Minicircle v Plasmids. Time to think again.

Monday, April 11th 2016

Print this page

pGM169, our clinical trial plasmid.Dr Ian Pringle, University of Oxford.

Comment on:- Transgene sequences free of CG dinucleotides lead to high level, long-term expression in the lung independent of plasmid backbone design, Biomaterials (2016) In Press.


Plasmid vector design is something we have always taken very seriously in the Consortium. When developing vectors for what would become our MD clinical trial we were very determined that only the most appropriate, most optimal design of plasmid could be used. Obviously we wanted the most optimal design to try and generate positive results, but moreover, we knew that we only had 1 chance to make a clinical trial plasmid. Once we committed to large scale production we would have no opportunity to change the design before the single dose trial or from single dose to multi dose study.

It may seem odd, but such a focussed determined effort to really optimise your vector is actually quite rare in gene therapy. For example, one of the most crucial aspects of vector performance in vivo is the choice of promoter used to drive the expression of the transgene. In the case of non viral airway gene therapy we wanted to find a promoter that resulted in long lived expression at high levels. So we went about testing many different promoters in our model systems (Gill et al 2001, Hyde et al 2008, Pringle et al 2012). This may seem like an obvious approach but when one looks at promoters generally used in gene therapy clinical trials they are drawn from a very small pool of promoters (Pringle et al 2012). Eventually we settled on or 'hCEFI' promoter which can drive gene expression for at least 5 months from a single aerosol dose in the mouse lung (Alton et al 2013) though we uncovered other promising candidates in the UbC promoter and the murine FABP4 promoter.


Promoter choice in gene therapy clinical trials (Pringle et al 2012)

Minicircles v Plasmids with Bacterial Backbones.

A technology we were keen to investigate was minicircles. Conventional plasmid vectors have to have a 'backbone' sequence to allow them to be grown in E.coli. The plasmid backbone contains an antibiotic resistance gene (for plasmid selection) and an origin of replication (for plasmid replication). However none of these sequences are useful when once the vector is delivered to target tissues such as the respiratory epithelia. Therefore at least ~1000 bp of your plasmid is not required. Minicircles are plasmids that have been originally designed with recombination sequences to allow the excision and destruction/removal of the backbone. The final product therefore has no bacterial sequences and will be smaller which is potentially a huge advantage for delivery.

A downside of minicircles is the cost and possible yields of production therefore early in our programme we disregarded them as we needed to produce so much plasmid for our aerosol studies. Instead we focussed heavily on plasmid optimisation and the creation of CpG-free plasmids (Hyde et al 2008, Pringle et al 2012, Davies et al 2012, Bazzani et al 2012) to reduce any possible inflammation associated with CG dinucleotides in vivo. We created a series of modular plasmid vectors that allowed the easy switching of component elements to allow us to rapidly produce large amounts of plasmid and compare them in our model systems.


Are Minicircles Superior to Plasmids?

Nevertheless, since then many other research groups have used minicircles with great success. There is evidence that minicircles are more effective than plasmids due to their reduced size and also that they can express for longer in animal models (Mayrhofer et al 2009Gracey Maniar et al 2013Munye et al 2016).


Comparing Minicircles with a Clinically Suitable, Optimised Plasmid Design.

Therefore, we wanted to see what would happen if we took our optimised plasmids and compared them to a minicircle. We had two different minicircles made. One with the hCEFI promoter and a normal Luciferase (Lux) gene and one with a CpG-free Lux gene. We compared these minicircles to two of our plasmids that have the same expression constructs, but the CpG-free backbone we developed for our clinical trial.


Minicircle and plasmid maps (View in new tab ).

Plasmids and minicircles were complexed in PEI and aerosolised to groups of mice. In the first instance, when we had CpG containing transgenes, our results were similar to other research groups. The minicircle vector does indeed show higher levels of activity than the plasmid, despite them having the same expression cassette. However, the expression from this cassette is not optimal and falls over time.


Duration of Lux activity following aerosolisation of PEI complexes. Plasmid and minicircles with CpG containing Lux (View in new tab ).

When we compared the activity from our preferred CpG-free construct to a minicircle with the same CpG-free construct there was no difference. Both vectors displayed expression at a higher level and this persisted for at least 28 days post aerosol. Therefore we do not see any advantage to the minicircle in the case when we compare it to an optimal plasmid construct.


Duration of Lux activity following aerosolisation of PEI complexes. Plasmid and minicircles with CpG containing Lux (View in new tab ).


Transgene sequence much more important than plasmid backbone choice.

We conducted a number of further studies that demonstrated that the actual choice of plasmid backbone does not seem to have an major influence on the expression from the otimal CpG-Free Lux transgene. Different plasmid backbones all result in persistent expression, so long as the transgene was CpG-Free. The same effect was true when we used CFTR instead of Luciferase. Activity from a CpG containing (natural) CFTR cDNA was transient while we saw persistent activity at a much higher level from the CpG-Free form (the plasmid used in our clinical trial).



We suspect that the mechanism responsible for these differences is de novo methlation of the CpGs in the transgenes. This has been observed for other vectors previously in the muscle and liver (Brooks et al 2004Argyros et al 2008) but our previous attempts to detect de novo methylation in the lung were unsuccessful (Pringle et al 2005). Why minicircles show an advantage in these situations is also not clear except maybe simply the presence of a small plasmid backbone allows them to avoid acting as a template for protein binding and histone remodelling than when a backbone is present.


Light microscope image of a human airway liquid interface cultures. Dark patches are mucous.


DNA fragments being cut from an agarose gel exposed to UV.


Large scale lentivirus production in suspension culture.


Proposed 3D model of the CFTR protein.


Mouse lung large airway (cell nuclei blue) transduced with an adenoviral vector (green).


E.coli from a large scale industrial production of our clinical trial plasmid pGM169.


Human airway liquid interface cultures transduced with a lentivirus expressing Luciferase.


Schematic diagram of the large human airways.


Purifying mRNA from tissue samples.


Sheep lung parenchyma (cell nuclei blue) transduced with an adenoviral vector (green).


A CFTR Western blot, to confirm protein production in cell culture.


A cake that only some of us got to enjoy!