Team:William and Mary/Part Collection



Part Collection


Our team is nominating our sensor circuit part collection, including both our set of Central Dogma Burden sensors and our “Beyond Burden” sensors*, for the parts collection special prize. These circuits are a part of our orthogonality assessment toolkit, including both these sensor circuits and a model. When the fluorescence output from our circuit in a host transformed with an additional circuit is input into the model, teams will be able to receive an orthogonality assessment. Our Central Dogma Burden sensor consists of a set of circuits that are able to assess burden at the transcriptional, translational, and post-translational levels which can then be input into our model. Our “Beyond Burden” sensors were developed from our analysis of RNA-seq data. In this analysis, we found that there were certain genes that were differentially expressed in hosts with and without circuits. Therefore, our team was able to use the expression of these genes as a “marker” of host orthogonality. These sensors are able to indicate the gene expression of these “marker” genes with fluorescence, providing an input for our model.

.

Our Toolkit


Our toolkit consists of both a data-driven mathematical model and a system of sensor circuits. Our model has two integrable components; the first is a mechanistic model using an ODE system which dynamically models metabolism of an E. coli cell, and outputs concentrations of molecules involved in metabolic processes. After conducting a literature review of published RNA-seq data, our team found that certain genes were consistently differentially expressed between a host with and without a circuit, and coupled equations describing concentrations of these “marker” genes with the purely metabolic system. We first simulate an unburdened E. coli cell, then simulate induction of a circuit by modifying parameters derived from measurements from our sensor circuits to reflect increased burden due to induction of a circuit. Based on these outputs, we quantify burden-based orthogonality due to changes in their steady states between simulation conditions.

In addition, we derived a completely data-driven “beyond burden” model which directly uses RNA-seq datasets to find correlations between “marker” genes and an overall increase in differential gene expression. This model uses “marker” genes which are highly correlated with overall increased differential gene expression as a proxy for detecting increased non-orthogonality. Our models are data-driven, requiring teams to input data from their circuits to receive an output, and are able to utilize three different kinds of data: RNA-seq data, qRT-PCR data, and fluorescence output data from our sensor circuits. The mathematical model is available on this wiki as well as on GitHub as well and therefore accessible to all teams.

Our team designed a system of sensor circuits to be used in conjunction with our mathematical model as the second part of our toolkit. These sensors produce a fluorescence output that when input into our model can provide an orthogonality assessment. Within our system, there are two separate sets of circuits; the Central Dogma Burden sensors and the “Beyond Burden” sensors. The Central Dogma Burden system senses burden at the transcription, translational, and post-translational levels. Decreasing levels of fluorescence between the host with and without an additional circuit will indicate levels of burden on the host. Our second set of sensors, the “Beyond Burden” sensors, utilizes the “marker” genes revealed by our RNA-seq analysis. These sensors also produce a fluorescence output in which changing levels indicates a lack of circuit-host orthogonality. Teams can input the fluorescence output from these systems into our model, resulting in an orthogonality assessment.

.

Individual Circuits and Designs

Transcriptional Burden Sensors (BBa_K3773511 & BBa_K3773512)

After a thorough literature search, it became clear that the best method with which to sense transcriptional changes was through aptamers. When deciding which aptamer to use for our sensor, we focused on three aptamers specifically: Mango, Spinach, and Broccoli. These aptamers are widely regarded as “the most promising fluorescent RNA aptamers” (Shanaa et al. 2021). As Broccoli binds to a non-toxic, membrane-permeable dye (DFHBI-1T) that has minimal effects on host activities, is widely used in vivo, and has a relatively simple testing procedure, our team decided to use Broccoli as our aptamer (Okuda et al., 2017). However, we still needed to choose which form of Broccoli to use. We eventually decided on F30-Broccoli, as it has been shown that its scaffold is not recognized by nucleases for degradation like other aptamers, that it is regarded as the “best-performing aptamer,” and that it is widely used (Filonov et al., 2015; Thorn et al., 2017). Using this aptamer, we designed two transcriptional burden sensors. Our first design, WM21_011 (BBa_K3773511), places control of F30-Broccoli expression under the strong, constitutive Anderson promoter BBa_J23119 (Yan and Fong, 2017), without incorporation of any RBS or spacers. Our second design, WM21_012 (BBa_K3773512), places control of F30-Broccoli expression under the strong, constitutive P70a promoter without incorporation of any RBS or spacers. In both circuits, F30-Broccoli (Filonov et al.) is directly followed by the strong, synthetic, bidirectional terminator BBa_B1006.

In addition, we used the 3WJ dimeric Broccoli (3WJdb) circuit (pUC19-P70a-3WJdB-T), which was designed and constructed entirely by researchers Alam et al. This circuit contains the dimeric Broccoli aptamer placed within the three-way junction (3WJ) RNA motif, and this aptamer forms a secondary structure that is able to bind to a membrane-permeable, fluorescent dye known as DFHBI-1T. DFHBI-1T only fluoresces after binding to the Broccoli aptamer. This circuit places 3WJdb under the control of the constitutive P70a promoter. This circuit was acquired from Addgene (#87311).

Once the correct aptamer dye has been applied to cells transformed with this circuit, a fluorescence output will be produced from the aptamer. As the circuit does not include an RBS, this circuit is able to detect decreases in burden on the transcriptional level. Therefore, based on the amount of aptamer transcribed, the fluorescence output will fluctuate, providing a measure of the test circuit’s impact on the host.

Translational Burden Sensors (BBa_K3773516 & BBa_K3773513)

Our team decided to use codon optimized sfGFP as our marker gene in this circuit. As the amount of sfGFP translated is directly proportional to the amount of fluorescence produced, sfGFP can be used as an indicator of changes in translation levels within the host cell. The translational burden sensor aims to detect changes in host translation after addition of a test circuit into the host. Changes in our sensor circuit’s fluorescence levels that occur after cotransformation of the host with the desired test circuit will reflect the impact of the test circuit on host function, therefore reflecting circuit-host burden level at the translational level.

For our translational burden sensor circuits, we tested two designs. The first version of this circuit, BBa_K3773516, was designed by researchers Ceroni et al. (Ceroni et al., 2015) to test translational burden of genetic constructs within a cell. However, while researchers Ceroni et al. incorporated their circuit into the host genome, we utilized their circuit in plasmid form. Researchers Ceroni et al. used the strong constitutive promoter BBa_J23100 and an RBS designed (using an RBS calculator) specifically for codon-optimized sfGFP. The sequence for codon-optimized sfGFP was created for high expression levels using DNA2.0, and sfGFP was followed by a spacer and the synthetic terminator BBa_B1002. Based on Ceroni et al.’s circuit concept and design, our team designed a translational burden sensor (BBa_K3773513) consisting of the strong, constitutive Anderson promoter BBa_J23119, the RBS B0034 with a spacer included, the codon optimized sfGFP sequence used by researchers Ceroni et al. (2015), and the strong synthetic, bidirectional terminator BBa_B1006.

Post-Translational Burden Sensors (BBa_K3773515, BBa_K3773523, BBa_K3773524, & BBa_K3773522)

Our team developed our post-translational sensor circuits based on three different post-translational modifications: phosphorylation, glycosylation, and protease cleavage. Given that phosphorylation is so common in the cell, our team wanted to develop a sensor of this post-translational modification. In order to sense phosphorylation, our team utilizes the EnvZ/OmpR system found natively in E. coli. EnvZ is an outer membrane protein that responds to changes in the osmolarity of the surrounding environment. When this protein detects osmolarity changes, it autophosphorylates, then passes the phosphate to the OmpR protein. Depending on the osmolarity of the surrounding medium, OmpR will either bind to the ompC promoter in high osmolarity states or to the ompF promoter in low osmolarity states, producing these genes coding for additional outer membrane proteins (Cai et al., 2002). Our post-translational modification sensor uses this EnvZ/OmpC system to look at phosphorylation rates in the cell, as the phosphorylation of EnvZ causes OmpC production. This circuit, WM21_015 (BBa_K3773515), consists of the ompC promoter and BBa_B0034 as the RBS, driving production of sfGFP with Bba_B0015 as the terminator (Clifton et al., 2018; Ceroni, et al., 2015; Hartvig et al.; 1996, Genbank V01146). Our construct builds off of the 2012 Tokyo-NoKoGen construct (BBa_K769001) using their same promoter sequence; however, our team is using sfGFP instead of the original GFP (Tokyo-NoKoGen 2012 iGEM Team, http://parts.igem.org/Part:BBa_R0082).

When we discussed our project with Dr. Barrick from UT Austin, he suggested that phosphorylation may not be an accurate measure of circuit-host burden due to the high levels of phosphate in the cell. To address this concern, our team also designed two sensors to indicate changes in levels of proteolysis. Each of these sensors places sfGFP after a promoter which normally controls the expression of a proteolysis-related protein. For the first sensor (BBa_K3773523), the lon promoter was chosen due to its importance as a protease and frequent upregulation upon introduction of a heterologous circuit. In E. coli, this promoter controls expression of the ATP-dependent Lon protease, which selectively degrades certain proteins. This protease is also upregulated in response to other forms of stress (https://www.uniprot.org/uniprot/P0A9M0, Oct 10. 2021). Lon has been observed to degrade about 50% of misfolded proteins in E. coli (He et al., 2018). For the second sensor (BBa_K3773524), the hslVU (a.k.a. clpQY) promoter was chosen, as the hslU gene it controls was also found to be even more frequently upregulated than lon in response to a heterologous circuit. hslU, also known as clpY, is the ATPase subunit of the HslVU complex. The HslVU complex forms a protease which is also part of the heat shock response (Lien et al., 2009). These two circuits consist of the lon and hslVU promoters driving production of sfGFP (Ceroni, et al., 2015). To promote transcription, the strong RBS BBa_B0034 was used and transcription was halted by the BBa_B1002 terminator (Prakash, 2018).

In addition to phosphorylation and proteases as post-translational modifications, our team also looked at glycosylation to quantify the effect of circuit-imposed burden on the post-translational modification stage of the central dogma. The mur operon was the best candidate for protein glycosylation in non-pathogenic strains of E. coli because of its role in peptidoglycan synthesis. The ftsLp2 promoter was chosen for our circuit focusing on glycosylation (BBa_K3773522) because it initiates the transcription of genes responsible for a N-acetylglucosaminyltransferase (murG) and a UDP-N-acetylmuramate-L-alanine ligase (murC) in peptidoglycan synthesis (Barreteau et al., 2008; Vicente et al., 1998). This promoter is regulated by the DNA-binding transcriptional repressor LexA. This circuit consists of the ftsLp2 promoter, which is responsible for the N-acetylglucosaminyltransferase (murG) and a UDP-N-acetylmuramate-L-alanine ligase (murC) in peptidoglycan synthesis (Barreteau et al., 2008; Vicente et al., 1998). Downstream of ftsLp2 is a codon-optimized sfGFP gene that allows fluorescence to be expressed at levels proportional to the mur operon for easy quantification of post-translational modification activity in response to circuit-imposed burden. This circuit also utilizes BBa_B0034 as an RBS driving production of sfGFP with BBa_B0015 as the terminator (Clifton et al., 2018; Ceroni, et al.; 2015, Hartvig et al.; 1996; Genbank V01146).

Heat Shock Response Sensors (BBa_K3773518 & BBa_K3773521)

Our Venn diagram analysis revealed that genes coding for heat-shock proteins were the most consistently differentially expressed genes among a variety of experiments using different circuits and conditions. These common DE genes were ibpA, ibpB, htpG, dnaK, dnaJ, groS, and clpB. In general, these genes code for small heat shock proteins, chaperone proteins, and chaperonins that play a role in bacterial response to stressors such as fluctuating temperatures, limited availability of nutrients and exposure to harmful chemicals (Cha et al., 2020; Ceroni et al., 2018). With the reliance on chassis machinery to express products of synthetic circuits, it is expected that the host resources would be unnaturally consumed in this process and induce the heat-shock response (Ceroni et al., 2018). Thus, we aimed to use the changes in the expression of these genes as a marker for measuring how much synthetic constructs burdened their host cells and contributed to unwanted interactions between the host and the construct.

We first looked into the literature to see if sensors for these genes were already available. We found that in their 2018 paper “Burden-driven feedback control of gene expression,” Ceroni et al. had successfully created and tested GFP-expressing sensor plasmids that used promoters for ibpA and ibpB (both in the ibpAB operon) and htpG (regulated by two overlapping promoters htpG1 and htpG2). In addition to these genes which Ceroni et al. found to be commonly DE from their own RNA sequencing analysis, they had created another sensor for the gene groS which was also consistently highly upregulated in our Venn diagram analysis. Although groS was not commonly DE for Ceroni et al. (2018), they utilized its expression as a burden marker because all the heat-shock genes identified as commonly DE were under the control of σ32. The groSL operon encodes heat-shock genes that control the amount and activity of σ32 in cells (Ceroni et al., 2018). Overall, we were able to acquire their sensor constructs ibpAB-GFP, htpG1-GFP, htpG2-GFP, and groSL-GFP from Addgene. Our team designed the remaining heat shock sensor circuits for dnaK (BBa_K3773521), dnaJ (BBa_K3773521), and clpB (BBa_K3773518).

Our overall design for the remaining heat-shock sensor circuits for dnaK, dnaJ and clpB consisted of a promoter that was upstream of the individual heat-shock gene coding region or its operon in the native E. coli genome followed by sfGFP. Since the dnaK and dnaJ genes are under the control of the same dnaK operon promoter (Cowing et al., 1985), we designed a single sensor for measuring the expression levels of both genes. The promoter sequence obtained from Cowing et al. (1985) included the spacer and RBS regions, thus we only added the sfGFP and terminator (BBa_B1006) sequences (Ceroni et al., 2015; Cambray et al., 2013). For clpB, we created two versions of our sensor circuit. Each version uses the promoter for the native gene in E. coli followed by one of two RBS regions: either BBa_B0034 or the native RBS preceding the gene in E. coli. The promoter and RBS control the expression of sfGFP, which is halted by BBa_B1002 in both circuits (Ceroni et al., 2015; Prakash, 2018).

Secretion Sensors (BBa_K3773519 & BBa_K3773520)

This circuit senses the downregulation of AG43 biofilm production in response to the burden imposed on a cell by the addition of a circuit. The flu gene is part of an operon including the isrC and flu genes. The operon contains three promoters: flup, flup1, and flup2. Flup activates both isrC and flu, whereas flup1 and flup2 each only activate the flu gene, therefore we eliminated the possibility of using flup as our promoter due to its lack of specificity (RegulonDB, 2019). We designed two versions of the Secretion Circuit, one using flup1 as the promoter (BBa_K3773519), and the other using flup2 (BBa_K3773520). Since flup2 is downstream and thus closer to the start of the gene, we expected this circuit to perform better, however we tested both versions in case this hypothesis was incorrect.

This circuit also utilizes BBa_B0034 as an RBS driving production of sfGFP with BBa_B0015 as the terminator (Clifton et al., 2018; Ceroni, et al., 2015; Hartvig et al., 1996; Genbank V01146).

Our Venn diagram analysis revealed that the flu gene is usually significantly downregulated when a circuit is introduced to an E. coli cell. This consistent downregulation means that we can use flu gene transcription levels as yet another means to assess the effect of circuit-host interactions on the host cell. The flu gene controls biofilm production, specifically that of AG43, and we chose to use this gene as an indicator of cellular secretion being affected by burden.



*Our team understands that many of these circuits can fit into both the Central Dogma sensor circuit category and "Beyond Burden" sensor circuit category; therefore, these circuits were grouped together for clarity.

References



Alam, K. K., Tawiah, K. D., Lichte, M. F., Porciani, D., & Burke, D. H. (2017). A Fluorescent Split Aptamer for Visualizing RNA-RNA Assembly In Vivo. ACS synthetic biology, 6(9), 1710–1721. https://doi.org/10.1021/acssynbio.7b00059

Barreteau, H., Kovac, A., Boniface, A., Sova, M., Gobec, S., & Blanot, D. (2008). Cytoplasmic steps of peptidoglycan biosynthesis. FEMS Microbiology Reviews, 32(2), 168–207.

Cai, S. and Inouye, M. (2002). EnvZ-OmpR Interaction and Osmoregulation in Escherichia coli. Journal of Biological Chemistry. https://doi.org/10.1074/jbc.M110715200

Cambray, G., Guimaraes, J., Mutalik, V., Lam, C., Mai, Q., Thimmaiah, T., Carothers, J., Arkin, A., & Endy, D. (2013). Measurement and modeling of intrinsic transcription terminators. Nucleic Acids Research. 41(9):5139-5148.doi: 10.1093/nar/gkt163

Ceroni, F., Algar, R., Stan, G., & Ellis, T. (2015). Quantifying cellular capacity identifies gene expression designs with reduced burden. Nature Methods, 12(5):415-418. Doi: 10.1038/nmeth.3339

Ceroni, F., Furini, S., Gorochowski, T.E., Boo, A., Borkowski, O., Ladak, Y., Awan, A.R., Gilbert, C., Stan, G.V., & Ellis, T. (2018). Burden-driven feedback control of gene expression. Nat Methods, 15, 387–393. https://doi.org/10.1038/nmeth.4635

Cha, H., Srivastava, R., Vakharia, V., Rao, G., & Bentley, W (1999). Green Fluorescent Protein as a Noninvasive Stress Probe in Resting Escherichia coli Cells. Applied and Environmental Microbiology, 65(2):409-411. https://doi.org/10.1128/AEM.65.2.409-414.1999

Clifton, K., Jones, E., Paudel, S., Marken, J., Monete, C., Halleran, A., Epp, L., Saha, M. (2018). The genetic insulator RiboJ increases expression of insulated genes. J Biol Eng. doi: 10.1186/s13036-018-0115-6

Conway, J., Lex, A., & Gehlenborg, N (2017). UpSetR: An R Package for the Visualization of Intersecting Sets and their Properties. Bioinformatics. 33(18):2938–2940. doi: https://doi.org/10.1093/bioinformatics/btx364

Cowing, D.W., Bardwell, J.C., Craig, E.A., Woolford, C., Hendrix, R.W. & Gross, C.A. (1985). Consensus sequence for Escherichia coli heat shock gene promoters. Proc. Natl. Acad. Sci. U.S.A. 82(9), 2679-2683

Filonov, G., Kam, C., Song, W., and Jaffery S. (2015). In-gel imaging of RNA processing using broccoli reveals optimal aptamer expression strategies. Chem Biol. 22(5). Doi: 10.1016/j.chembiol.2015.04.018.

Genbank Accession Number V01146, https://www.ncbi.nlm.nih.gov/nuccore/V01146

Hartvig, L., & Christiansen, J. (1996). Intrinsic Termination Of T7 RNA polymerase mediated by either RNA Or DNA. The EMBO Journal. 15(17):4767-4774.

He, L., Luo, D., Yang, F., Li, C., Zhang, X., Deng, H., & Zhang, J. (2018). “Multiple Domains of Bacterial and Human Lon Proteases Define Substrate Selectivity”. Emerging Microbes & Infections, 7(1)1–18. doi:10.1038/s41426-018-0148-4.

https://www.uniprot.org/uniprot/P0A9M0

Lien, H., Yu, C., Liou, C., & Wu, W. (2009). “Regulation of ClpQ+Y+ (HslV+U+) Gene Expression in Escherichia Coli”. The Open Microbiology Journal, 3(1):29–39. doi:10.2174/1874285800903010029.

Okuda, M., Fourmy, D., & Yoshizawa, S. (2017). Use of Baby Spinach and Broccoli for imaging of structured cellular RNAs. Nucleic Acids Research, 45(3):1404-1415. https://doi.org/10.1093/nar/gkw794

Prakash, S. (2018). De Novo Engineering of trans-activating riboswitches in E. coli (thesis).

Shanaa, O., Rumyantsev, A., Sambuk, E., & Padkina, M. (2021). In Vivo Production of RNA Aptamers and Nanoparticles: Problems and Prospects. Molecules. 26(5): 1422, https://doi.org/10.3390/molecules26051422

Thorn, K. (2017). Genetically encoded fluorescent tags. Molecular Biology of the Cell. 28(7). https://doi.org/10.1091/mbc.e16-07-0504

Tokyo-NoKoGen 2012 iGEM Team, http://parts.igem.org/Part:BBa_R0082

Vicente, M., Gomez, M. J., & Ayala, J. A. (1998). Regulation of transcription of cell division genes in the Escherichia coli dcw cluster. Cellular and Molecular Life Sciences, 54(4), 317-324.

Yan, Q., & Fong, S. S. (2017). Study of in vitro transcriptional binding effects and noise using constitutive promoters combined with UP element sequences in Escherichia coli. Journal of biological engineering, 11, 33. https://doi.org/10.1186/s13036-017-0075-2