Motivation & Idea

Doing the iGEM lab work, cloning some stuff with 3A assembly. Hasn’t everyone wished you could just take your parts, cut the plasmids, mix and ligate them and be done with it? No need for time intensive (and expensive!) purification of the cut plasmids.

We designed a system that will allow you to do exactly that. The idea is very simple: If we would combine this with parts in the normal iGEM plasmids, we’d get re-ligation products of the original plasmids as well as possible side products (see figure 1). How can we get rid of all of these in one simple step?

Overview over the possible products and side products. By inducing the kill genes in side products or wrong backbones, we can select against bacteria carrying the false constructs.

Quite easily, since the common feature of all the undesired ligation products is that they have (at least) one backbone of the two plasmids, which were source for the parts we want to fuse together. If we put a negative selection marker into the backbone of these plasmids, we can easily get rid of them - not during the ligation, but directly after the transformation.

This method will of course reduce the effective transformation efficiency, because only one out of 10 or more ligations products is desired and able to form a colony after transformation. However, we can easily put a lot more DNA into the ligation reaction, since we can use the whole restriction mixture.

And if we are already improving cloning techniques, why stop at restriction and ligation? Can we not also make plasmid purification easier, which is quite expensive - especially for small iGEM teams with little funding - and also complicated to do using pipetting robots.

How could a one-step purification of only (desired) plasmids from the lysate of bacteria work? We’d need something, that can bind plasmid-DNA, but does not bind genomic DNA. This can only be achieved through a sequence specific binding, so we’ll have to use nature’s toolkit for this.

There are a lot of very specific DNA binding proteins out there, for example the zinc-finger proteins, which are already well characterised from their use in zinc-finger-nucleases. However, proteins are not very stable and putting them directly on a column is - while obviously possible - pretty much the opposite of what makes an method easy to use or cheap.

So how can we use proteins to bind plasmids, without having them prepared? Well - as always, we could say - our friends the bacteria can help us. If we already need to design a plasmid with a specific binding sequence in it, why not make that plasmid express a protein, that directly binds it? That way we only need a way bind our protein of interest with a specific tag. For strong binding and maximum efficiency we can use biotin and fuse our DNA-binding protein with streptavidin.

The combination of 3D printing and some good old chemistry gives us the possibility to create our own columns with a biotinylated surface, that should be able to pull our plasmids from E. coli lysates.

Design

The idea is simple: by integrating an inducible kill gene into the backbone, we can specifically select for our desired combination.

Using different inducers and kill switches, which can be found in the registry by loads, we can create a series of plasmids which can also be combined to bring more than one plasmid into the same culture of bacteria.

Kill cassettes

The first idea which comes to mind when thinking about kill switches is the combination of Holin and Endolysin (Young, 2002), which was already used by several iGEM teams. However, other proteins such as PezT (pneumococcal epsilon-zeta toxin, Mutschler et al., 2011) , part of a prokaryotic toxin/antitoxin system, were already used in iGEM as well, e.g. by Team Hamburg in 2015. PezT is smaller than Holin, which makes it easier to integrate into the plasmids, and also makes it easier to transform. We decided to look at both variants.

Looking at different induction systems, we chose three well-known systems: the LacI and LacZ cassettes, as well as a rhamnose-inducible promoter and the TetR (tet-on) system (Chan et al., 2015).

Combination of the inducers with the kill genes and introduction of restrictions sites which would allow us to integrate the cassettes into backbones without interfering with the RFC standards (see below) gives rise to three different cassettes:

Apa-pRha + RBS-PezT + Terminator-Bam-Hind
Apa-const-Promoter + LacI cassette + LacI promoter + RBS-PezT + Terminator-Bam-Hind
Apa-const-Promoter + TetR casette + TetR promoter + RBS-PezT + Terminator-Bam-Hind

Plasmid Purification

As described below, the plasmid purification depends on zinc-fingers which bind to special sequences on the DNA. By using fusion proteins of zinc fingers with streptavidin, we can pull out the desired plasmid bound to the zinc finger on biotin-covered surfaces.

We selected three zinc fingers which are reported to bind strongly to their respective binding sequences.

We decided to use the following zinc finger proteins to bind our plasmids, since they were reported to work well and have a strong binding affinity:

C7C7 (source: mouse transcription factor, Liu et al., 1997)
K230R (source: de-novo design against human CCR2 / CCr7, Lee et al., 2010)
ZFN-L (source: de-novo design against human CCR5, paper: Lombardo et al., 2007)

We integrated the respective binding site for each zinc-finger-streptavidin fusion protein (from now on referred to only as zinc-finger) behind the promoter of the respective zinc-finger, so that binding of the protein will partially inhibit its own transcription. This will prevent over-production of the zinc-fingers in the bacterium, as we only need as many copies of the protein as there are the plasmids that contain them.

Integration into Plasmid Backbone

Starting from the standard iGEM plasmids (pSB1C3, pSB1A3), we need to introduce restriction sites into a non-coding, non-regulating sequence in the backbone that will allow us to clone our finished constructs (possibly one of each kind) into the backbone. Since we want to create RFC compatible plasmids, we have to use non-RFC restriction enzymes for that.

We decided to use ApaI, BamHI and HindIII for this, as they are easily available to us. To find the best position for the insertion of new restriction sites, we used a python script (see here). This allowed us to find the position in the backbone where the lowest number of mutations would be necessary for any combination of two of the three enzymes in all non-annotated (as analysed by SnapGene, see fig. 3) sequences in both plasmids. This allows us now to insert any combination of a negative selection cassette and plasmid purification cassette into the mutated backbones, that have the additional restriction sites.

Plasmid maps of pSB1A3 (A), pSB1C3 (C) and the mutant plasmids (B and D, respectively) containing new restriction sites to include the kill cassettes and/or zinc fingers.

Results

We tested three different 3D printer materials concerning their stability in solvents which could be used during the linking reaction of biotin to a 3D-printed microchip. After two hours incubation of the materials in the solvents, we observed that all the materials were either completely dissolved or were very soft to the touch (see fig. 4). Since we could not leave out piperidin in the coupling reaction, we had to re-think on better alternatives for coupling our biotin. The Intavis AG in Tuebingen helped us out by giving us a derivatized cellulose membrane, originally used for their SPOT synthesis, which stay stable during the coupling process. Sadly, due to time problems, we could not test the coupling of biotin to the membrane.

Different 3D printing materials (right side) were incubated in the indicated solvent for two hours. None of the materials was stable enough to keep its state over the time. Pip = piperidine

(References are listed here)