Difference between revisions of "Team:William and Mary/RiboJ"

In the past many teams have rigorously characterized parts both new and old, so what separates the Circuit Control Toolbox from being just another characterization project? The answer to that is multifold, but mainly consists of the fact that the Circuit Control Toolbox not only rigorously characterizes novel or useful tools to alter transfer functions, but also allows these tools to be generalizable to any circuit, not just to the specific circuit in which they were characterized.

A key problem of generalizing the results of characterizations is that the dynamics of gene expression are influenced by the protein coded for. For our toolbox to be able to to be used orthogonally at the end of an arbitrary genetic circuit, we must be able to use the characterization done in fluorescent proteins on any other protein. That is, we have to be able to insulate a circuit from its genetic context.

Fortunately, Lou et al. from the Voigt lab at MIT discovered a ribozyme which does just that. RiboJ is ribozyme derived from a hammerhead ribozyme, which self cleaves, removing the upstream region, most importantly the 5’ untranslated region. (Figure 1) When Lou et al. 2012 (“Ribozyme-based insulator parts buffer synthetic circuits from genetic context”) measured two different variants of GFP (sfGFP and cl-sfGFP fusion), they found that the relative expression for the induction curves was not always the same, likely due to upstream effects. However, when they added RiboJ before the coding region, the relative expression levels for the induction curves collapsed, indicating that the relative expression had been homogenized for a given promoter (Figure 2). This means that the gene expression dynamics are controlled only by the promoter, and not by the coded for protein, which means as long as the promoter is kept the same and RiboJ is added before all coding regions, our characterizations can be used for different proteins.

Figure 1: Figure 1: Structure of RiboJ, the loop in uppercase is derived from the sTRSV ribozyme, while the additional loop is in lowercase. The catalytic core of the active ribozyme is circled in red and the smaller loop serves to promote translation by exposing the RBS. Figure adapted from Lou et al.

Characterization was done on 07/26/16 using flow cytometry, which means that our data shows single cell resolution of the impact of various RBSs. Characterization was done in BL21 E. coli, which is a standard protein expression strain. A high copy plasmid (1C3) containing one of the pLacO1 RiboJ RBS variants (ex. Bba_K206636) was co-transformed with a low copy plasmid (3K3) which contained a constitutively expressed LacI (Bba_K206616) to repress the promoter. Individual colonies were picked from antibiotic selection plates, and colony PCRed to ensure the correct plasmids were contained. For each RBS variant 3 colonies containing both plasmids were inoculated in M9 and incubated till midlog growth was reached. At that point the cells were induced with various concentrations of IPTG until steady state was reached. Cells were then FACSed and their fluorescence was reported in MEFL calculated using Spherotech calibration beads. (Figure 2).

Figure 2: Average (3 biological replicates) population level fluorescence of RBS library over different induction conditions. Note the wide range of RBS strengths in the library, and that the strongest RBS B0035, is not actually the strongest until induced at a concentration of 1000µM IPTG. This illustrates the importance of evaluating over a transfer function, not just at an arbitrary steady state.

All of the individual RBS plots are available on our measurement page, but here are a few examples of individual RBS measurements.

Figure 3A

Figure 3B

Figure 3A and 3B: Fluorescence measurements of replicate 1 of B0031 (BBa_K2066036) un-induced (A) and induced at [IPTG] 100µM (B). Note the relative crispness of peaks, indicating a nearly normal distribution of florescence. The majority of all replicates are similarly unimodel in their distribution.

Figure 4A

Figure 4B

Figure 4A and 4B: Fluorescence measurements of replicate 3 of B0031 (BBa_K2066036) un-induced (A) and induced [IPTG] 100µM(B). Note the bimodal distribution compared to figure. In this case induction occurs, but a subset of the population is maximally induced at all induction conditions. All induction conditions of number 3 contained this distribution, while other replicates of this biological circuit had a more normal fluorescence distribution, indicating that something unusual, likely the loss of the repressor is going on in this replicate. Of note, this bimodal distribution would not be noticeable without the use of FACs to observe at a single cell level.

While teams have measured RBSs in the past (Warsaw 2010), to our knowledge no team has ever measured the entire community library over a variety of induction conditions. As illustrated by B0035’s variable relative strength at different induction conditions, it is vital to perform characterization at different induction conditions rather than just constitutively. However, our findings have been consistent with the measurements on the registry, and much like Warsaw 2010 we found that B0030 was stronger than reported in previous measurements on the registry. Additionally, measurements on the registry by Kelly and Rubin in 2007 indicate that B0035 is stronger than B0034. Given that their data was recorded from a strong constitutively expressed promoter, their measurement makes sense in the context of our findings that B0035 is a stronger RBS only when expressed at a high level.

While the community RBSs are designed for and most commonly used for expression in E. Coli, we thought that it would be useful to characterize them in a variety of conditions. So as a collaboration we sent a blinded version of the library to University of Pittsburgh’s iGEM team, who then proceeded to characterize the RBSs in the cell free system S30 (Figure 5). We also sent a blinded library to the high school team Alverno, who characterized them using Richard Murray’s TX-TL cell free system (Figure 6). In both cases, the majority of RBSs had relative expressions similar to our characterization. However, Team Alverno characterized B0064 as having a significantly higher relative expression than was found in existing registry characterization as well as our characterization. We talked to UPitt iGEM, and they noted that due to technical concerns they were not confident about the validity of their first set of measurements (Figure 7). When we looked at the unblinded data from the first set of measurements, we noted that they had a similarly high level of B0064 expression. Future experiments are needed to determine whether that is an artifact of a common procedural error in cell free systems, or is representative of a novel result in a cell free system.

Figure 5: Unblinded graph of UPitt iGEM’s measurements of our RBS library in the S30 cell free system. Their data is in line with both existing measurements, as well as our measurements.

Figure 6A: Unblinded graph of Team Alverno’s characterization of our RBS library in the TX-TL cell free system.

Figure 6B: Time course measurements of Team Alverno’s characterization of our RBS library in the TX-TL cell free system.

Our RBS characterization highlights the importance of rigorous characterization, especially in the context of our Circuit Control Toolbox. To be able to use RBSs to tune the amplitude of any given transfer function, you need to be sure that you can apply your characterization to a given situation. For example, using our RBS characterization it becomes clear that it is important to consider both the system you are working in as well as the number of mRNA transcripts created. Someone trying to use B0035 for high expression of a protein under the control of a weak constitutive promoter needs to be aware that B0035 is only the strongest relative RBS when there are a large number of copies of mRNA. Keeping this in mind, it is now possible to use our characterization to tune the amplitude and relative max of various transfer functions. Using RiboJ to remove 5’ untranslated regions, and keeping in mind the various limitations of any set of characterizations, our characterization of this will be of use in the Circuit Control Toolbox, and for teams at large.

Figure 7: Team Pitt’s first series of measurements

1. C. Lou, B. Stanton, Y.-J. Chen, B. Munsky, C. A. Voigt, Ribozyme-based insulator parts buffer synthetic circuits from genetic context. Nat. Biotechnol. 30, 1137 (2012). doi:10.1038/nbt.2401 pmid:23034349