PUF, The Magic RNA Binding Protein: Programmable RNA Binding Protein with Custom Functions

Abstract: RNA has characteristics that are important in human gene expression (i.e. alternative splicing of mRNA, noncoding RNA). Therefore, a modular RNA binding protein is an invaluable tool for gene regulation. The PUF domain of human PUM1 gene contains eight tandem repeats, each recognizing one of the four nucleotide bases. In theory, a PUF protein can be programmed to recognize any 8-nt ssRNA sequence. Here we demonstrate that PUF can be tethered with other functional domains for applications in E. Coli. Specifically, we show that a PUF/endonuclease fusion protein acts as RNA scissors, silencing gene expression through site specific mRNA cleavage. PUF/endonuclease activity was also characterized with an RNA scaffold in order to test PUF’s binding specificity. PUF biobricks offer a wide range of possible functions including gene expression modulation and scaffolding of metabolic pathways.

RNA Scissors Experimental Design In designing our project, we based our quantitative tests on fluorescence measured by a fluorescence plate reader. Our constructs were created in ways that best suit providing evidence for our hypothesis of the PUF-PIN fusion protein showing endonuclease activity. Generally, our results were collected from quantifying two main PUF-PIN fusion protein types, a wild type and a mutant type, in different conditions. These two fusion proteins had recognition sites that differed by one base pair.

In the figures above we show our PUF-PIN constructs. The top row shows the PUF-PIN and mutant *PUF-PIN complexes that have been translated and transcribed. The middle row shows how PUF and *PUF both bind their own specific binding site on mRNA. The third row of figures shows the action of PIN endonuclease after PUF or *PUF has bound to its respective mRNA base recognition site. The endonuclease will cut the mRNA disrupting the reporter gene.

RNA Scaffold Overview In order to provide a direct application for the RNA binding abilities of PUF, a RNA scaffold was designed with the idea of serving as a platform for an enzyme conveyor belt. The array of enzymatic pathways which could be enhanced by a scaffold are numerous, though, we projected to increase efficiency and production of a resveratrol derivative called piceatannol. The project consists of a couple parts, each a proof of concept and build-up of previous ones. The start of the project consisted of designing a RNA scaffold which is best tailored to PUF binding in a spatially specific manner. Once the scaffold was designed, synthesized, and purified it was important to show not only that PUF can bind specifically to its designated sites, but that the scaffold can support a concept such as a biological conveyor belt. One assay which was designed to prove this was incubation of the RNA scaffold with a non-specific endonucleases bound to PUF. The length and number of digested RNA parts would prove that PUF was binding specifically and appropriately to the designated sequences. Another assay would include tethering a split-fluorescent protein to wild-type and mutant PUF. An in-vitro gel-shift assay, or EMSA, would once again prove that PUF is binding the appropriate sites. More importantly, an in-vivo experiment which shows fluorescence with presence of the scaffold and darkness without the scaffold would prove efficient enzymatic pathways could be achieved.

[3]: Yeming Wang, Laura Opperman, Marvin Wickens, and Traci M. Tanaka Hall. Structural basis for specific recognition of multiple mRNA targets by a PUF regulatory protein. PNAS 2009 ; published ahead of print November 9, 2009,doi:10.1073/pnas.0812076106

RNA Scaffold Design

Fig. 3

Fig. 5 The sequence in Fig. 5 is the DNA sequence coding for the d0 scaffold. The first highlighted potion is the T7 promoter followed by the MS2 binding site. The next highlighted region shows the PP7 binding site followed by the T7 terminator.

Fig. 3[1] The design of our RNA scaffold was based upon a scaffold created by the Pam Silver research lab at Harvard. This group was the only one to develop such a construct and prove its effectiveness. Our scaffold was built upon the Silver group’s scaffold in order to serve as the application for our own RNA binding proteins. The d0 scaffold has two hairpin loops with binding sites for two distinct RNA binding proteins, MS2 and PP7. Although the Silver group developed even more complicated scaffolds, we decided to work with the simplest (d0) as a way to prove not only that PUF can bind specifically, but that the scaffold can be used for efficient production of compounds in bacterial cells.

Fig. 6 The scaffold was further modified after research suggested that PUF binds best to nucleotides with an angle of curvature of approximately 20o turn per repeat[3]. The hairpin loops were changed in order to accommodate this from 8 nucleotides to 18 nucleotides achieving a 20o turn per nucleotide effect. Sequences of the stem loop were further modified in order to keep GC content from being too high and to make a more stable structure. This DNA sequence was synthesized through IDT’s miniGENE option.

Fig. 4

Fig. 4 The final structure of the scaffold used in the UIUC iGEM 2012 project.

Fig. 7[1] The figure above shows an important theoretical concept of the RNA scaffold. Assuming that the RNA binding proteins bind specifically, a spatial control of “cargo” can be made to serve various functions. In our project, we envisioned enzymes of the piceatannol pathway to churn out product due to the spatial proximity of one enzyme next to another, yielding increased reaction kinetics. However, in order to prove that such a concept is possible, further assays must be executed.

Fig. 8[1] A gel-shift in-vitro assay as the one pictured would properly demonstrate that distinct RNA binding proteins are binding to the scaffold and thus proper scaffold functioning. With addition of the scaffold, each separate binding protein causes a change in overall size resulting in a different band from each protein by itself. Due to assay shown being ran on a native gel, secondary structures and changes in electronegative affinities might explain the unusual effect of larger constructs being localized further down the gel.

Fig. 9[2] This data was received from Prof. Wang’s lab in UNC which characterized PUF-PIN (PIN, a non-specific endonuclease) functioning at various pH levels. A similar assay with our scaffold would show specific binding of PUF to its destined sites on the scaffold due to the presence of RNA fragments of expected lengths.

[1]: Organization of Intracellular Reactions with Rationally Designed RNA Assemblies. Camille J. Delebecque, Ariel B. Lindner, Pamela A. Silver, and Faisal A. Aldaye. Science 22 July 2011: 333 (6041), 470-474.Published online 23 June 2011 [DOI:10.1126/science.1206938]

[2]: Engineering RNA Endonucleases with Customized Sequence Specificities. Rajarshi Choudhury, Daniel Dominguez, Yang Wang and Zefeng Wang. Department of Pharmacology and Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, April 11, 2012

Fig. 10/11 1 L culture incubated at 37oC till 0.5 nm optical density after inoculation with 5 mL of overnight culture. 2 mM IPTG induction for 2 hours. His-Tag Ni-NTA purification, centrifuged with Millipore 30kDa cutoff ultracentrifuge tubes. SDS-PAGE 6% native gel stained with Coomassie Brilliant Blue for 1 hr. and destained with H20 for 1 hr.

Fig. 12 10 well 10% 1mm urea denaturing acrylamide gel, post 2μg/mL EtBr staining for 20 min., destaining for 20 min. 1X TBE buffer, 120V for 55 min. 3 mM MgCl2 ions used in all lanes. RNA samples were denatured for 3 min. at 95oC and then let to fold at 4oC for 5 min. Addition of 1 μL of annealing buffer (EDTA/Tris) before addition of protein. 30 min. incubation time with protein at 37oC and then addition of 2X 80% formamide/EDTA to stop the reaction. 66 nM concentration of RNA, 305 nM concentration of WT, 6-2/7-2 PUF-PIN in first four lanes, 152.5 nM concentration of WT, 6-2/7-2 PUF-PIN in last four lanes. RNA Scaffold Conclusion

In order to start the endonuclease assay mentioned in the design portion, a few necessary components needed to be collected. Both WT & 6-2/7-2 PUF-PIN were purified by Ni-NTA columns and then quantified through BCA analysis. Although the BCA analysis gave quantitative numbers, the SDS-PAGE gel showed large amounts of impurities in the elution fractions. Therefore, the concentrations determined by the BCA analysis (3.05 μM for WT PUF-PIN, 0.46 μM for 6-2/7-2 PUF-PIN) should be taken as approximate due to the actual purity of the desired proteins being roughly 30-40%. Even though the purity of the proteins is sub-optimal, the RNA endonuclease enzymes used are multiple turnover, meaning they can cleave multiple RNA strands rather than cleaving one and ending the reaction. Therefore, the low concentrations of protein can be used in further assays. Before the endonuclease assay could be run, the RNA scaffold designed needed to be synthesized which was done through in-vitro transcription with the Invitrogen MEGAscript® T7 Kit. The expected length of the scaffold was 126bp and the NEB Low Range ssRNA Ladder showed homology with the 150bp band. However, with presence of only one band and approximate length homology, it can be concluded that the expected scaffold construct was produced. The endonuclease assay in Fig. 12 showed promising results of specific PUF binding. There was cleavage of the scaffold seen most distinctly in wells with WT PUF-PIN and when both proteins, WT & 6-2/7-2, were present. 6-2/7-2 PUF-PIN showed equal efficacy in cleavage, yet there was smearing of RNA which might suggest unspecific cleavage and 6-2/7-2 PUF binding. From this observed effect, it could be said that certain derivatives of PUF are more specific to their designated binding sequence than others. In order to make this a more conclusive assay, a negative control such as the non-specific endonuclease (PIN) by itself should be used as comparison to endonucleases bound to PUF. However, due to the endonuclease being non-specific and data showing presence of single bands, not smears, it can be said that PUF provides specificity to these otherwise non-specific endonucleases and specifically binds RNA.

Human Practices Project Overview Synthetic biology is a relatively new field, but it’s rapidly growing and has the potential for a large societal impact. UIUC iGEM fully realized this when executing our human practices project, and responded with a tripartite plan. First we examined our own foray into synthetic biology. By doing media research and examining formal reports on the ethics of synthetic biology, we began to grasp the public perception of genetically engineered organisms. We summarized our research in a mini-database on our wiki, with the hopes of helping inform the public. Second, we went out into the Champaign-Urbana community for outreach. We focused on explaining the basics of genetic engineering and DNA. In our biggest event, we presenting at UIUC’s 2012 Engineering Open House. We presented a poster on DNA, showed off some fluorescent E. coli, and showed visitors how to extract their own DNA. Third, we continued our public outreach by producing 2 videos on GMOs – one on transgenic corn and the other on gene therapy. It is our hope that these will prove thought provoking and serve as educational materials for those more interested in genetic engineering and synthetic biology. Our videos are available on YouTube via our team wiki.

Institution: University of Illinois Urbana-Champaign Team members: Aditya Malik, Anthony Chau, Asha Kirchhoff, Bob Chen, Cara Schornak, Divya Tankasala, Hsiao-Han Chen, Isiah Ramos, Uros Kuzmanovic Instructions: Chris Rao, Courtney F. Evans, Kaustubh Bhalerao, Yong-Su Jin Advisors: Ahmet Badur, Brad Evans, Kori Dunn, Todd Freestone Division: Americas East

PetroBrick Characterization Overview As a side project, we decided to characterize a previous team’s work on an existing biobrick. For that purpose, we chose to characterize the University of Washington’s Petrobrick. The Petrobrick, once transformed into E. coli, acts as a microbial alkane production pathway. Two enzymes are co-transformed to create this biobrick: Acyl-ACP Reductase (AAR - Bba_K90032) and Aldehyde De-Carbonylase (ADC - Bba_K90031). AAR reduces cellular fatty acyl-ACP from bacterial fatty acid via into fatty aldehydes. ADC then removes the carbonyl group on the fatty aldehyde, resulting in an odd number alkane chain one carbon shorter than the original Acyl-ACP fatty acid. In turn, both of the enzymes convert fatty acids into an odd number alkane by means of a constitutive protein expression plasmid.

PetroBrick Characterization Design It was noted that the alkane production is enhanced when growing expression strains using the optimized growth conditions developed by the 2011 University of Washington team, so we followed the protocol to the best of our ability. After analyzing their results, we decided to reproduce the experiment specifically to test for the production of C15 alkanes, which were the most abundant. In order to do so, four samples of empty E. coli cells grown in TB media were injected with known concentrations of C15 alkanes (obtained from Sigma-Aldrich). They were used as control samples with the corresponding concentrations: 1 mg/L, 10 mg/L, 50 mg/mL, and 100 mg/L. After injecting the cells with the known C15 alkanes, the cells were then incubated for 48 hours in M9-Glucose media to ensure nothing changed in the development of the control samples. After incubation, ethyl acetate was used to extract 200 uL of the alkane samples to be analyzed with GCMS. Gas-Chromotography Mass-Spectometry (GCMS) was used to create a standard curve of the four known concentration and their corresponding peak areas. For the actual samples, the Petrobrick-transformed DH5α E. coli cells were grown in TB media overnight. After growth, the cells were spun down and re-suspended in M9-Glucose media for 48 hours. Ethyl acetate was used to extract the produced alkanes. 200 uL of each of the samples were used for GCMS analysis.

PetroBrick Characterization Conclusion The average yield for C15 alkanes as determined by our results was 160.2 mg/L. Our maximum yield was 190.6 mg/L. The average C15 alkane yield for the UW team was 160.3 mg/L. Based on our results, we were able to successfully reproduce the results from the UW iGEM team’s work on the Petrobrick, effectively proving its function.

Characterization Results

Fig. 15. Standard curve created from the results of GCMS analysis of the four controlled known concentrations of C15 alkanes and the corresponding peak areas.

Figure 16. Concentration yields of C15 alkanes from the four experimental samples based on standard curve measurements at corresponding retention time for pentadecanoic acid (C15 alkane)

Fig. 13. University of Washington (2011). Fatty Acid Biosynthesis Pathway. [Image]

Fig. 14. University of Washington. (2011). Diagram showing the process of alkane extraction. [Image].

In order to test site-specific cleavage of the two fusion proteins, we propose to match both the PUF-PIN and *PUF-PIN proteins with a non-specific binding site with a YFP reporter for control experiments as shown above.

Fig. 1 Our primary test was done on wtPUF-PIN protein. With regards to our hypothetical results, our experimental fluorescence measurements were substantially faithful to our predictions. Looking at columns 5, 6, 7, and 8, our observations appear to be in line with what we have predicted in introducing the wild type PUF-PIN to constructs containing a control recognition site and a specific recognition site. Columns 1 and 2 are negative fluorescence controls while columns 3 and 4 are positive fluorescence controls measured without the influence of any sort of binding site. Note: PUF-PIN was cloned in pBAD and the YFP+control and YFP+PBS (PUF binding site) were cloned in protet. Fig. 2 In order to facilitate more reliable results, we also tested our constructs with the expression of an RFP, mCherry. If in the case that confounding factors influenced our results in our experiments using YFP, we could minimize inaccurate influences on our data by using a different reporter.

RNA Scissors Conclusion The YFP fluorescence data has shown that the PUF-PIN domain has the ability to silence YFP expression. However, YFP expression was silenced regardless of the binding site type present in the YFP expression constructs. Assessing the specificity of PUF requires further experimental research. The mCherry data suggests that it is likely due to a problem with the YFP reporter, but due to the fact that there was no control binding site cloned into the mCherry expression construct, more experiments will be required for conclusive evidence. The specificity and customizability of our gene silencing through RNA scission both have extensive ongoing experiments.

Fig. 1 Fig. 2 Future Directions: Enzymatic Assembly Earlier this year, research at the Kee-Hong Kim lab of Purdue University had preliminary evidence[4] showing that a trans-stilbene compound, piceatannol, had an ability to inhibit the development of human adipose cells. The mechanism is based around the idea that piceatannol interacts with a preadipocyte's (immature fat cell) insulin receptors in such a way that suppresses it's growth into a mature adipose cell. Piceatannol is a metabolite of resveratrol, a compound currently under investigation for possible anti-cancer properties. Piceatannol differs from resveratrol by one hydroxyl group on one of the aromatic rings. Piceatannol is currently very costly to synthesize. On the advent of such a discovery, we felt that if we were to engineer a pathway to optimize the production of piceatannol from cheaper substrates through the utilization of our PUF and RNA scaffold projects, we could show the versatility of our PUF toolkit working with an RNA scaffold.

As shown in our concept drawing above, the sequence of enzymatic activity begins at Tyrosine Ammonia Lyase (TAL)[5], which converts the naturally present amino acid in E. coli, tyrosine, into p-coumaric acid. P-coumaric acid along with malonyl-CoenzymeA are converted into resveratrol by the CoA-Ligase:Stilbene Synthase (4CL:STS) fusion protein[6]. Lastly, resveratrol is metabolized into piceatannol by Cytochrome P450 BM3 (BM3)[7]. A substrate could be hypothetically processed by these sequential proteins, resulting in a molecular, in vivo assembly line in E. coli. We hope to join the RNA Scisssors work and the RNA Scaffold work together with this concept to create a kinetically optimized small-scale model for piceatannol production in vivo.

Fig. 17 A concept drawing of the enzymatic assembly line in vivo

Fig. 18 Chemical specifics of the enzymatic assembly line

[4]: Piceatannol, natural polyphenolic stilbene, inhibits adipogenesis via modulation of mitotic clonal expansion and insulin receptor-dependent insulin signaling in early phase of differentiation. Kwon JY, Seo SG, Heo YS, Yue S, Cheng JX, Lee KW, Kim KH. Department of Food Science, Purdue University. J Biol Chem. 2012 Mar 30;287(14):11566-78. [5]: Received from Dr. Fevzi Daldal's lab of the University of Pennsylvania. [6]: Part received from the Parts Registry; created by 2008 Rice University iGEM team. [7]: Received from Dr. Chul-Ho Yun of Chonnam National University, South Korea