My thesis research was conducted in Ben Glick’s laboratory at the University of Chicago, Department of Molecular Genetics and Cell Biology. I had two main projects. The first was studying a red fluorescent protein called DsRed. The second project involved studying subdomains of the Endoplasmic Reticulum called transitional Endoplasmic Reticulum (tER) sites (also called ER exit sites) in the budding yeast Pichia pastoris.
Fluorescent proteins, as the name suggests, are proteins that fluoresce. When you shine one color of light on them, they absorb that light – some physical chemistry happens (e.g. electrons move to higher orbitals, then relax) – and light of a different color is emitted. The color difference between the absorbed light and the emitted light means that one can use optical filters to selectively pass only the emitted light to your eyes or an electronic detector (e.g. a camera). Because they are proteins, these fluorescent molecules can be expressed in cells and they can also be directly fused to other proteins you are interested in, rendering them a visible molecular tag.
The use of fluorescent proteins as reporters for gene expression and as fusion tags has revolutionized cell biology because now we can watch individual molecules in live cells. We don’t want the tags themselves to influence the cellular processes we are examining, therefore we want the fluorescent proteins to be as neutral as possible. The first fluorescent protein ever discovered is called GFP for Green Fluorescent Protein and it came from a jellyfish called Aquoria victoria. As luck would have it GFP is pretty neutral, specifically it is usually a monomer, meaning each copy of GFP stays by itself and doesn’t bind to other copies of GFP. However, there is a need for colors that will complement green, such as red. In 1999 Matz and colleagues cloned new fluorescent proteins from corals and identified a red protein they called DsRed. The [original] wild-type DsRed had a host of problems limiting it’s utility as a research tool. A couple of research groups including the Glick lab undertook projects to make it a better research tool using a technique called directed evolution to engineer improved variants. Directed evolution is just like natural evolution but sped up. Mutations are made on purpose and large libraries containing thousands of variants are screened in a high-throughput manner so that only improved variants are selected. Those improved variants are then mutated, screened, selected etc. When I started on my DsRed project, the Glick lab had just engineered the most advanced DsRed variant which they called DsRed.Express. There were two remaining problems with DsRed.Express, limiting its utility. The most significant was that it was a tetramer, which means that four copies of the protein bind together in a specific orientation (see figure). If tetrameric DsRed is used as a fusion tag it will drive the tertamerization of the protein its fused to which is likely to disrupt its funtion. Therefore my first goal was to engineer DsRed into a monomer (single copy). The second problem with DsRed.Express was that it aggregated a little bit, that is it non-specifically stuck to itself and other proteins in the cell, particularly at high concentrations and this too could interfere with the cells we are monitoring.
To make DsRed.Express into a monomer I first took advantage of other labs’ structural studies of DsRed. They identified the amino acids that made up the tetramerization interfaces on the surface of the protein, i.e. they identified which amino acids were acting like glue. I then used rational, site-directed mutagenesis to specifically disrupt those amino acids and change them to amino acids that were less likely to mediate binding. It so happens that after I made those mutations, that DsRed variant was no longer fluorescent. I therefore used several rounds of directed evolution, incorporating both random and rational mutations, to recover the fluorescence. The end result was a variant we called DsRed.M1 or DsRed.Monomer that had 45 mutations compared to the wild-type protein (there are 225 amino acids in the protein). In collaboration with Bob Keenan’s lab, also at the University of Chicago, we solved the x-ray crystal structure of both DsRed.Express and DsRed.Monomer and were able to rationalize several key properties of these proteins. These results are published HERE.
We then used a similar strategy, i.e. rational mutagenesis informed by structural data combined with random mutagenesis and a novel screening platform to engineer non-aggregating variants of DsRed.Express that we naturally called DsRed.Express2. This work, which I began, was moved forward in dramatic fashion by Rita Strack a new graduate student (at that time) in the lab. A more detailed description of the work can be found in the paper Rita and I wrote.
While engineering DsRed I became interested in the chemical basis for chromophore formation. The chromophore is the part of the fluorescent protein that absorbs and emits light. A key feature of fluorescent proteins is that their chromophores form autocatallytically from amino acids within the protein. This means that a series of chemical reactions rearrange the bonds between atoms within amino acids inside the proteins without the help of additional enzymes or cofactors. Others had previously proposed that DsRed matures along a linear pathway that includes a GFP-like chromophore as an intermediate. There had been a puzzling result that DsRed tetramers occassionally contained copies of the protein with a green fluorescing chromophore instead of a red one and the linear model with GFP as an intermediate was initially an attractive one. However several pieces of data never quite fit this model, then one day we did a journal club on a paper by Rebekka Wachter’s lab about GFP maturation and I hypothesized that some of their findings might apply to DsRed and that the green and red fluorescing chromophores in DsRed might represent distinct endpoints of a branched maturation pathway. Again in collaboration with Rita, we then tested this branched model. A key prediction of the model was that there exists a kinetic competition between the green and red branches, meaning at the split point (3 to 4r or 3 to 4g, in the diagram above) whichever reaction happens first drives the chromophore irreversibly down that branch. We modified the chemical environment for chromophore maturation to push it down one path or the other and serendipitously all of our predictions were validated by the data. Thus we demonstrated that the DsRed chromophore matures along a branched pathway.