Our project involves engineering E.coli to invade tumour cells in order to deliver drugs to kill the tumour from within. This essentially requires the engineered cells to display pathogenic properties when two conditions localising the engineered cells to the tumour core are present - lack of oxygen and sufficient number of E.coli present (Figure 1). Please refer to the Project page for more details on the genes and proteins involved.

The two conditions required for expression of the listeriolysin and invasin genes are the two forms of controls exerted on the system. We were concerned that this control might not be tight enough, i.e. the genes may be expressed even when the two conditions are not met. This is usually termed ‘leaky’ expression. For insights on how to address this, as well as other issues elaborated upon below, we consulted two principal investigators, Dr Matthew Chang and Dr Ian Cheong, and respective members of their labs, Adison Wong and Adrian Ng.

Dr Matthew Chang and Dr Ian Cheong are principal investigators at the Centre for Life Sciences (CeLS) in National University of Singapore (NUS), and the Temasek Life Sciences Laboratory (TLL) respectively. Dr Chang’s laboratory deals with synthetic biology of microbial systems while Dr Cheong’s laboratory focuses on finding out novel cancer therapeutics. The former is also programme leader for the NUS Synthetic Biology for Clinical and Technological Innovation programme. While their valuable opinions are not reflected in our current project due to time constraints, we do intend to pass this knowledge on to a future iGEM team.

When we asked Dr Chang about controlling leaky expression, he suggested that we could carry out further characterisation of the anaerobic pNirB promoter we intend to use (see our BioBrick part submissions), to see if it responds to any other input.

Dr Ian Cheong likened promoters to diodes and resistors - diodes only conduct electricity after a threshold voltage (‘all-or-nothing’) whereas resistors conduct electricity incrementally. A resistor-like promoter would continually express the gene at a basal level but increase expression level with more inducer. For instance, in our system, the promoter controlling the effector genes should ideally behave in a diode-like manner. On the other hand, the anaerobic pNirB promoter can be allowed to have some non-anaerobic expression, since a high number of E. coli and hence high AHL concentration is needed before the second switch can be turned on.

As we were talking, Adison from Chang’s lab suggested that we could rethink our initial two-plasmid system. The original concept was to construct this system on two separate plasmids (Figure 2). A ‘transcription factor’ plasmid would constitutively express the esaR repressor and express the esaI protein under anaerobic conditions. A ‘effector’ plasmid would contain the esaR-binding box, and express the effector genes (invasin, listeriolysin, drug of choice) when repression is removed.

Adison proposed that we could place our system in one plasmid for easier transfection, as two plasmids with the same backbone may out-compete each other within the same bacterium (Figure 3[c]). Initially, the two-plasmid system was conceived with the intention that the drug on the effector plasmid might be easily changed and customised for different types of tumours. However, this suggestion does bear consideration, especially as we may want to propose integrating the entire system into the chromosome to reduce likelihood of horizontal gene transfer.

In view that the effector genes, invasin and listeriolysin, are bacterial virulence factors; and that not many therapeutic bacteria have reached clinical trials, we also consulted with Dr Cheong and Dr Chang about the safety of the system if it were to reach clinical trial. We had identified two key challenges: [1] preventing severe immune response in the short-term, and [2] avoiding horizontal gene transfer with commensal bacteria in the long-term.

The use of bacteria in therapies is still a controversial strategy, and to date there has not been any FDA-approved bacterial strains for use in disease treatment. This could mainly be due to the concern of eliciting a massive immune response in patients. When injected intravenously, the therapeutic, but foreign, bacteria will be able to travel to all parts of the body causing a systemic wide immune response. Throughout the body, huge amounts of interleukin-2 (IL-2) and tumour necrosis factor-alpha (TNF-α) will be released, which are key mediators of an immune reaction. Following this, the systemic leakage in the vessels will cause blood pressure to dip. These are the signs of an individual experiencing septic shock and if not treated immediately, the extremely low blood pressure can eventually lead to shock and death.

However such issues can be avoided if the bacteria is injected intratumourally instead. Dr Cheong suggested that it is then important to engineer our therapeutic strain to be extremely antibiotic sensitive so that treatment can be aborted with the use of antibiotics the moment the bacteria invade into other body parts and the patient starts showing signs of fever. Interestingly, Dr Cheong also pointed out that we could put the immune response triggered by our therapeutic bacteria to our advantage by using them as an adjuvant instead. The bacteria, on top of being able to secrete the drug directly in tumour cells, could be used to stimulate the immune response to act against tumour regions, thus enhancing the anti-cancer effects. He supported this with examples of Coley’s toxin and the use of adjuvant chemotherapy to improve survival rate of bladder cancer patients.

Our mentors, Linda and Stuti, have also suggested that instead of a chemical compound, the drug could be something non-toxic (Figure 3[a]), such as short hairpin RNA (shRNA), which acts to silence target gene expression via RNA interference (RNAi). This would serve to minimise toxicity to the bacterium host and reduce safety risks when constructing the system.

Horizontal gene transfer describes processes by which nearby bacteria transfer genes to each other - the term horizontal refers to the fact that this gene transfer does not occur ‘vertically’ (as in a family tree). Many bacteria species are known to acquire antibiotic-resistance genes via horizontal gene transfer from non-related species, making it a clinical challenge.

In our system, we require invasin and listeriolysin for intracellular drug delivery, but these bacterial proteins are also virulence factors. Upon receiving such genes, originally harmless cells/vectors can essentially become pathogens. This raises the potential for horizontal gene transfer to create pathogens in our body, from the normal array of microorganisms the human body naturally hosts. To address this, our interviewees had some suggestions to reduce its likelihood.

First, extra-chromosomal elements are easily shared amongst bacteria - the obvious solution is to integrate the system into the chromosome (Figure 3[d]). Even then, recombination is still a possibility for gene transfer. Subsequently, Adrian from Cheong’s lab suggested that the likelihood for successful chromosomal recombination can be reduced, if we deliberately introduce restriction enzyme sites into the system, either between the genes or within them (Figure 3[b]). This increases the likelihood of the engineered gene cassette being cleaved when the bacterium vector dies and releases free DNA to the environment - thus reducing the likelihood of other bacteria taking up the engineered genes.

Although we were unable to implement most of the suggestions from our interviewees due to time constraints, we have been able to improve upon the design of our system. The ideas are summarised in Figure 3.