Enhancing the immunotherapeutic potential of mycobacteria by transfection with tumour necrosis factor-α

Abstract

In an attempt to enhance the anti-tumour properties of mycobacteria we have developed recombinant forms of Mycobacterium smegmatis which express and secrete biologically active human tumour necrosis factor-α (TNF-α). This was achieved by transfecting M. smegmatis using shuttle plasmids incorporating the cDNA sequence for the human TNF-α mature peptide. In vitro experiments on a panel of human bladder tumour cell lines (EJ18, MGH-U1, RT4, RT112) indicate that our genetically modified mycobacteria are more effective than wild-type at inducing or up-regulating the expression of intracellular adhesion molecule-1 and the secretion of an array of proinflammatory cytokines [interleukin-1 (IL-1), IL-6, IL-8, granulocyte–macrophage colony-stimulating factor]. We have also demonstrated increased adhesion molecule and cytokine expression in response to mycobacteria transfected with vector containing no gene insert. However, this was not as pronounced as that observed following tumour cell stimulation by the TNF-α-transfected strain. In contrast, in three out of four tumour cell lines all M. smegmatis strains were found to down-regulate the secretion of the anti-inflammatory cytokine transforming growth factor-β₁. Our studies have also confirmed that M. smegmatis is a powerful inhibitor of bladder tumour cell growth and revealed that its antiproliferative potency is enhanced by transfecting with human TNF-α and, to a lesser extent, with vector alone. All M. smegmatis strains were effective in the activation of peripheral blood leucocyte cultures. However, no differences were observed in the ability of the TNF-α-transfected, mock-transfected and wild-type mycobacteria to induce tumour cell killing activity. These results suggest that the immunomodulatory effects of M. smegmatis can be enhanced by transfection with vectors which allow the secretion of human TNF-α, thus increasing mycobacterial immunotherapeutic potential.

INTRODUCTION

Live bacillus Calmette–Guérin (BCG), an attenuated strain of Mycobacterium bovis, has been used as an immunotherapeutic agent against many forms of malignancy with varying degrees of success.^1–3 Despite its proven value in treating carcinoma in situ (CIS) of the bladder,⁴ its mechanism of action is not fully understood and reasons for therapy failure in almost 20% of CIS patients remain elusive. However, we do know that the mycobacteria enter epithelial cells, probably by endocytosis⁵ and once inside the bladder wall, they initiate a large local influx of lymphocytes, macrophages, polymorphonuclear cells and natural killer cells.⁶ It is believed that these cells give rise to the majority of cytokine molecules detected in the urine from patients, including interleukin-1 (IL-1), IL-2, IL-5, IL-6, IL-8, IL-10, IL-12, tumour necrosis factor-α (TNF-α) and interferon-γ (IFN-γ).^7–12 However, it has recently emerged that cultured tumour cells themselves, can be induced to secrete IL-1, IL-6, IL-8 and TNF-α in response to mycobacterial stimulation.^13–15

The release of cytokines during immunotherapy is believed to be responsible for the marked changes in tumour cell surface phenotype observed following mycobacterial instillation, especially in major histocompatibility complex (MHC) class II and intracellular adhesion molecule-1 (ICAM-1) expression.^16,17 It is thought that the induction or up-regulation of these molecules may play a role in tumour recognition and destruction. For example, enhanced ICAM-1 expression is believed to favour binding of lymphokine-activated killer (LAK) cells to tumour cells prior to killing,^18–20 while MHC class II expression on tumour cells has been associated with antigen presentation to CD4⁺ T cells²¹ propounding an alternative specific immune response pathway.

A number of workers have begun to employ recombinant technology for the purpose of elucidating the mechanisms behind mycobacterial immunotherapy/vaccination.^22–24 Modifications to the immunostimulatory properties of mycobacterial strains could provide clues to understanding the difference between therapy success and failure and may ultimately provide a means of enhancing the clinical efficacy of BCG therapy.

The aim of this study was to genetically modify Mycobacterium smegmatis, a live, non-pathogenic bacterium, to express human TNF-α. TNF-α was selected because of its known antiproliferative effect against cancer cells in vitro,²⁵ including bladder tumour cells,²⁶ its ability to induce LAK activity²⁷ and to regulate MHC class II and ICAM-1 expression.^19,28,29Mycobacterium smegmatis was selected as it is less pathogenic than BCG and is more easily handled in the laboratory, having a doubling time of 2–6 hr as opposed to 22 hr for BCG. In addition, we have shown it to be almost 200 times more potent with respect to inhibiting bladder tumour cell growth in vitro (unpublished observations).

Here, we present details of gene transfection into mycobacteria using a shuttle plasmid strategy, analysis of the concentration and biological activity of the human recombinant protein and results of in vitro investigations into tumour ICAM-1, human leucocyte antigen (HLA)-DR and cytokine induction and killing by wild-type and recombinant M. smegmatis clones.

MATERIALS AND METHODS

Gene amplification

Primers for amplification of the mature peptide sequence of human TNF-α were designed incorporating appropriate restriction enzyme sites for unidirectional cloning into the multiple cloning site of Escherichia coli–mycobacteria shuttle plasmids pMOD8 and pMOD12gfp. The TNF-α cDNA was amplified using reverse transcription–polymerase chain reaction (RT-PCR) on mRNA from human donor lymphocytes stimulated with 4β-phorbol 12β-myristate 13α-acetate (PMA). Isolation of mRNA was achieved on oligo-dT spin columns (Qiagen Ltd, Crawley, West Sussex, UK).

Vector technology

The pMOD8 and pMOD12gfp shuttle expression vectors used in this study were a gift from Dr M. O'Donnell, Division of Urology, Beth Israel Hospital, Boston, MA. These include separate origins of replication for both E. coli and mycobacteria, an α-antigen signal peptide sequence driven by the heat-shock protein (hsp 60) mycobacterial promoter and the 12CA5 epitope derived from the influenza haemagglutinin (HA) protein which serves as a tag for the detection of the recombinant protein. The pMOD series of vectors also carry a kanamycin resistance gene for selection of transformants. In addition, the pMOD12gfp vector incorporates a second promoter (hsp 70), two multiple cloning sites and a gene encoding the jellyfish (Aequorea victoria) green fluorescent protein (gfp).³⁰ Recombinant gfp fluoresces in ultraviolet illumination and is used as a selectable marker.

Cloning of TNF-α

Digestion of the purified PCR product with appropriate restriction enzymes (Promega UK Ltd, Southampton, UK) was followed by ligation into the shuttle vectors and transformation into competent DH5α cells (Gibco BRL, Life Technologies Ltd, Paisley,UK). The transformants were selected on kanamycin–agar and incorporation of the gene insert was verified by PCR using the original TNF-α primers. Successfully transformed E. coli colonies were grown up overnight in Luria–Bertani (LB) liquid medium containing 30 μg/ml kanamycin (Gibco BRL) and the cultures were maxi-prepped (Maxi-kit tip-500; Qiagen Ltd) to harvest the cloned vector constructs. After confirmation of the TNF-α sequence by diagnostic restriction analysis and automated sequencing, the plasmid constructs were introduced into M. smegmatis by electroporation (easyjecT PLUS; Flowgen Instruments Ltd, Lichfield, Staffs.,UK). Aliquots of M. smegmatis [200 μl at ≈5×10⁸ clony-forming units (CFU)/ml], washed and resuspended in 10% glycerol (Sigma Chemical Co., Poole, Dorset,UK), were transferred to precooled electroporation cuvettes (Flowgen) and 1–2 μg of purified vector construct DNA was added. Transfection reactions using empty, circular pMOD vectors were conducted to produce direct controls for subsequent assays. Electroporation conditions were set at 2·5 kV, 25 μF and 700 Ω, this gave a time constant of ≈17 milliseconds. To the electroporated mycobacteria, 1 ml of Middlebrook 7H9 liquid medium containing 10% Middlebrook albumin–dextrose complex (ADC) enrichment (Difco Laboratories, Detroit, MI) was added and the cell suspension was incubated for 2 hr before plating onto Middlebrook 7H11 agar (Difco) containing 0·5% glycerol, 10% Middlebrook ADC enrichment and 30 μg/ml kanamycin. Following 3 days incubation at 37°, transfectants were selected by their antibiotic resistance and their fluorescence under ultraviolet illumination (pMOD12 series).

Analysis of recombinant protein

A number of kanamycin-resistant M. smegmatis colonies were used to establish liquid cultures and growth was monitored by absorbance at 600 nm. Supernatant samples from mid-log phase cultures were analysed using sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and Western blotting techniques, probing with monoclonal anti-[HA] (Boeringer Mannheim, Sussex, UK) and anti-human (hu)TNF-α antibodies (R & D Systems Europe Ltd, Abingdon, Oxon, UK) and enzyme-linked immunosorbent assay (ELISA; Quantikine HS immunoassay kit; R & D Systems) for the presence of huTNF-α. An L929 bioassay employing recombinant huTNF-α standards (R & D Systems) was subsequently undertaken to assess the biological activity of the recombinant protein.³¹

Mycobacterial culture

Two transfected M. smegmatis clones, referred to as pMOD8–TNF-α and pMOD12gfp–TNF-α, demonstrated to be the most efficient secretors of recombinant human TNF-α, were chosen and grown in 500 ml Middlebrook 7H9 medium containing 0·2% glycerol, 0·05% Tween-80 (Sigma) and 30 μg/ml kanamycin supplemented with ADC enrichment. Empty vector (mock-transfectants) and wild-type control M. smegmatis cultures were established concurrently. Wild-type cultures were grown in the absence of antibiotics. Large-scale cultures were incubated at 37° for 2–5 days and the mycobacterial cells were then washed to remove the 7H9 medium and antibiotics in RPMI-1640 medium supplemented with 25 mm HEPES, 2 mm l-glutamine, 1 mm sodium pyruvate and 5% heat-inactivated fetal calf serum (FCS) (all from Sigma). Dilutions of the resuspended cultures were prepared for CFU determination, 100 μl of each dilution was plated onto 7H11 agar and the resultant colonies were counted after 3 days incubation at 37°. The number of CFU/ml was calculated in each case. The remaining M. smegmatis cells were frozen in aliquots at −70°.

Cell culture

Four human transitional cell carcinoma lines, EJ18, MGH-U1, RT4 and RT112, a gift from Dr J. Masters, Institute of Urology, London, were employed as an in vitro model for bladder cancer. The cell lines were maintained in complete RPMI-1640 medium without antibiotics at 37° in a humidified atmosphere of 5% CO₂ in air. Adherent monolayers were routinely recovered by trypsinization with 0·5 mg/ml trypsin, 0·2 mg/ml tetrasodium ethylenediamine tetraacetic acid (EDTA; Sigma) and manipulations were carried out in a class II microbiological safety cabinet (Envair Ltd, Rossendale, Lancs, UK).

Measurement of cytokine induction

The effects of wild-type and recombinant M. smegmatis on the expression of cytokines by cultures of bladder tumour cells and peripheral blood mononuclear cells (PBMC) were investigated. Supernatant samples from naive cell populations and cultures incubated with M. smegmatis clones were collected at 72 hr (tumour cells) or at 6 days (PBMC) and analysed for TNF-α concentration and the presence of other cytokines (IL-1β, IL-2, IL-6, IL-8, GM-CSF and RANTES) using commercially available ELISA assays (R & D Systems).

Cytofluorometric analysis

Single-colour immunofluorescence was conducted on bladder tumour cells stained with monoclonal antibodies to human ICAM-1 (BBIG-I1; R & D Systems) or HLA-DR (fluorescein isothiocyanate-conjugated CR3/43; DAKO Ltd, Ely, Cambridge, UK). The cells, incubated with an initial concentration of 5×10⁴ CFU/ml mycobacteria, recombinant TNF-α (positive control), or with medium alone, were harvested at 24, 48 and 72 hr, washed in phosphate-buffered saline, pH 7·4, supplemented with 1% heat-inactivated FCS and 0·01% sodium azide (all from Sigma) and incubated with optimal concentrations of primary antibody for 30 min at 4°. The binding of the unconjugated ICAM-1 antibody was detected by incubating the washed cells with goat anti-mouse IgG (F(ab′)₂-fluorescein isothiocyanate-conjugated; Sigma) at 4° for a further 30 min and controls were incubated with secondary antibody alone. Following antibody incubations, the cells were washed and resuspended in washing buffer containing 1% formaldehyde (Fisher Scientific UK, Loughborough, Leics., UK). Fluorescence was measured on an EPICS-XL flow cytometer (Coulter, Luton, UK). Live cells were gated within a window and 5000 events were accumulated using logarithmic amplification of fluorescence intensity.

Proliferation assay

Cell proliferation was monitored by measuring tritiated thymidine incorporation into the DNA of replicating cells. The four bladder tumour cell lines were seeded at 1000–5000 cells/well in 96-well flat-bottomed plates (Costar UK Ltd, High Wycombe, Bucks., UK). Optimum concentrations of wild-type or transfected mycobacteria (5×10⁴ CFU/ml) were added to mid-log phase cultures. At 16 hr prior to the harvest time-point, 0·5 μCi methyl [³H]thymidine (Amersham International plc, Little Chalfont, Bucks., UK) was added to each well under sterile conditions. The cells were trypsinized and recovered at 24, 48 and 72 hr with a semiautomatic cell harvester (Skatron, Lier, Norway). Filter discs were dried and counted using a scintillation analyser (Tri-Carb 1900CA; Canberra Packard, Pangbourne, Berks., UK).

Cytotoxicity assay

Lysis of bladder tumour cells by M. smegmatis-activated killer cells (designated as SMAK cells) was investigated in a chromium (^51*sref51*Cr) – release assay. Effector cells were induced by stimulating PBMC, isolated from normal donor buffy coat by density gradient centrifugation on Lymphoprep (Nycomed, Oslo, Norway), with 2×10² CFU/ml wild-type or recombinant mycobacteria over a period of 4–6 days. LAK cells were generated concurrently by incubating PBMC from the same donor with 1000 U/ml recombinant IL-2 (PeproTech EC Ltd, London, UK). Target cells were labelled by adding Na₂^51*sref51*CrO₄ (Amersham) at 100 μCi per 10⁶ tumour cells in complete RPMI-1640 medium and incubating under normal cell culture conditions for 1 hr. The cells were then washed three times in complete medium, resuspended at a concentration of 5×10⁴ cells/ml and seeded into 96-well V-bottomed microtitre plates (Costar) at 100 μl per well. Effector cells were added to achieve a previously determined optimum effector:target ratio of 40:1 producing a final volume of 200 μl. After centrifugation at 100 g for 30 seconds, the plates were incubated for 4 hr, centrifuged again at 100 g for 5 min and finally 100 μl of supernatant from each well was transferred into tubes for measurement in a gamma-counter.

Results are calculated as percentage specific toxicity using the formula:

Spontaneous release was measured by determining the counts in supernatants from target cells cultured in medium alone and total release was measured by determining counts from target cells lysed with 1% Triton X-100 (Sigma)

RESULTS

Expression and secretion of biologically active recombinant TNF-α

The secretion of TNF-α was initially verified by probing nitrocellulose blots of SDS–PAGE-separated culture supernatants with antibodies to the 12CA5 epitope sequence tag and to human TNF-α. These studies revealed the presence of bands at 17 000 MW in transfected culture supernatants corresponding to the human recombinant protein. ELISA analysis confirmed levels of TNF-α in the same culture supernatant samples ranging from 13 to 80 ng/ml following 7 days of culture. No TNF-α was detected by either technique in supernatants from bacterial cultures transfected with empty vector. Higher secretion levels were found to be produced by clones containing the single promoter/cloning site (pMOD-8) vector construct. Supernatant samples from the most efficient TNF-α-secreting M. smegmatis clone (pMOD8–TNF-α), seeded at 5×10⁴ CFU/ml in RPMI medium without antibiotics, were collected at 24, 48 and 72 hr and found to contain biologically active human TNF-α which increased in concentration from 230±65 pg/ml (1·6±0·5 IU/ml) to 9·8± 0·7 ng/ml (69±14 IU/ml) over 3 days.

The effects of wild-type and recombinant M. smegmatis on tumour cells in vitro

In previous studies we have shown that BCG directly effects human bladder tumour cell lines in vitro, inducing and/or up-regulating ICAM-1 expression,¹⁷ the secretion of cytokines^13–15 and inhibiting tumour cell growth.³² We have therefore extended these investigations to study the effects of wild-type and recombinant M. smegmatis on these parameters.

Cytokine induction by wild-type and transfected mycobacteria

Table 1 presents the concentrations of TNF-α, IL-1β, IL-6, IL-8, GM-CSF, RANTES and TGF-β₁ detected in supernatants from 3-day tumour cell cultures incubated alone or with 5×10⁴ CFU/ml wild-type, vector-transfected or TNF-α-transfected M. smegmatis.

Table 1.

Cytokine expression by tumour cell cultures incubated for 72 hr with medium alone or with 5×10⁴ CFU/ml wild-type or recombinant M. smegmatis clones

TNF-α concentrations in supernatants from MGH-U1, RT4 and RT112 incubated with TNF-α-transfected M. smegmatis reflected levels obtained from 5×10⁴ CFU/ml of the TNF-α-transfected clone grown alone without antibiotic selection. This suggests that the TNF-α was secreted exclusively from the recombinant mycobacteria. For EJ18 however, a concentration almost 2·5 times higher indicates an induction of TNF-α protein from the tumour cells themselves. Low concentrations of TNF-α were also found in supernatants from EJ18 and RT112 cell lines incubated with M. smegmatis transfected with empty vector. As no TNF-α was detected from mock-transfected M. smegmatis cultured in isolation or from tumour cells cultured alone or with wild-type clones, it appears TNF-α is released by the tumour cells in response to the presence of the vector.

No significant induction of IL-1β (except RT4 cells incubated with TNF-α-transfectant) or RANTES was seen, however, a distinct pattern emerged for IL-6, IL-8 and GM-CSF expression. We were able to show greater induction/up-regulation of expression from cell lines infected with recombinant M. smegmatis strains compared with wild-type and levels were always higher in supernatants from cells subjected to the TNF-α-transfectants. A reversal of this trend was demonstrated for TGF-β₁, with three out of four cell lines infected with all mycobacteria preparations producing lower levels of TGF-β₁ compared to tumour cells cultured in medium alone. It should be stressed that similar effects were reproducibly observed on a minimum of two to five separate occasions depending on the cytokine.

Induction of ICAM-1 and HLA-DR expression

Previous studies from our laboratory have reported the constitutive surface expression of ICAM-1 by the four bladder tumour cell lines to be in the range 4–80% positive cells,¹⁹ with the greater levels in higher grade tumours dependent upon cell density and autocrine factors.³³ The current study revealed up-regulation of ICAM-1 expression on all four cell lines following infection with the TNF-α-transfected M. smegmatis (Fig. 1). Wild-type and vector-transfected mycobacteria were shown to be less effective at augmenting cell surface ICAM-1 levels in every case and failed to induce any expression on the RT4 cell line. Additional experiments on the MGH-U1 cell line indicated that the induction of ICAM-1 by TNF-α-transfected M. smegmatis could be largely, though not completely, inhibited by inclusion in the culture of excess neutralizing antibody to TNF-α (Fig. 2).

Figure 1.

The effect of wild-type and recombinant forms of M. smegmatis on ICAM-1 expression. Bladder tumour cells were stained following 24, 48 and 72 hr incubation with 5×10⁴ CFU/ml mycobacteria and ICAM-1 levels were compared with cells incubated with medium alone. Results show significantly greater up-regulation of ICAM-1 expression by the TNF-α-transfectant compared with wild-type and vector-transfectant. A typical experimental outcome is presented for each cell line from four similar repeat determinations and means±standard deviations are calculated from triplicate assessments.

Figure 2.

The quenching of ICAM-1 expression by anti-huTNF-α antibody. MGH-U1 tumour cells were incubated for 48 hr with medium, TNF-α-transfected or mock-transfected M. smegmatis clones both with and without the addition of 5 μmg/ml anti-huTNF-α. The mean percentage ICAM⁺ cells±standard deviation are calculated from triplicate determinations.

Incubation of the RT4 cell line with TNF-α-secreting M. smegmatis also produced an increase in MHC class II expression (2·6-fold) which was not achieved with either wild-type or mock-transfected species (data not shown).

Anti-proliferative effects of wild-type and transfected mycobacteria

Our early observations revealed the antiproliferative potency of M. smegmatis to be up to 200 times that of BCG. Using initial concentrations of 5×10⁴ mycobacterial CFU/ml, a gradual reduction in tumour cell proliferation could be demonstrated over 72 hr. This concentration was employed in all subsequent studies.

Results for all four transitional cell carrcinoma cell lines show a similar trend. Both TNF-α-transfected and mock-transfected M. smegmatis were more effective than wild-type mycobacteria in inhibiting tumour cell replication. The TNF-α transfectant was slightly, but consistently, more effective than the mock-transfectant in this regard (Fig. 3).

Figure 3.

A comparison of wild-type and recombinant mycobacterial antiproliferative potency measured by incorporation of [³H]thymidine by bladder tumour cell lines. Cells were incubated with medium alone or with 5×10⁴ CFU/ml wild-type, TNF-α-transfected or vector-transfected M. smegmatis over 72 hr. Results show TNF-α-transfected M. smegmatis clones are significantly more effective than wild-type and marginally more effective than vector-transfected strains at arresting tumour cell growth. A typical result is presented for each cell line showing the mean±standard deviation of six replicate wells. Each experiment was performed on four separate occasions with reproducible results.

The effects of SMAK cells

Due to its remarkable potency and rapid replication rate, which resulted in overgrowth of cultures accompanied by poor survival of lymphocytes, initial concentrations of 2×10² CFU/ml M. smegmatis were chosen to study the activation of PBMC over 4–6 days. Resultant SMAK effector cells produced specific cytotoxicity values of 21–43% (day 5) in the four bladder cell lines, compared with 55–87% for LAK cells. However, no significant enhancement of cytotoxicty was observed with recombinant mycobacteria compared with wild-type (Table 2).

Table 2.

The induction of activated killer cells (SMAK cells) by wild-type and transfected M. smegmatis

No cell surface phenotypic differences between wild-type and recombinant SMAK cells were revealed by flow cytometry using monoclonal antibodies to CD3, CD4, CD8, CD25, HLA-DR and γδ. In fact, the number of CD25⁺ and HLA-DR⁺ cells remained unchanged from non-activated PBMC populations. All M. smegmatis-activated killer cells however, demonstrated an increase in CD8⁺ cells.

Cytokine measurement showed notable induction of IL-6, IL-8 and RANTES but no significant difference was observed between the WT-SMAK and recombinant-SMAK cells. IL-2 production was not detected from any of the SMAK cell cultures.

DISCUSSION

Mycobacterial strains incorporating the genetic sequence of human TNF-α have been engineered using an extra-chromosomal plasmid system. TNF-α conforms to all predictable criteria for efficient mycobacterial expression.²⁴ It contains no N-glycosylation sites, its protein sequence gives rise to a single disulphide bond³⁴ and it appears non-toxic to the mycobacterial host. ELISA and bioassay analysis of supernatants from TNF-α-transfected M. smegmatis cultures have confirmed high levels of bioactive recombinant protein secretion.

An in vitro model based on four human bladder transitional cell carcinoma cell lines was chosen to examine differences in immunomodulatory activity between wild-type and recombinant M. smegmatis clones. We have demonstrated significant differences in tumour response to wild-type and transfected strains. TNF-α-secreting mycobacteria proved to be reproducibly more effective in augmenting/inducing proinflammatory cytokine secretion (IL-1β, IL-6, IL-8, GM-CSF) and cell surface ICAM-1 expression. We have also shown enhancement of mycobacterial antiproliferative properties as a result of TNF-α transfection but failed to observe any differences in ability to induce SMAK activity. Interestingly, our experiments also reveal that mock-transfected mycobacteria possess greater immunostimulatory properties than the wild-type.

These studies with TNF-α-transfected M smegmatis extend previous observations on the enhanced immunomodulatory activities of mycobacteria transfected with other human and rodent cytokine genes.^22–24

Experiments demonstrating increased up-regulation of cytokine expression by TNF-α-transfected mycobacteria appear to confirm documented views concerning the role of TNF-α in immune modulation. Most of the cytokines investigated here (IL-1, IL-6, IL-8, GM-CSF) are known to be induced by TNF-α or at least stimulated by the same exogenous or endogenous signals.³⁵ It is not unreasonable to assume that the enhanced ability of TNF-α-transfected M. smegmatis to induce cytokines such as IL-8 (which enhances chemotaxis and activation of neutrophils) and GM-CSF (which increases production of granulocytes and macrophages) should endow it with superior anti-tumour properties in vivo. Of additional potential benefit in this regard are the observations that M. smegmatis (transfected or otherwise) down-regulates the production of TGF-β₁ by tumours. This cytokine is overproduced in many tumours and may provide growth advantage due to its stimulation of extracellular matrix formation and its local immunosuppressive effect on cytotoxic T cells and natural killer cells.^36,37

It has been previously reported that cytokines found in the urine of patients receiving BCG immunotherapy augment ICAM-1 and HLA-DR expression on bladder tumour cells.²⁶ Our results show wild-type M. smegmatis to be an extremely effective up-regulator of ICAM-1, verifying the suitability of this strain as a model for BCG. In accord with previous observations, augmentation of ICAM-1 by the TNF-α-secreting strain is likely to promote increased recognition of the tumour cells by activated leucocytes via ICAM-1–LFA-1 interactions.^18–20

An increase in MHC class II expression on RT4 cells in response to TNF-α-transfected M. smegmatis suggests secretion of the recombinant protein may engage a second pathway involving presentation of mycobacteria-derived antigens by the MHC-expressing tumour cells producing a specific T-cell response.²¹

The antiproliferative activities of mycobacteria themselves and of a number of immune system-derived cytokines, including TNF-α, have been described in cultured bladder tumour cells.^26,32 Wild-type M. smegmatis exhibited remarkably potent inhibitory effects on tumour cell growth. Early experiments demonstrated the effect of 5×10⁴ CFU/ml M. smegmatis to be equivalent to ≈1×10⁷ CFU/ml wild-type BCG and only modest improvements in cytostatic ability could be attributed to TNF-α secretion.

In all of the experiments described, there is evidence of an improvement in immunological activity of mycobacteria transfected with vector alone. The reason for this is unclear but modification of growth characteristics as a consequence of the transfection protocol has been ruled out as no differences in replication rate between wild-type and recombinant species were apparent in the absence of antibiotic selection. However, it is not unexpected that the vector-transfectant generates greater adhesion molecule expression and antiproliferative activity given the higher levels of cytokines induced from tumour cells cultured with mock-transfected mycobacteria compared with wild-type. At the present time we can only speculate on why mock-transfected M. smegmatis is more active than wild-type with respect to a number of the effects we have investigated. One remote possibility is that the use of vectors containing hsp promoters may result in the induction or enhanced production of hsp by the tumour cells themselves which in turn leads to enhanced immunomodulatory and cytotoxic effects, though this remains to be tested. Neverthless, these experiments highlight the importance of including such controls in all experiments utilizing recombinant micro-organisms.

The tumoricidal activity of PBMC coincubated with BCG has been described and termed as the BCG-activated killer (BAK) phenomenon.³⁸ Our experiments have shown marked SMAK cell cytotoxicity against bladder tumour cells and experiments using phytohaemagglutinin-stimulated PBMC (blasts) from allogeneic and syngeneic sources as target cells, demonstrated the tumour specificity of SMAK cells. No killing of the stimulated blast cell populations from either source was observed, suggesting the effector cells were not indiscriminantly cytotoxic.

Investigations into changes in cell phenotype following M. smegmatis activation showed an expansion of CD8⁺ cells, a characteristic in BAK cell development³⁹ but no increases in IL-2 receptor or MHC class II expression.

Failure to demonstrate any differences between wild-type M. smegmatis-induced (WT-SMAK) and recombinant M. smegmatis-induced (TNF-α-SMAK/mock-SMAK) killer cells may be attributed to subactive levels of TNF-α secretion. As a consequence of the extremely potent cytolytic properties of M. smegmatis, colony numbers used were probably insufficient to produce effective levels of TNF-α. Also, in the absence of selective antibiotics, loss of the non-integrated vector from the transfectants was rapid (70% drop-out over 3 days).

The production of IL-6, IL-8 and RANTES but not IL-2 from M. smegmatis-stimulated PBMC indicates that M. smegmatis is not inducing T-cell activation. These results are in conflict with current theories of a BCG T helper type 1 (Th1)-type activation pathway⁴⁰ and throws doubt on the suitability of M. smegmatis as a substitute for BCG in these experiments. Whether or not this failure to induce T-cell activation is also a reflection of the extremely small numbers of M. smegmatis organisms (2×10² CFU/ml) used in these studies remains to be established.

Our results illustrate the immunomodulatory potential of genetically modified mycobacteria. Production and secretion of cytokines at the primary tumour site, thereby mimicking physiological expression, has advantages over systemic administration in that comparatively small concentrations of these extremely toxic molecules are introduced locally into the patient. We have presented evidence of enhancement of the in vitro immune response and tumour response to mycobacteria secreting TNF-α. This may be of significant clinical benefit in the improvement of mycobacterial immunotherapy protocols and animal models are currently being developed to determine in vivo therapeutic prospects.

Abbreviations

ADC

albumin dextrose complex

BAK

bacillus Calmette–Guérin (BCG)-activated killer (cell)

hsp

heat-shock protein

GM–CSF

granulocyte–macrophage colony-stimulating factor

ICAM-1

intracellular adhesion molecule-1

LB

Luria–Bertani (medium)

SMAK

Mycobacterium smegmatis-activated killer (cell)

TCC

transitional cell carcinoma

TGF-β

transforming growth factor-β