iGEM RESEARCH ARTICLE: Development and Characterization of Fluorescent and Luminescent Biosensors for Estrogenic Activity
Note for Reviewers
This work has been submitted for open review as an iGEM Research Article
Reviewers, please consider the following questions when reviewing the article:
Research articles should meet the publication criteria for PLOS ONE (http://journals.plos.org/plosone/s/criteria-for-publication).
We ask that reviewers consider the following questions when reviewing an iGEM Research Article:
- Does the study present the results of primary scientific research?
- Are the manuscript and analysis technically sound?
- Are the conclusions supported by the data?
- Does the manuscript adhere to the PLOS Data Policy?
- Is the manuscript presented correctly and well written?
See iGEM Reviewer Instructions for more information.
Please leave your review as a comment below the article.
Development and Characterization of Fluorescent and Luminescent Biosensors for Estrogenic Activity
Ruchi Asthana* (1), Donna Lee (1), Michelle Yu (1), Maxwell R. Telmer (3), Niteesh Sundaram (5), Dominique MacCalla (3), (4), Wei Mon Lu (2), (4), Jordan Tick (5), (6), William Casazza (1), Kenneth Li (1), Diana Marculescu (5), Marcel P. Bruchez (1), (7), Natasa Miskov-Zivanov (5), Cheryl A. Telmer (1)
1 Department of Biological Sciences, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
2 Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
3 Department of Materials Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
4 Department of Biomedical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
5 Department of Electrical and Computer Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
6 School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
7 Department of Chemistry, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America
*Corresponding Author: Ruchi Asthana (rasthana@andrew.cmu.edu)
Author Contributions
Lab Experiments: R.A, D.L, M.Y, M.R.T, D.M, W.M.L, W.C, K.L & C.A.T
Modeling and Device Design: N.S, J.T, M.R.T & N.M
Manuscript Preparation: R.A, D.L, M.Y, M.R.T, N.S & C.A.T
Reviewing: C.A.T, N.M & M.P.B
Editing: D.M, D.M, W.M.L & J.T
Funding Acquisition: D.M, M.P.B, N.M & C.A.T
Abstract
The scientific community has become increasingly aware of the negative effects compounds with endocrine disrupting activity have, for example fish feminization. The EPA has a sophisticated program to analyze chemicals for estrogenic activity, however it is neither portable nor inexpensive enough to enable widespread testing of water supplies. The goal of this study was to create a rapid, sensitive, low cost, and portable biosensor for estrogenic activity. Fluorescent proteins including BFP, GFP, YFP, OFP and RFP were purified and evaluated in an effort to standardize measurements that are useful for modeling and sharing this data across the synthetic biology community. Luciferase proteins from Photinus (firefly), Renilla and Gaussia were also characterized and tested as reporters. The fluorescent proteins, luciferases, and a simplified protocol for protein purification were then shared with a local community lab. An estrogen sensor bacterial strain containing a two plasmid biosensor was designed. The sensor plasmid contains a phage T7 RNA polymerase (T7 RNAP), which is conditionally activated by estrogen binding to the human estrogen receptor ligand binding domain (ER LBD). The sensor plasmid expresses YFP as an indicator of the amount of sensor synthesized. The reporter plasmid encodes a reporter protein (RFP or Gaussia luciferase) under control of T7 RNAP. Upon binding of estrogen and related compounds, the ER LBD alters conformation and reconstitutes the activity of the T7 RNAP, resulting in transcription and translation of the reporter. The RFP biosensor detected a range of 10-100 µ M estrogen and ethinylestradiol. The Gaussia biosensor detected the presence of these two compounds at nM concentrations, but was not sensitive enough to differentiate between the varying concentrations. Modeling of the system helped predict and verify laboratory results. The code along with instructions, software, and parts for creating an inexpensive 3D printed luminometer/fluorimeter were made publically available to promote Do It Youself (DIY) projects.
Financial Disclosure
Funding was generously provided by Carnegie Mellon University, College of Engineering, and Mellon College of Science, Departments of Electrical and Computer Engineering and Biological Sciences, the Molecular Biosensor and Imaging Center, and the Council of Electronic Design Automation and the National Sciences Foundation. Reagents were donated by IDT, RegisTechnologies Inc. and Thermo Fisher.
Competing Interests
The authors declare that no competing interests exist.
Ethics Statement
N/A
Data Availability Statement
All data are fully available without restriction on both our github pages and our website. Links include:
https://github.com/niteeshsun/iGEM2015, https://github.com/rasthana522/iGEM2015
http://2015.igem.org/Team:CarnegieMellon
Introduction
In 2014, the CMU iGEM team initiated the development of STREAM (Sensor That Reports Endocrine Activating Molecules), an inexpensive, fast, and simple sensor to detect molecules in water that bind to an estrogen receptor (http://2014.igem.org/Team:CarnegieMellon). Their approach was (1) to construct a biosensor using biological parts and (2) to take advantage of rapid bacterial growth and simple machinery for signal amplification. The sensor was designed such that the binding of estrogen to a receptor ligand binding domain would result in an intein splicing (Liang et al. 2010); this would cause the production of functional T7 RNAP (Liang et al. 2007). T7 RNAP would then bind to a T7 promoter causing subsequent transcription and translation of red fluorescent protein (RFP), resulting in significant signal amplification. It should be noted that the level of estrogen is reflected by the production of functional T7 RNAP, and reported using the expression of RFP. The susbequent inclusion of yellow fluorescence protein (YFP) on the estrogen responsive sensor plasmid later provided a sensor reference level, making it extremely useful in confirming expression levels. A BioNetGen Model (Faeder et al. 2009) was developed and run in Rulebender (Smith et al. 2012) to quantitatively model the performance of the sensor with the addition of estrogen.
This BioNetGen model predicted that the intein-based sensor would fail due to slow splicing kinetics. This was confirmed by biological data when the sensor did not function in the presence of E2. So inteins were removed and the ER-LBD residues 281-595 (McLachlan et al. 2011) were tested at three positions within T7 RNAP (Appendix). From these experiments, it was determined that 179/180 gave some response. Pieces from McLachlan were subsequently tested and the 312/595 fragment, showed an improved response, and thus it was used in our study. BioNetGen models were modified for these new sensors and fluorescent proteins were quantified to improve accuracy of the model.
Materials and Methods
Restriction enzymes, ligase, phusion polymerase, Ni-NTA agarose beads, and chemicals were supplied by Thermo-Fisher. Oligos and gBLOCKS came from IDT. Octylthioglucoside was manufactured by Amresco (CAS# 85618-21-9). Table 1 shows the plasmids for expressing fluorescent proteins and luciferases used in this study. Additional cloning details can be found in the Appendix.
A. Cloning, bacterial strains, and vectors
We used the Escherichia coli (E. coli) bacterial strain MACH1-T1 (Invitrogen) as the host for cloning. A pSB1C3 (www.igem.org) was first modified so that it could include additional restriction sites. Using overlap PCR, the sequence was introduced between the EcoRI and PstI sites in the following restriction site: EcoRI-NotI-XbaI-SphI-HindIII-BshTI-MunI-SpeI- NotI-PstI. The final product was: 5’ G/AATTC GCGGCCGC T T/CTAGA G GCATG/C CTT A/AGCTT GGCGGGTCA A/CCGGT GGAGGTTCT CAC/AATTGT TA/CTAGTAGCGGCCGCTGCA/G.
A BBa J23115 promoter, and a BBa B0034 RBS were also included in our design. The N and C terminus of T7 RNAP (BBa K145001) were cloned with the ERLBD and inserted between residue 179 and 180 of T7 RNAP. Lastly, the BBa B0034 RBS and BBa K1491004 YFP were added after the stop codon of T7 RNAP. To con- struct the reporter plasmids, pSB3K3 (www.igem.org) was used. A BBa S04423 T7 promoter, BBa B0034 ribosome binding site (RBS), BBa E1010 mRFP1, and BBa K1732003 Gaussia luciferase were followed by a BBa B0016 T7 termination sequence.
Fluorescent proteins and luciferases were ordered as codon-optimized gBLOCKS with the BBa J23100 promoter, BBa B0034 RBS, and BBa B0015 termination sequences. They were then cloned into pSB1C3 to give the following devices (Table 1).
B. Fluorescent protein purification
Fluorescent proteins were isolated to quantify and measure fluorescence. The protocol to extract soluble proteins was optimized for 1.5mL of culture. It was then simplified and placed onto cards for easy access to the methods (Figure 2). These instructions were then tested on naive students to troubleshoot the language and clarity of the steps. Fluorescence was measured with a TECAN M1000 using wavelengths for excitation (nm)/emission (nm) (ex/em). Gain was set at 80 and band- width 5nm (RFP 10 nm). The results were: 399/456 (BFP), 488/509 (GFP), 514/527 (YFP), 548/562 (OFP) and 584/607 (RFP).
C. Fluorescent protein quantitation
The Microassay Procedure for Microtiter Plates protocol from the supplier Bio-Rad was followed (http://www.bio-rad.com/LifeScience/pdf/Bulletin_9004.pdf). A standard curve was generated using Bradford Standard Assay (BSA), and proteins were analyzed by adding reagent to dilutions of the purified proteins. Absorbance was measured using a TECAN M1000 at 595nm.
D. Luciferase localization
To determine where the luciferases were located, cells expressing luciferase were first centrifuged. The supernatant of these samples was subsequently collected and pellets were tested with coelenterazine (50µ M). This mixture was then analyzed in the Bioluminescence mode of the TECAN M1000, using a 100ms integration time.
E. Estrogen Sensor Response
A single colony of bacterial sensor strain containing the sensor plasmid (BBa K1732015) and either the RFP (BBa E1010) or the Gaussia luciferase
(BBa K1732003) reporter was cultured overnight in kanamycin 50µ g/mL and chloramphenicol 34µ/mL. This mixture was then diluted 1/20 and compounds were added. The 17-β -ethinylestradiol (E2) and 17-α – ethinylestradiol (EE2) were made as stocks in ethanol and added 1:1000 into the cultures for concentrations ranging from 100µ M to 10pm. Control cultures received ethanol only to eliminate the carrier effects of other com- pounds. Cells were incubated at 37◦ C with shaking at 300rpm. After 24 hours, 100µ L of culture was analyzed for fluorescence using a TECAN M1000. To detect YFP, an ex/em of 514/527 with a gain of 80 (5nm bandwidth) was used. To detect RFP, and ex/em of 584/607 with a gain of 100 (10nm bandwidth) was used. Triplicate cultures were analyzed; ratios were calculated and graphed using GraphPad Prism software (GraphPad Software). The procedure used to prepare the sample was the same for the Gaussia luciferase biosensor as it was for the RFP biosensor. However, 10µ l of 750µ M coelenterazine (75µ M final) in water was added to the sample 1 minute prior to reading luminescence.
F. Modeling
Two models were created to characterize the estrogen sensor. The first model is based on using red fluorescence protein (RFP) as an intracellular reporter, while the second is based on using Gaussia as an extracellular reporter. The biosensor models were written in BioNetGen Language (BNGL), a rule-based modeling language. Rule-based modeling is a type of modeling in which differential equations are generated from a description of how various biological components and systems interact with one another (Figure 3). For more details about the modeling, please see the Appendix.
The models were run in RuleBender version 2.0.382 (http://visualizlab.org/rulebender/), an interactive design environment which is dedicated to analyzing, visualizing, and debugging BNGL models. A useful feature of rule bender is its ability to graphically display the inter- actions between each of the components in the form of a contact map. The contact map helps the user visualize how each interaction fits together in the overall scheme of the system (Figure 3).
Results
A. Codon Optimized Fluorescent Proteins
In order to provide quantitative data for modeling it is important to be able to relate fluorescent readings with amounts of protein. Fluorescence is a measurement of light that is detected and changed to an electrical signal that is then amplified; values are obtained in relative fluorescent units (RFU). RFU are relative to some background; data collected with respect to RFU are then compared to a control sample to obtain meaningful data. These measurements are not always useful for modeling therefore fluorescent proteins were purified and fluorescence was related to quantity. Figure 4 shows the fluorescence of bacterial cultures expressing fluorescent proteins. As you will notice codon optimized GFP cultures showed 3-4 times more fluorescence than the BBa˙E0040 GFP cultures. YFP, EGFP, BFP and OFP showed decreasing signal, and RFP expressed from the strong J23100 promoter showed more fluorescence than RFP expressed from the weak J23115 promoter (Figure 4A). Figure 4B shows that the amount of protein per culture varied 2-3 fold across fluorescent proteins. Codon optimization produced at least twice as much protein (GFPco vs. E0040). A 3-4 fold change was observed with promoter differences. These numbers can be used in models involving fluorescent proteins. Note that differences in fluorescence are not the same as differences in amount due to spectral properties of the specific fluorescent proteins. If labs were to calibrate their fluorescence readings to protein amount it would allow for standardization of fluorescence data.
B. Luciferases
Fluorescent proteins are very convenient reporters because they do not require a substrate; however, there is significant autofluorescence at the wavelengths used for detection. Therefore we investigated the use of luciferases for reporters. Luciferases that remained in the cytoplasm such as firefly and Renilla were compared to that from Gaussia which is an extracellular luciferase. Additionally, two extracellular domains were tested, one predicted to target the media (CDcel) and the other predicted to target the periplasmic space (PelB).
Renilla and firefly luciferase did not contain a targeting sequence and were localized intracellularly as expected. Gaussia luciferase with the targeting domain derived from the catalytic domain of cellulase (Gao et al. 2015), called CDcel-Gaussia, was found mainly in the media as expected. Gaussia luciferase with the PelB leader sequence (Lei et al. 1988), called PelB-Gaussia, was expected to remain in the periplasmic space however 95% of it was found in the media (Table 2).
Of the coelenterazine using luciferases, PelB-Gaussia expressing cells produced the highest level of light output. The firefly luciferase produced the highest light out- put but uses luciferin, a different substrate, and the concentration required mM instead of uM as for the coelenterazine utilizing luciferases. It is important to note that we were unable to determine the concentration that saturates the firefly culture because our stock solution of luciferin is 30mM and we could not test any higher concentrations. There is a possibility that a higher concentration of luciferin would produce even greater light output. We also had problems with CDcel-Gaussia, which did not produce much light (Figures 5).
C. Response of estrogen sensor
The biosensor is a bacterial cell containing two-plasmids. The sensor plasmid is a high-copy plasmid, which has the ligand binding domain of the human estrogen receptor alpha (ER-LBD) inserted into T7 RNAP and YFP is transcribed from the same message and used for normalization. When the ER-LBD binds estrogen, it causes a conformational change (McLachlan et al. 2011) that brings together the separated domains of T7 RNAP and the activity of the T7 RNAP is reconstituted (Shis and Bennet et al. 2012). The second plasmid is the reporter plasmid and is a low-copy plasmid, which has the T7 promoter driving expression of RFP. When the T7 RNAP is reconstituted upon binding to estrogen, it allows for binding to the T7 promoter on the reporter plasmid, transcription of the RFP mRNA and then translation to produce RFP. Several luciferases were then characterized and the RFP was substituted for a Gaussia luciferase reporter. Two EDCs with high activities are natural estrogen hormones known as 17-β -ethinylestradiol (E2) and a synthetically produced estrogen 17-α -ethinylestradiol (EE2). E2 is a major estrogen used to naturally regulate the female reproductive system and to maintain sexual characteristics. EE2 is derived from E2 and is used mainly as a component of contraceptives. It is important to note that the negative control (labeled as µ M estrogen) contained only 100% ethanol as it was the solvent in which the estrogenic compounds were resuspended in. The relative fluorescent units (RFU) of RFP was ratioed to the RFU of YFP (Figure 6).
The Gaussia luciferase biosensor also shows a response to the addition of two different types of estrogenic compounds, E2 and EE2 (Figure 6). Higher concentrations (100 uM) are detected more strongly whereas concentrations lower than 10uM do not seem to be easily distinguishable. It is important to note that while the readings from the RFP biosensor were very close to zero, making it more difficult to distinguish between the varying concentrations. The Gaussia Luciferase biosensor readings remain above 100, allowing for more sensitive detection among the varying concentrations of EE2.
Discussion
This report describes the development of a fast and inexpensive estrogenic responding biosensor. Fluorescent proteins and luciferases were characterized for use as reporters and an estrogen responsive T7 RNAP was developed, tested and modeled. Additionally, software, a user interface and parts were 3D printed in an effort to construct a portable, low cost luminometer/fluorimeter detection device (for a greater description see igem2015. org/Team:Carnegie_Mellon).
The simple protein purification protocol that was developed can have widespread application as a method to produce standards for fluorescent proteins. This protocol is also very useful as a teaching tool to familiarize scientists of all ages with the concepts and methods involved in protein purification. These cards are in use at the Citizen Science Lab in Pittsburgh, PA, USA (http://www.thecitizensciencelab.org/).
The RFP biosensor and Gaussia luciferase biosensor were both successful in detecting two different estrogenic compounds, E2 and EE2. The intracellular RFP detection of both compounds was similar, with the strongest response at 100µ M and indistinguishable responses at concentrations below 10µ M. The extracellular luciferase reporter detected E2 above 10µ M but not below and detected the presence of EE2 at pM concentrations but could not distinguish between concentrations.
While the two biosensors responded to the presence of estrogen, there is a need to improve sensitivity and test additional estrogenic compounds. This will expand upon the functionality of the sensor as the environment is influenced by many endocrine disrupting chemicals aside from estrogen. When used in industrial field work to detect concentrations in the environment, amounts of EDCs are very low due to the large bodies of water they are found in (ppt or fM concentrations). Increased sensitivity will allow for more accurate readings and inference of potential effects they may have. To further increase the efficiency of the biosensor for industrial purposes, a device should also be designed to host the biosensor to more conveniently detect EDCs and generate real-time measured. Taking advantage of the robustness of T7 RNAP it would be of interest to utilize an in vitro RNA output system to have a more rapid test.
Appendix
Appendix files can be accessed here: http://staging-blogsplosorg.kinsta.cloud/collections/files/2016/08/iGEM-2015-Appendix-Carnegie-Mellon.docx
References
[1] Bistan M, Podgorelec M, Marinek Logar R, Tiler T. Yeast Estrogen Screen Assay as a Tool for Detecting Estrogenic Activity in Water Bodies. Food Technol Biotechnol. 2012;50(4):427-433.
[2] Brenzel S, Kurpiers T, Mootz H. Engineering Artificially Split Inteins for Applications in Protein Chemistry: Biochemical Characterization of the Split Ssp DnaB Intein and Comparison to the Split Sce VMA Intein . Biochemistry. 2006;45(6):1571-1578.
[3] BioNetWiki [Internet]. Bionetgen.org. 2016. Available from: http://bionetgen.org/index.php/Citation˙guidelines
[4] Endocrine Disruption — US EPA [Internet]. Epa.gov. Available from: https://www.epa.gov/endocrine-disruption
[5] Faeder JR, Blinov ML, Hlavacek, WS. Rule-based modeling of biochemical systems with BioNetGen, Methods Mol. Biol. 2009; 500, 11367.
[6] Gaido K, Leonard L, Lovell S, Gould J, Baba D, Portier C et al. Evaluation of Chemicals with Endocrine Modulating Activity in a Yeast-Based Steroid Hormone Receptor Gene Transcription Assay. Toxicology and Applied Pharmacology. 1997;143(1):205-212.
[7] Gao D, Wang S, Li H, Yu H, Qi Q. Identification of a het- erologous cellulase and its N-terminus that can guide re-combinant proteins out of Escherichia coli. Microb Cell Fact. 2015;14(1):1-8.
[8] Hampl R, Kubtov J, Strka L. Steroids and Endocrine Disruptors History, Recent State of Art and Open Questions. The Journal of Steroid Biochemistry and Molecular Biology. 2016;155:217-223.
[9] Lei S, Lin H, Wang S, Wilcox G. Characterization of the Erwinia carotovora pelA gene and its product pectatelyase A. Gene. 1988;62(1):159-164.
[10] Liang R, Liu X, Liu J, Ren Q, Liang P, Lin Z et al. A T7-expression system under temperature control could create temperature-sensitive phenotype of target gene in Escherichia coli. Journal of Microbiological Methods. 2007;68(3):497-506.
[11] Liang R, Zhou J, Liu J. Construction of a Bacterial Assay for Estrogen Detection Based on an Estrogen-Sensitive Intein. Applied and Environmental Microbiology. 2011;77(7):2488-2495.
[12] McLachlan M, Katzenellenbogen J, Zhao H. A new fluorescence complementation biosensor for detection of estrogenic compounds. Biotechnol Bioeng. 2011;108(12):2794-2803.
[13] niteeshsun/iGEM2015. GitHub. 2016. Available from: https://github.com/niteeshsun/iGEM2015
[14] Rich R, Hoth L, Geoghegan K, Brown T, LeMotte P, Simons S et al. Kinetic analysis of estrogen receptor/ligand interactions. Proceedings of the National Academy of Sciences. 2002;99(13):8562-8567.
[15] Routledge E, Sumpter J. Estrogenic activity of surfactants and some of their degradation products assessed using a recombinant yeast screen. Environmental Toxicology and Chemistry. 1996;15(3):241-248.
[16] Schaerli Y, Gili M, Isalan M. A split intein T7 RNA polymerase for transcriptional AND-logic. Nucleic Acids Research. 2014;42(19):12322-12328.
[17] Shis DBennett M. Library of synthetic transcriptional AND gates built with split T7 RNA polymerase mutants. Proceedings of the National Academy of Sciences. 2013;110(13):5028-5033.
[18] Smith A, Xu W, Sun Y, Faeder J, Marai G. RuleBender: integrated modeling, simulation and visualization for rule-based intracellular biochemistry. BMC Bioinformatics. 2012;13(Suppl 8):S3.
[19] Team:Carnegie Mellon 2015.igem.org. Available from: http://2015.igem.org/Team:Carnegie˙Mellon