This article is mentioned in:

Introduction

Cancer is a complex disease that requires the use of different methods and treatment modes to cure. Surgery, radiotherapy and chemotherapy, alone or in combination, are currently the prevalent treatment modalities (1). However, there is an urgent need for targeted therapies, which may provide more curative prospects.

Claudins (CLDNs) are a family of 17–27 kDa integral membrane tight junction (TJ) proteins that pass through the paracellular space in epithelial and endothelial tissues (2). The CLDN protein structure consists of a cytoplasmic N-termini, a C-termini and two extracellular loops. One of the loops interacts with CLDNs on adjacent cells to form an adhesive structure with other TJ proteins (3). The other extracellular loop in CLDN3 and CLDN4 serves as a binding site for Clostridium perfringens (C. perfringens) enterotoxin (CPE) (4).

As altered CLDN expression is frequently found in cancer cells, the hypothesis of a correlation between altered CLDN expression and cancer has been suggested (5). The loss of CLDNs and other TJ proteins in cancer has been interpreted as a mechanism for the loss of cell adhesion which is an important step in cancer progression and metastasis. However, many CLDNs, such as CLDN3 and CLDN4, are typically upregulated in many types of cancer, such as ovarian, breast, prostate, colon and pancreatic cancer, suggesting that they may have a positive effect on tumourigenesis and may lead to an increase in invasion, motility and cell survival (5). The potential value of CLDN in cancer therapy has been the subject of a number of studies (6) and is based on the fact that CLDNs are expressed at the cell surface and contain two extracellular domains that serve as potential target sites.

CPE is commonly associated with C. perfringens type A food poisoning. CPE is a single polypeptide of 35 kDa, which, upon binding to its receptors, causes cytolysis through its effects on membrane permeability (5,7). Both CLDN3 and CLDN4 are receptors for CPE. The intra-tumoural administration of CPE in cancer cells has been shown to result in tumour regression concomitant with a large degree of tumour necrosis (5). It was first demonstrated by Michl et al that the treatment of xenograft pancreatic cancer cells with CPE, led to a significant reduction in tumour burden, accompanied by necrosis, in CLDN4-expressing tumour cells (8).

Pseudomonas aeruginosa exotoxin A (ETA) functions by binding to nicotinamide and releasing the adenine dinucleotide (ADP-ribose) in mammalian cells. This ADP-ribose unit stays attached to ETA and is transferred to elongation factor-2 (EF-2), a protein involved in the translation and elongation of proteins. This, in turn, blocks protein synthesis in host cells, causing damage to target tissues (9). In this study, we demonstrate that HN5 head and neck squamous carcinoma cells overexpress CLDN4 and that targeting the CLDN4 receptor by CPE-ETA’ (an immunotoxin created by fusing the c-terminal CLDN4-binding domain of CPE to the ETA domain) efficiently and specifically kills HN5 cells. Furthermore, our findings show that CPE-ETA’ can be expressed and secreted by the oncolytic bacterial strain, Clostridium ghonii (C. ghonii), and that the secreted protein is potent against CLDN4-expressing cells.

Materials and methods

Cell culture conditions

The cell lines used in this study were the HN5 human head and neck squamous carcinoma, MCF-7 breast ductal carcinoma, A549 non-small cell lung cancer, MRC-5 normal foetal lung fibroblast, HT29 and HCT116 colon cancer, HeLa cervical cancer and Huh-7 hepatocarcinoma cells. The cancer cells were cultured in completed medium consisting of 500 ml DMEM (Gibco), 10% FBS, 12.5 ml HEPES buffer solution (1M) and 1 ml of penicillin (5,000 U) and streptomycin (5,000 μg) antibiotic mixture. Culture flasks were placed in a sterile tissue culture incubator under a humidified atmosphere at 37°C and 5% CO2. Tissue culture was performed by routine procedures (10). All cell lines were purchased from the American Type Culture Collection, apart from HN5 which was kindly provided by Dr Hong-Jian Zhu, University of Melbourne, Melbourne, Australia.

Confirmation of CLDN4 expression in cancer cell lines

Cells were seeded at appropriate densities and were cultured until 80–90% confluency. Subsequently, the cells were washed three times in PBS, scraped, centrifuged and resuspended in 100 μl of cell extraction buffer (Invitrogen) with 1 mM phenylmethanesulfonylfluoride (PMSF) and protease inhibitor on ice for 30 min while vortexing every 10 min. The lysate was clarified by centrifugation at 13,000 rpm at 4°C and the supernatant stored at −80°C until further use. Total protein concentration was measured by spectrophotometry using the DC protein assay kit (Bio-Rad), and equal amounts of proteins were loaded onto SDS-PAGE gels for western blot analysis.

For qPCR analysis, RNA isolated from the cultured cells was converted to cDNA using SuperScript III (Invitrogen). Primers used for qPCR were as follows: CLDN4 forward, 5′-AGT GCA AGG TGT ACG ACT CGC T-3′ and reverse, 5′-CGC TTT CAT CCT CCA GGC AGT T-3′. GAPDH and β-actin were used as the internal reference genes.

SDS-PAGE and western blot analysis

For SDS-PAGE, the following protein ladders were used: Precision Plus Protein Dual Color Standards (Bio-Rad), PageRuler Plus Prestained Protein Ladder (Fermentas). For western blot analysis, the primary antibodies used were: anti-His antibody, anti-CLDN4 antibody and anti-α-tubulin antibody. The secondary antibody used was goat-anti-mouse IgG antibody. SDS-PAGE and western blot analysis were performed according to standard procedures (11).

Bacterial strains and plasmids

The properties of the bacteria and plasmids used in this study are listed in Table I.

Table I Bacterial strains and plasmids.

Table I

Bacterial strains and plasmids.

Strain or plasmid	Relevant characteristics	Source
Plasmids
p10His-cCPE-ETA’	Amp, T7, N-His-tag, MCS	(12)
pMTL-555	repL, traJ, ermB, fac2	(33)
pMTL-10His-cCPE-ETA’	repL, traJ, ermB, fac2	This study
Strains
E. coli
DH5a	Φ80dlacZΔM15, Δ(lacZYA-argF)U169	Promega
CA434	HB101 carrying the IncPβ conjugative plasmid, R702	(33)
BL21 A1	F-ompT hsdSB(rB-, mB-) gal dcm araB::T7RNAP-tetA	Invitrogen
Clostridia
C. ghonii		This study

Construction of Clostridial CPE-ETA’ fusion vectors

To construct pMTL-10His-cCPE-ETA’, p10His-cCPE-ETA’ (12) was used as the template for PCR. The forward primer SfiI_CPE (5′-GAG GGC CCA GCC GGC CCA TCA TCA TCA TCA TCA TC-3′) and the reverse primer SmaI_ETA (5′-GTC CCG GGA GTT ACT TCA GGT CCT CGC-3′) were used to amplify 10His-cCPE-ETA’ which incorporated the SfiI and SmaI sites, respectively. PCR products were amplified using Phusion High-Fidelity DNA Polymerase (Finnzymes) to minimise sequence errors. Finally, amplicons were excised from gels and cloned into the Clostridial shuttle vector, pMTL-555, using the restriction sites, SfiI and SmaI. The fusion protein is preceded by a secretion signal which is cleaved upon extracellular export of the protein. The recombinant plasmid pMTL-10His-cCPE-ETA’ was verified by sequencing.

Conjugal transfer of plasmids into C. ghonii

For conjugation Escherichia coli (E. coli) cultures were grown aerobically at 37°C, while Clostridium cultures were grown anaerobically at 37°C. After 25 μl spots of C. ghonii overnight cultures were absorbed by HI agar (without antibiotic), CA434 donor cell suspensions (E. coli strain able to transfer the pMTL-555 vector to Clostridium via its helper plasmid capabilities) with recombinant vectors were spotted (25 μl) on these Clostridium spots and grown at 37°C, overnight in anaerobic conditions. To select Clostridia which had taken up the recombinant plasmid, the spots were spread on HI agar plates selected by 10 μg/ml erythromycin, 250 μg/ml cycloserine and 10 μg/ml polymyxin B. The plates were grown in anaerobic conditions until colonies were visible. To further verify that the Clostridium had taken up the plasmids, 15 colonies were selected from these plates and each colony was cultured under aerobic and anaerobic conditions, respectively. Since Clostridium can only grow under strict anaerobic conditions, there should only be growth under anaerobic conditions.

Preparation of Clostridial secreted proteins

Clostridia were cultured in 30 ml of HI medium with erythromycin 10 μg/ml and D-cycloserine 250 μg/ml overnight at 37°C under anaerobic conditions. The cultures were spun at 5,000 × g and the supernatant (medium) was filtered through a 0.2-micron Millex-HV Syringe-driven filter unit. The filtered medium was concentrated using an Amicon ultra centrifugal filter (ultracel-30k), at 5,000 rpm for at least 30 min at 4°C. The proteins were washed by PBS and concentrated 30-fold to 500 μl. The DC protein assay kit (Bio-Rad) was used to measure protein concentrations.

Expression of CPE-ETA’ in E. coli

To produce 10His-cCPE-ETA’, the p10His-cCPE-ETA’ plasmid was used as previously reported by us (12). After transfection into E. coli, protein purification was performed using a Ni-NTA Fast Start kit (Qiagen) following the manufacturer’s instructions. Purified protein was desalted and concentrated in PBS using an ultrafiltration filter (Amicon Ultra-15–30 kDa cut-off). Finally, protein samples were stored at −80°C in 20% glycerol PBS and when required protein concentrations were measured using the DC protein assay kit (Bio-Rad).

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) cytotoxicity assay

Cancer cells were seeded at a density of 1×104 cells per well of a 96-well plate. After overnight incubation, recombinant proteins were added to the cells and incubated for 48 h. To ascertain the specificity of CPE-ETA’ for the CLDN4 receptor, HN5 cells were incubated with anti-CLDN4 antibodies (blocking of the CLDN4 receptor) for 1 h prior to the addition of CPE-ETA’. MTT assays were performed according to a standard procedure (13). The absorbency was measured by using a POLARstar Omega spectrophotometer from BMG Labtech. The results were converted to percentage proliferation compared to the control PBS group.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 5 software. The significance level was 0.05 (P<0.05) and a Student’s t-test was used to analyse the data. Experiments were performed three times and the data are presented as the means ± standard error of the mean.

Results

Identification of high levels of CLDN4 expression in cancer cell lines

Cell lysates from cancer cell lines were separated by SDS-PAGE, transferred onto PVDF membranes and probed with anti-human CLDN4 and α-tubulin antibodies. Subsequent analysis revealed that CLDN4 expression in the MCF-7, HN5, HT29 and HCT116 cancer cell lines was significantly upregulated (Fig. 1A). Furthermore, a weak expression was observed in the A549 cells, while CLDN4 expression was undetectable in the HeLa, MRC-5 and Huh-7 cell lines. Therefore, for all subsequent proliferation experiments, HeLa cells were used as the negative control. In addition, even protein loading was confirmed by the expression of the housekeeping protein, α-tubulin. The results from real-time PCR analysis of the CLDN4 transcript were consistent with those obtained from western blot analysis (Fig. 1B).

Figure 1 Detection of CLDN4 expression in cancer cells. (A) Proteins probed with anti-CLDN4 antibody followed by secondary antibody (goat-anti-mouse). Equal protein loading of selected cancer cell lines (MCF-7, HN-5, A549, HeLa, MRC-5, HT-29, HCT116 and Huh-7) was confirmed by the housekeeping gene, α-tubulin. (B) Real-time expression analysis of CLDN4 in cancer cell lines.

CPE-ETA’ protein expression and purification

His-tag purification was employed to purify CPE-ETA’ (Fig. 2A) expressed in E. coli. Analysis of the purified protein by SDS-PAGE revealed a protein of the expected size (58 kDa) in elution 1 and elution 2 (Fig. 2B). The purification step yielded 3 mg/ml of protein and a total of 6 mg of protein was isolated.

Figure 2 Construction and purification of 10His-cCPE-ETA’. (A) p10His-cCPE-ETA’ was constructed by Yao et al(12) in our laboratory. The backbone is pBM1.1 (Kanr) (21) and is induced by IPTG and L-arabinose for protein expression. (B) Protein purified by His-tag purification. Following electrophoresis, proteins were revealed by staining with Coomassie blue. M, page ruler plus prestained protein ladder; L, total cell lysate; F, column flowthrough; W1 and W2, wash one and two; E1 and E2, elution one and two.

Efficacy of purified CPE-ETA’

The toxicity of purified CPE-ETA’ was examined by employing MTT proliferation assays (13). MTT is converted to formazan by living cells and can be detected by spectrophotometric quantification. Purified protein was diluted from 0 to 40,000 ng/ml to give a dose response and subsequent calculation of the 50% inhibitory concentration (IC50) values (Fig. 3). PBS with 20% glycerol was used as the no-drug control. DMEM medium only (without cells and proteins) was used as the blank for MTT assay. The results showed that CPE-ETA’ was very effective against HN5, MCF-7, HT29 and HCT116 cells with an IC50 between 8–160 ng/ml (Fig. 3). Furthermore, the A549 cells showed a moderate sensitivity against the targeted toxin (IC50 ∼270 ng/ml), while the CLDN4-negative cell line, HeLa, had an IC50 of ∼17,000 ng/ml.

Figure 3 MTT assay was carried out on selected cancer lines. 10His-cCPE-ETA’ protein toxicity tested in the cancer cell lines, A549, MCF-7, HN-5, HT-29, HTC116 and HeLa. Protein concentrations were from 0 to 40,000 ng/ml. Each concentration of drug was repeated four times. Data represent the means ± standard error. 10His-cCPE-ETA’ exhibited potent toxicity towards CLDN4-positive cells but had little effect on CLDN4-negative cells. HN5 and MCF-7 cells were the most sensitive (IC50 ∼10 ng/ml), while HeLa cells were unaffected at concentrations >10 μg/ml.

Specificity of CPE-ETA’ for CLDN4 receptors

To examine the specific biniding of CPE-ETA’ to the CLDN4 receptor, HN5 cells were incubated with an anti-CLDN4 antibody prior to the addition of CPE-ETA’. It was found that CPE-ETA’ had no effect on cells pre-treated with the antibody compared to cells that were not treated with antibody (Fig. 4). It was also found that the CLDN4 antibody alone had no effect on the proliferation of the cells.

Figure 4 Anti-CLDN antibody abrogation of CPE-ETA’ binding capacity. (A) MTT assays were carried out to determine the binding specificity of CPE-ETA’ for CLDN4. HN5 cells were seeded overnight in culture plates and treated as follows: PBS, no drug control; anti-CLDN4 (100 ng/ml); CPE-ETA’ (100 ng/ml); anti-CLDN4 (100 ng/ml) plus CPE-ETA’ (100 ng/ml). (B) Microscopic analysis of HN5 cells clearly show that pre-treatment with anti-CLDN4 antibody abrogates the effects of CPE-ETA’.

Construction of CPE-ETA’

Fig. 5 shows the map of CPE-ETA’ constructed in the pMTL-555 backbone. pMTL-555 allows for the expression of proteins in Clostridium under the fac2 promoter. This plasmid is compatible in both E. coli and Clostridium. Furthermore, it contains elements for the conjugal transfer of plasmids from E. coli to Clostridium. 10His-cCPE-ETA’ was amplified from p10His-cCPE-ETA’ by PCR, SfiI and SmaI restriction enzyme sites (Fig. 5A) were incorporated into the amplicon for subsequent cloning into pMTL-555. The PCR fragment was cloned into pMTL-555 (digested by SfiI and SmaI) to produce pMTL-10His-cCPE-ETA’ (Fig. 5A). Sequence analysis was used to verify the correct colonies in pMTL-555.

Figure 5 Construction of pMTL-10His-cCPE-ETA’ and expression in Clostridium ghonii. (A) Indicated are restriction sites and important motifs highlighted are the promoter, the N-terminal secretion signal (SS), 10His, cCPE, and ETA’. Cloning of 10His-cCPE-ETA’ into pMTL-555 was achieved with SfiI and SmaI. Identification of transconjugants and detection of 10His-cCPE-ETA’ expression in Clostridium. (B) Bacterial cultures were replica-patched and grown under aerobic and anaerobic conditions at 37°C overnight. Growth was observed under anaerobic conditions only, indicating the presence of Clostridium. (C) Western blot analysis of 10His-cCPE-ETA. The primary antibody used was anti-His antibody; the secondary antibody used was goat anti-mouse IgG antibody. Lanes 1 and 4, Clostridial colonies containing pMTL-10His-cCPE-ETA; lanes 2 and 3, Clostridial colonies containing the empty vector, pMTL-555. Protein expression was detected in both the cell lysates and secretions in the medium in pMTL-10His-cCPE-ETA only.

Transfer and expression of pMTL-10His-cCPE-ETA’ in C. ghonii

The E. coli donor strain, CA434, was used to transfer plasmids into Clostridium by conjugation. Clostridium is not easily amenable to heat- or electro-transformation of plasmid DNA (14). However, we demonstrate that conjugation can be used to transfer plasmid DNA into C. ghonii and, more importantly, this is the first report of the successful DNA transfer into the oncolytic C. ghonii strain. Transfer of the plasmid was achieved from CA434 by conjugation into C. ghonii after the selection of the plasmid by erythromycin and counter selection of E. coli by cycloserine and polymyxin. Clostridium is an obligate anaerobe and cannot grow in the presence of O2. Therefore, to confirm the identity of the recombinant Clostridium, 15 colonies were selected, plated on HI agar and grown under aerobic and anaerobic conditions, respectively. It was found that all colonies were able to grow under anaerobic conditions but were unable to grow under aerobic conditions (Fig. 5B), suggesting that all colonies were Clostridium.

For confirmation of recombinant protein expression, Clostridium strains were grown under anaerobic conditions overnight in HI medium. Protein from cell lysates secreted into the growth medium was analysed by western blot analysis using an anti-His antibody. Subsequent western blot analysis showed that 10His-cCPE-ETA’ was expressed in Clostridium and was secreted into the medim (Fig. 5C).

Efficacy of CPE-ETA’ secreted by C. ghonii

To examine the effects of CPE-ETA’ secreted protein from C. ghonii, bacteria were grown in HI mediun overnight at 37°C under anaerobic conditions. The growth medium was concentrated using ultrafiltration and buffered in PBS. The proteins were used directly in MTT assays and the results expressed as the percentage survival. It was found that medium from non-recombinant Clostridium (pMTL-555) was able to kill all cells tested (Fig. 6), possibly due to endogenous toxins, protease and lipases produced by C. ghonii(15). Previous data from our group has shown that C. ghonii has oncolytic activity when administered in vivo in tumour-bearing mice with high specificity and safety profiles (unpublished data).

Figure 6 Cell proliferation assays of Clostridium-secreted proteins. Assays were performed in triplicate in 96-well plates. Black bars represent the empty vector control group (pMTL-555), while the pMTL-10His-cCPE-ETA group is represented by white bars. The cell lines tested were the HN5, HT29, MCF-7 cell lines, as well as HeLa cells (used as the negative control). Cells were seeded overnight and proteins were added. MTT assays were conducted 48 h after the addition of proteins. 10His-cCPE-ETA’ was significantly different compared to the empty vector control at the highest protein concentration in HN5 (P=0.01), MCF-7 (P=0.0002) and HT29 cells (P=0.011), while the results for HeLa cells (P=0.07) were not significant. Data are presented as the means ± standard error of the mean.

Furthermore, CPE-ETA’ increased the killing capacity of C. ghonii-secreted protein in HN5, MCF-7 and HT29 cells (Fig. 6). It was found that CPE-ETA’ did not affect the proliferation of CLDN4-negative HeLa cells.

Discussion

The current study investigated the expression profiles of CLDN4 in a number of cancer cell lines. CLDN4 overexpression has been implicated in a variety of cancers, including breast, colon, prostate, pancreatic and ovarian cancers (8,16–19). Furthermore, we investigated potential therapeutic strategies to target CLDN4. To this end, the binding domain of CPE (20), a natural toxin with high binding affinity for CLDN4, was employed to ‘piggyback’ the ETA toxin domain of Pseudomonas aeruginosa(21) to cancers overexpressing CLDN4 (12). The engineered immunotoxin upon binding to the CLDN4 receptor is expected to translocate into the cytoplasm via the Golgi apparatus and endoplasmic reticulum where it will bind to EF-2 and inhibit protein synthesis (Fig. 7). The fusion toxin was transfected ino E. coli and the purified protein was assayed against CLDN4-expressing cells. Finally, an oncolytic Clostridial(22) strain was engineered to secrete the CPE-ETA’ fusion protein and the secreted protein was tested on several cancer cell lines.

Figure 7 Mechanism of action of the CPE-ETA’ immunotoxin. The native ETA protein binds the LPR1 receptor or in the case of CPE-ETA’ it binds CLDN4. Upon binding the proteins undergo receptor-mediated endocytosis and subsequently within the endosome the T subunit is cleaved by furin proteases into two domains. The cleaved C+T domain is translocated into the cytoplasm via the Golgi apparatus and endoplasmic reticulum. Finally, the C domain inactivates translation elongation factor-2 by enzymatic ADP-ribosylation. R, receptor-binding domain; T, translocation domain; C, (ADP)-ribosyltransferase A domain; LPR1, low-density lipoprotein receptor-related protein 1.

Analysis of the expression levels of CLDN4 in the following cancer cell lines were found to be consistent with data from previous reports using these cell lines: MCF-7 human breast cancer (4), A549 human lung cancer (23), HeLa human cervical cancer (19), HT29 human colon cancer (24), HCT116 human colon cancer (25) and Huh-7 human liver cancer cells (26). On the other hand, the expression of CLDN4 in the HN5 human head and neck cancer and the MRC-5 human fibroblast cell line has not been previously reported.

Our findings show that purified CPE-ETA’ was highly toxic to almost all the cancer cell types apart from the negative control cell line, HeLa. In addition, blocking the CLDN4 receptor with an anti-CLDN4 antibody abrogated the effects of CPE-ETA’, suggesting that the killing capacity of CPE-ETA’ is dependent on binding to the CLDN4 receptor (Fig. 7). The IC50 data indicated that this fusion protein had a significant effect on the HN5 human head and neck squamous carcinoma cancer cells, the MCF-7 human breast cancer and the HT29 human colon cancer cells. Furthermore, the high sensitivity of HN5 cells opens up the possibility of further research for the clinical application of CPE-ETA’ as a therapeutic agent for the treatment of head and neck cancers.

Fifty percent of patients presenting with head and neck cancer are at an advanced stage of the disease, limiting effective treatment regimes (27). The standard therapy for head and neck cancer is chemo-radiotherapy which is often associated with serious side-effects (28). Targeted therapies have been developed which target the EGFR receptor, disrupting angiogenesis (an important process in the growth and progression of solid tumours) (29). Cetuximab, a monoclonal antibody, is clinically used for the targeted therapy of head and neck cancer (29). In this study, we show that CLDN4 is overexpressed in HN5 head and neck cancer cells and that CPE-ETA’ is a potent inhibitor of HN5 cells by targeting the CLDN4 receptor. Furthermore, this killing capacity of CPE-ETA’ was found to be specific and targeted to CLDN4-expressing cells, as shown by the inability of CPE-ETA’ to inhibit the proliferation of CLDN4-negative HeLa cells (Fig. 3).

Having established the killing capacity of CPE-ETA’ in HN5 cells, we sought to increase the specificity and availability of the CPE-ETA’ in the tumour microenvironment. Clostridial strains have shown promise in seeking and colonising solid tumours. Since Clostridial species are obligate anaerobes, they can only replicate under hypoxic conditions, a hallmark of all solid tumours (22,30). Furthermore, Clostridial species are spore-forming, which makes them ideal for carrying therapeutic payloads to solid tumours, as spores are known to elicit minimal immune response in the host (31). In order to examine the expression of functional CPE-ETA’ fusion proteins in Clostridium, CPE-ETA’ was used to construct the Clostridium expression vector, pMTL-555 (32). A Clostridium secretion signal was incorporated at the N-terminal of CPE-ETA’ for extracellular secretion. The final construct was transferred to C. ghonii by conjugal transfer. Clostridium is not easily amenable to genetic modifications and is not readily transformed by heat shock or electroporation. This is the first report of successfully transforming C. ghonii and expressing functional therapeutic proteins.

Proteins secreted in the medium by recombinant C. ghonii were examined for cytotoxicity using MTT assay and were found to exert effects on several CLDN4-positive cancer lines. Furthermore, C. ghonii modified to express CPE-ETA’ showed increased killing capacity in the HN5, HT29 and MCF-7 cells, but had little effect on HeLa cells (CLDN4-negative). These data suggest that CPE-ETA’ has high specificity for CLDN4-expressing cells. Therefore, in the tumour environment, the expression of this protein in C. ghonii will have little effect on CLDN4-negative cells. Furthermore, secretion of this protein from the tumour may be useful in eliminating distant small metastatic tumours that cannot be colonised by C. ghonii, as these tumours will not have the level of hypoxia to sustain the growth of C. ghonii.

In conclusion, this study confirms the upregulation of CLDN4 expression in certain cancer cell lines and identifies CLDN4 overexpression in the HN5 head and neck squamous carcinoma cell line. Furthermore, we demonstrate that the targeting of HN5 cells with the CLDN4 binding protein, CPE-ETA’, not only shows extreme potency, but high specificity. In addition, we show that oncolytic Clostridia are capable of expressing and secreting the functional CPE-ETA’ fusion protein and that this protein is capable of eliciting cell death in a number of CLDN4-positive cancer cells. The data presented in this study warrant further investigation for using oncolytic Clostridia to deliver therapeutic proteins locally to head and neck and breast cancer cells with high specificity, efficacy and safety.

Abbreviations:

Acknowledgements

This study was supported by the Dr Jian Zhou Smart State Fellowship from the Queensland state government and by grants from the National Health and Medical Research Council and Cancer Council of Queensland to M.Q.W. We would like to thank other members of the Wei Laboratory for their support and helpful comments.

References