Conventional wisdom is proved wrong thanks to new technology developed at Church lab to easily and quickly measure thousands of designed sequences of DNA. I spoke to NSF Graduate Research Fellow Daniel Goodman to find out about the two papers he produced with George and former post-doc Sriram Kosuri (now a UCLA professor).
How does the technology work?
Before you could only really optimize genetic systems by trying a few designs at a time, or by randomly mutating DNA, and then seeing what grows the fastest or performs the best. We are designing and sequencing libraries on the scale of millions of DNA bases to discover how individual DNA elements affect expression or any number of other phenotypes in the cell.
“Measuring all these sequences simultaneously would be extremely difficult, even for our lab, so we came up with a measurement technology called FlowSeq, which uses microarray oligonucleotides as the raw material. Each cell contains a unique member of our library with a different promoter and RBS driving a green fluorescent protein, and then on the same plasmid there’s also a red fluorescent protein being expressed at a constant level, so by comparing the green to the red we can see how much the RBS and promoter are driving the expression. We use fluorescence activated cell sorting (FACS) to sort the bacterial cells one at a time into logarithmically-spaced bins of florescence. Then we barcode the DNA from each bin and sequence them all on an Illumina HiSeq. By sequencing them separately we build back an expression profile for every individual DNA sequence. Leveraging cheap high-throughput DNA sequencing allows us to measure all this DNA on an unprecedented scale.
What did you find out?
“First, we wanted to know about the relationship between different gene regulatory elements – specifically promoters and ribosome binding sites (RBSs) – and to explore how different combinations of elements affected expression. We built a DNA library of around 110 synthetic promoters and 110 synthetic RBSs, giving us a matrix of over 12,000 unique sequences, each about 200 base pairs long. We showed that there are some interesting cross-talk effects where translation strength affects transcription. Additionally certain combinations of promoters and RBSs interfere with each other, which makes the whole notion of independent ‘parts’ a bit trickier. For the first time we can quantify all these effects in terms of how ‘composable’ these genetic elements are.
“Our second experiment asked how using different synonymous codons (different codons that produce the same amino acids) affect gene expression. When synthesizing a gene, it is best to optimize codon choice for a high expression level. The conventional wisdom is to use codons that are common in the organism you’re working with. We wanted to know how changing codon usage at the front of the gene affects expression, so we built a library of sequences that varied N-terminal codon usage (codon usage at the beginning of the gene). We took 137 natural E. coli genes, took their first 10 amino acids, and then varied those codons in a variety of expression context.
“Everyone thought that common codons increase expression because there are more tRNAs for them, but it turns out that at the beginning of the gene, using entirely the rarest codons increases expression by about 15-fold on average, which is pretty surprising. For some genes, the expression change was over 100-fold. To figure out what was going on, we took apart the data codon by codon, and it turns out the reason that rare codons are better is that they happen to be GC poor, not only in E. coli, but in all bacteria with greater than 50% GC content. A-T bonds in RNA are weaker than G-C bonds, so the more A’s and T’s you have, the less secondary structure is present and the more unfolded the RNAs. It’s all about optimizing the folding of the RNA at the beginning of the gene and not about optimizing codon frequency.”
Collecting data on such a large scale means we can get a much better expression with a vastly reduced development cycle.
“By making design and measurement cheaper and faster, we can truly begin to make Synthetic Biology into a real engineering discipline.”
Read the papers!
(0) Readers Comments
April 08, 2014
January 08, 2014
March 24, 2014
January 13, 2014