Background

After creating 55,000 novel zinc finger sequences, we need to determine which ones effectively bind to their respective target sites. One possible method is to screen for hits, relying on a marker like fluorescence to pick out colonies with zinc finger binding. Though effective, screening requires searching for hits among a large amount of background; accordingly, we chose to use a more efficient system of selection to identify working zinc finger arrays. By tying zinc finger binding to cell survival, all cells without successful binders would die, and thus every colony present would likely represent a viable hit.

We constructed two different one-hybrid selection systems. The first was based off the metabolic system designed by Meng et al (2005) which tied zinc finger binding to histidine production. Under the control of a zinc finger binding site (ZFB) are the genes His3 and URA3, the yeast analogs of the E. coli genes HisB and PyrF (both of which were knocked out in the selection strain). When grown in media without histidine, the cells can only survive if a zinc finger-omega subunit of RNA polymerase (also knocked out in the strain) fusion protein binds successfully to the ZFB and initiates expression of His3. Strong binders can be distinguished from weak ones by addding 3-AT, a competitive inhibitor of His3. Furthermore, to test for inherently leaky promoters, the strain can be grown without zinc fingers in a media containing 5-FOA, which is broken down by URA3 into a toxin. If the ZFB promoter is constitutively on, the cells will express URA3, break down 5-FOA, and die.

The second selection system used TolC, a membrane pump that can remove toxins like Sodium dodecyl sulphate (SDS) from the cell. TolC expression was placed under the control of a zinc finger binding site and the cells were grown in media containing SDS. If the zinc finger array successfully bound to the ZFB, TolC would be produced and SDS would be pumped out of the cell, allowing it to survive. Although we did successfully construct this system, we found that it was less sensitive than the His3 metabolic selection: while the His3 system could recognize hits diluted one in one million of background, TolC could only identify one in one hundred (see One-Hybrid Selection results for details). We accordingly only used the His3-URA3 system to test our zinc finger libraries.

Genome Editing

Where our selection departs from the one described by Meng and others is its use of a genome-based rather than plasmid-based system. Not only did we knock out HisB, PyrF, and rpoZ ourselves using the newly developed techniques of multiplex automated genome engineering (MAGE) and lambda red, we also inserted the zinc finger binding site-His3-URA3 construct directly into the gene instead of expressing it in the cell on a vector (see MAGE and Lambda red results for details). We then used lambda red to directly edit the zinc finger binding site on the genome. While plasmids are easily lost during cell division, the genome is a very stable location, and each cell will have exactly one copy of the selection system, preventing artifacts in the data from one cell by chance having more His3 genes than another. With this selection strain we present the genome as the next stage of Biobrick development: standardized parts, we have shown, can now be inserted directly into the bacterial genome instead of being assembled on multiple plasmids.

References

Xiangdong Meng, Michael H Brodsky, Scot A Wolfe. A bacterial one-hybrid system for determining the DNA-binding specificity of transcription factors. (2005). Nature Biotechnology, 23(8): 988-994.