISA. Results are expressed as U/ml IFNg.

A B

observed in both healthy donors (Fig. 2A) and tuberculo-sis patients (Fig. 2B).

Stimulation of PBMC with the soluble antigen PPD produced a strong PPD specific CTL response in patients(Fig. 2D) which was significantly increased in comparisonto that observed in the blood bank donors (Fig. 2C)(Student t-test, P = 0.007). The levels of CTL activity werelow at the effector target ratio of 5:1 but due to the limi-tation on the volume of blood collected it was oftenimpossible to use a higher E:T ratio. However, when suffi-cient numbers of PBMC were available, the ratio wasincreased to 10:1 which resulted in an increase in CTLactivity of 35±22% (n = 10)(Data not shown).

Correlation of T cell responses to Mantoux or BCGvaccination status

The CTL and IFNg responses of patients with tuberculosiswere extremely heterogeneous and the data were analysedfurther to identify any trends within the patient group. Allpatients on this study had previously been tested for skintest reactivity to PPD 1–2 weeks before initiation of ther-apy. In addition, previous BCG vaccination was identifiedfrom the presence of a vaccination scar. Patients weredivided by skin test status, or BCG vaccination. A signifi-
cant increase in the IFNg production induced by both PPD

(Student t-test, P=0.034) and live M. bovis BCG (Student t-test, P = 0.052) was seen in the group of patients whowere skin test positive. However, this relationship did notapply to CTL responses and in addition there was no cor-relation between IFNg production and CTL activity. Thelevels of IFNg production or CTL activity could not beattributed to the presence or absence of previous BCGvaccination.

IFNg production during and after the completion of drugtreatment

The levels of IFNg production in response to PPD (Fig. 3A;n = 7) remained unchanged throughout the study. IFNg

production in response to live M. bovis BCG (Fig. 3B; n = 12) increased after 2 months treatment to levels equalto those observed in the healthy blood bank group andthis increase was sustained at the 9 month timepoint. Dueto low cell yields it was not possible to test all the donorswith PPD, but in those tested (Fig. 3A; n = 7) the levels ofIFNg production in response to PPD remained unchanged.

CTL responses during and after completion of drugtreatment

After 2 months anti-mycobacterial drug treatment the



Fig. 2 PBMC from healthy donors (n = 16, 12) (A, C) or tuberculosis patients (n = 31, 32) (B, D) were incubated for 7 days in the presenceof live M. bovis BCG (A, B) or PPD (C, D). Stimulated effector cells were used in a 6 hour CTL assay using autologous macrophages

51 51 se from target macrophages was subtracted from each sample and

A B

DC

antigen pulsed target cellantigen pulsed target cell

antigen pulsed target cellantigen pulsed target cell

CTL responses to PPD (Fig. 4A, n = 11) and M. bovis BCG(Fig. 4B; n = 18) decreased in comparison to the activityseen at the time of diagnosis. CTL activity in response tolive M. bovis BCG (Fig. 4B) reduced to levels significantlylower than those of the healthy blood bank donors(Student t-test, P = 0.006), whereas the PPD specific

prepulsed with Cr plus antigen as target cells. Spontaneous Cr releathe data expressed as % specific lysis.

response dropped to a level which was no longer signifi-


cantly different to that observed in the healthy bloodbank (Student t-test, P = 0.8). However, 3 months after thecompletion of anti-mycobacterial treatment the CTLactivity against these two antigens returned to the levelsobserved at the time of diagnosis. Therefore PPD specificlysis returned to levels which were significantly elevated
(Student t-test, P = 0.024) in comparison to those seen in


Fig. 3 IFNg responses were compared in patients with tuberculosis at the time of diagnosis, after 2 months of drug therapy, and 3 monthsafter the completion of successful therapy. IFNg responses to PPD (n = 7) (A) or live M. bovis BCG (n = 12) (B) were measured asdescribed in Figure 1 and the results expressed as U/ml IFNg.

Fig. 4 CTL responses were compared in patients with tuberculosis at the time of diagnosis, after 2 months of drug therapy, and 3 monthsafter the completion of successful therapy. CTL responses to PPD (A) (n = 11), or live M. bovis BCG (B) (n = 18) were measured as

A

B

B

A BTime (months) Time (months)

Time (months)Time (months)

the blood donors, while lysis induced by live M. bovisBCG was again equal to that seen in the blood bankdonors.

DISCUSSION

The study set out to evaluate IFNg producing capacity,and CTL activity, of patients with tuberculosis in TheGambia. Although it is clear that T cells, and the produc-tion of certain cytokines such as IFNg, are central to pro-

described in Fig. 2 and the results expressed as % specific lysis.

tection against tuberculosis, much more information is


needed on the contribution of the various T cell subsetsto the ultimate control of bacterial growth. Such informa-tion is needed to aid vaccine design, to allow the identifi-cation of infected individuals who would with timereactivate latent disease, and to predict which patientswill relapse following therapy. In the current study, IFNg

and CTL responses were investigated in patients with pul-monary tuberculosis in The Gambia. These patientsshowed the typical symptoms of pulmonary disease, withcough, night sweats, weight loss and cavitation seen on
X-ray. The relative proportion of cases arising from reacti-


vation versus primary infection or reinfection was notknown.

As controls, a group of blood bank donors were used asthese individuals are healthy, and show no signs of infec-tion with M. tuberculosis. However, contact with M. tuber-culosis could not be ruled out as Mantoux skin testingdata was not available for these donors. The majority ofpeople in The Gambia have been previously vaccinatedwith BCG which induces a positive skin test result.16,17

The blood bank donors were therefore used as a group ofcurrently healthy asymptomatic individuals. However,these individuals may also be a suitable group of controlsin a country where most individuals are naturally ex-posed to M. tuberculosis, but the majority do not go on todevelop the disease.

Comparisons between symptomatic tuberculosispatients and healthy donors showed that during theactive disease, the production of IFNg in response to PPDwas comparable whereas the response to live M. bovisBCG was reduced in the patients. This reduced IFNg pro-duction did not reflect a genetic inability to respond aspatients were able to recover their IFNg producing capac-ities after 2 months of drug treatment, an effect whichwas sustained after a complete recovery. These antigenspecific IFNg secreting cells could be sequestered at thesite of infection, or circulating but unresponsive (anergic)during active disease. This reduction of IFNg productionin tuberculosis patients, and subsequent recovery of IFNg

producing capacities during and after treatment has beendemonstrated in other studies in tuberculosis patients18,19

and our data confirmed these observations. It has beenshown that IFNg producing cells can be found in thepleural fluid of tuberculous pleurisy patients20 which sug-gests that these cells may have migrated to the lung dur-ing active disease, thus accounting for their temporaryabsence in the peripheral blood. More invasive tech-niques for collecting cells from the lung of the infectedindividual would be necessary to confirm these hypothe-ses. The increased numbers of PPD specific CTL in theperipheral blood was a surprising observation as it mightbe assumed that these cells would be found in the site ofinfection. However, other groups have shown that thePPD specific CTL activity in patients with tuberculosis hasbeen elevated both in the peripheral blood10 and in thepleural fluid of tuberculosis pleuritis patients.11

Most of the tuberculosis patients studied showed anincrease in IFNg production after 2 months, an effectwhich was sustained 3 months after completion of treat-ment. In a few patients the IFNg response either fellbetween 0 and 2 months and then remained low, or wasmaintained at a high level throughout the study. A largergroup of patients, who could be followed up for a periodof several years, would be needed to evaluate whether
such response patterns might predict subsequent relapse

or re-infection. At present, it is not clear if this recovery inthe IFNg response may result from T cell stimulation withantigens released following bacterial killing by drugs, bya reversal of an antigen state, or by reduced traffic of antigen-specific T cells to the lungs.

There was a positive correlation between IFNg produc-tion and the presence of a positive Mantoux skin test inthe untreated tuberculosis patients. Skin test reactivity isa measure of the delayed type of hypersensitivityresponse and the production of IFNg is associated withthe ability to produce a positive skin test reaction.However the ability to mount a DTH reaction does notresult in protective immunity which may require a strongIFNg response in addition to other unidentified immuno-logical mediators. In this study, Mantoux skin testing hadbeen performed before chemotherapy was instituted,usually 1–2 weeks before the untreated blood sample was collected for the immunological studies. As adults inThe Gambia are Mantoux skin test positive and havetherefore been pre-sensitized by TB exposure, environ-mental mycobacteria or prior BCG vaccination,16,17 itseems unlikely that the skin test performed to aid diagno-sis would have significantly affected the PBMC CTLresponse to PPD, as although this was higher in theuntreated tuberculosis patients when compared to thecontrols, the IFNg responses in both groups were similar.In countries such as the USA where regular Mantoux skintesting is used to monitor infection in groups such ashealth case workers, repeated skin testing does notinduce sensitisation to PPD.21

One observation from this study which should be ofconcern is the high number of tuberculosis patients whohad evidence of a BCG vaccination scar. BCG vaccinationhas been routinely used in The Gambia since 197917 butclearly this appeared to have very little effect on the pro-tection of this population against M. tuberculosis. Themajority of patients diagnosed in the clinics had pul-monary tuberculosis, with very few cases of extra-pul-monary infection, so perhaps prior BCG vaccination canprevent the dissemination of M. tuberculosis to otherorgans of the body. In common with other countries thisinability of BCG vaccination to protect against tuberculo-sis highlights the urgent need for a more effective vaccine.

The disparity between the levels of IFNg produced inresponse to the antigens tested (PPD and live M. bovisBCG) may be due to the stimulation of different T cellpopulations by these antigens. PPD is known to primarilyactivate CD4+T cells, whereas live M. bovis activates boththe CD4+ and CD8+ cells in vitro.14 The CTL data againreflects the contrast between the different antigens withthe PPD specific CTL response being much higher thanthat seen with live M. bovis BCG. More recent work hasindicated that in BCG vaccinated healthy donors the
CD8+T cells activated by M. bovis BCG produce both IFNg


and TNFa and mediate the antigen-specific lysis of infect-ed monocyte derived macrophages.22 One would alsoexpect to see a stronger correlation between IFNg pro-duction and CTL activity using purified CD8+ T cells, assuch CD8+ cells when activated by BCG both make IFNg

and contain intracellular perforin.22 The elevated CTLresponse to PPD in tuberculosis patients is likely to beCD4+ T cell mediated as shown in other studies10 as PPDis known to predominantly stimulate this T cell popula-tion.14 Further work is needed to confirm the contributionof CD8+T cells to these responses in patients with tuber-culosis, but it appears that the use of whole live mycobac-teria may provide an insight into the immune responsewhich would be missed if only soluble antigens were used.

The study of our patients throughout their drug treat-ment and subsequent recovery revealed some interestingresults. The levels of CTL activity dropped significantlyduring the course of drug treatment, and then returned totheir original levels after complete recovery from tuber-culosis. The action of the mycobacterial drugs appears toaffect the CTL activity, possibly by generating a release ofmycobacterial antigens in the lung, inducing a stronglocal immune response, and in consequence the CTL mayhave migrated to the site of infection. An independentstudy has demonstrated that CTL activity is increased inthe population of cells obtained within the pleural fluid oftuberculosis pleurisy patients, in comparison to levelsobtained from the peripheral blood of the same patient.11

This suggests that CTL may localize at the infection site intuberculous pleuritis patients and a similar localization tothe lung might also occur in pulmonary tuberculosisonce bacteria are killed by anti-mycobacterial drugs. Itwas also noticeable that certain individuals did not recov-er antigen-specific T cell responses; large scale follow-upstudies would be needed to investigate if such individualsare at increased risk of relapse – or of reinfection.

One important observation in this study was the returnof the patient immune response to a level similar to thehealthy blood bank donors, 3 months after completion oftherapy. We would have hypothesized that after infectionand successful treatment the patients might have a boost-ed immune response which would respond to mycobac-terial antigen in a more rapid manner and to a greaterlevel. This does not appear to be the case – rather, theimmune response returned to that seen in the healthyblood bank controls. The only parameter which remainedaltered after recovery was the PPD specific CTL responsewhich initially was elevated and which then returned toelevated levels after the patient had recovered from theinfection. Larger longitudinal studies, and studies onhousehold contacts who do or do not develop diseasemight reveal if an enhanced ability to generate PPD spe-cific CTL response is associated with susceptibility rather
than protection.

ACKNOWLEDGEMENTS

This work was supported in part by a grant from the EuropeanCommunity [TS3*-CT93.0254]. Dr J. Turner was the recipient of aMRC studentship and the Gordon Smith Travelling Scholarship.We would like to thank all the patients from the MRC Gate Clinicwho volunteered to take part in this study. We are also gratefulto the fieldworkers, Ken Joof and Yerro Sowe, for their invaluableservice throughout this project.

REFERENCES

1. Cooper A M, Dalton D K, Stewart T A, Griffin J P, Russell D G,Orme I M. Disseminated tuberculosis in interferon g gene-disrupted mice. J Exp Med Immunol 1993; 178: 2243–2247.

2. Flynn J L, Chan J, Triebold K J, Dalton D K, Stewart T A, Bloom B R. An essential role for interferon g in resistance toMycobacterium tuberculosis infection. J Exp Med 1993; 178:2249–2254.

3. Newport M J, Huxley C M, Huston S et al. A mutation in theinterferon-gamma-receptor gene and susceptibility tomycobacterial infection. N Engl J Med 1996; 335: 1941–1949.

4. Mutis T, Cornelisse Y E, Ottenhoff T H M. Mycobacteria induceCD4+ T cells that are cytotoxic and display Th1-like cytokinesecretion profile: heterogeneity in cytotoxic activity andcytokine secretion levels. Eur J Immunol 1993; 23: 2189–2195.

5. Salgame P, Abrams J S, Clayberger C et al. Differing lymphokineprofiles of functional subsets of human CD4 and CD8 T cellclones. Science 1991; 254: 279–282.

6. Orme I M, Miller E S, Roberts A D, et al. T lymphocytesmediating protection and cellular cytolysis during the course ofMycobacterium tuberculosis infection. J Immunol. 1992; 148:189–196.

7. Lowrie D B. Is macrophage death on the field of battle essentialto victory, or a tactical weakness in immunity againsttuberculosis? Clin Exp Immunol. 1990; 80: 301–303.

8. Cooper A M, D’Souza C, Frank A A, Orme I M. The course ofMycobacterium tuberculosis infection in the lungs of micelacking expression of either perforin-or granzyme-mediatedcytolytic mechanisms. Infect Immun. 1997; 65: 1317–1320.

9. Laochumroonvorapong P, Wang J, Liu C C, et al. Perforin, acytotoxic molecule which mediates cell necrosis, is not requiredfor the early control of mycobacterial infection in mice. InfectImmun. 1997; 65: 127–132.

10. Kumararatne D S, Pithie A S, Drysdale P, Gaston J S, Kiessling R.Specific lysis of mycobacterial antigen-bearing macrophages byclass II MHC-restricted polyclonal T cell lines in healthy donorsor patients with tuberculosis. Clin Exp Immunol. 1990; 80:314–323.

11. Lorgat F, Keraan M M, Lukey P T, Ress S R. Evidence for in vivogeneration of cytotoxic T cells. Am Rev Resp Dis. 1992; 145:418–423.

12. Ottenhoff T H, Mutis T. Specific killing of cytotoxic T cells andantigen-presenting cells by CD4+ cytotoxic T cell clones. J ExpMed. 1990; 171: 2011–2024.

13. Boom W H, Wallis R S, Chervenak K A. Human Mycobacteriumtuberculosis-reactive CD4+ T-cell clones: heterogeneity inantigen recognition, cytokine production, and cytotoxicity formononuclear phagocytes. Infect Immun. 1991; 59: 2737–2743.

14. Turner J, Dockrell H M. Stimulation of human peripheral bloodmononuclear cells with live Mycobacterium bovis BCG activatescytolytic CD8+ T cells in vitro. Immunol. 1996; 87: 339–342.

15. Sorg T B, Cynamon M H. Comparison of four beta-lactamaseinhibitors in combination with ampicillin against Mycobacterium



tuberculosis. J Antimicrobial Chemotherapy 19(1) 59–64. 1987.

16. Huebner R E, Schein M F, Bass J B Jr,. The tuberculin skin test.Clin Infect Dis. 1993; 17: 968–975.

17. Corrah P T. Studies of tuberculosis in The Gambia. OpenUniversity, 1994.

18. Sanchez F O, Rodriguez J I, Agudelo G, Garcia L F. Immuneresponsiveness and lymphokine production in patients withtuberculosis and healthy controls. Infect Immun. 1994; 62:5673–5678.

19. Zhang M, Lin Y, Iyer D V, Gong J, Abrams J S, Barnes P F. T-cell


cytokine responses in human infection with Mycobacteriumtuberculosis. Infect Immun. 1995; 63: 3231–3234.

20. Barnes P F, Lu S, Abrams J S, Wang E, Yamamura M, Modlin R L.Cytokine production at the site of disease in humantuberculosis. Infect Immun. 1993; 61: 3482–3489.

21. Menzies D, Interpretation of repeated tuberculin tests. Am. J.Respir. Crit. Case Med. 1999; 159: 15–21.

22. Smith S M, Malin A S, Lukey P T, Atkinson S E, Content J,Huygen K, Dockrell H M. Characterisation of humanMycobacterium bovis Bacille Calmette-Guerin-Reactive CD8+ T
cells. Infect Immun 1999; 67: 5223–5230.

a longitudinal study of in vitro ifnγ production and cytotoxic t cell responses of tuberculosis...

Documents