Our goal is to produce light in the most ecological way possible. Due to the energy use, greenhouse gas emissions, and environmental effects of electrical lighting, we consider biological light production to be the perfect candidate as a sustainable alternative. Dinoflagellates were our original inspiration, as they are responsible for the famous glowing blue waves that initially attracted our attention. After extensive literature review, we had a much better understanding of the current knowledge of bioluminescence, and began planning the design of our engineered biological system.

Bioluminescence was our go-to solution as it allows us to generate light in a very efficient manner. The phenomenon utilizes an enzyme called luciferase along with a substrate called luciferin to emit photons. The oxidation of luciferin into a high-energy intermediate results in light emission upon decomposition. This process is nearly 100% efficient with negligible heat production, resulting in a “cold light”.¹ Other methods of producing light not involving combustion nor electricity are:

BIOLUMINESCENCE

Being able to control bioluminescence has been in the minds of many groups in recent decades. Many previous studies have expressed luciferases from a variety of species, modified them, and enhanced them beyond their natural counterparts. The problem is that synthesized luciferin must be externally added to these organisms to produce light. This is not only costly but also highly inefficient. That’s why we opted for an autobioluminescent system, meaning that luciferin is produced inside the organism without the addition of external substrates, allowing it to glow by itself. There are only two currently elucidated pathways for the complete biosynthesis of luciferin.^{2, 3}

Fungal Bioluminescence

Fungal bioluminescence is the result of a metabolic pathway referred to as the Caffeic Acid Cycle.

The system starts with the synthesis of hispidin from caffeic acid catalyzed by hispidin synthase (HispS). Then, hispidin is hydroxylated by hispidin-3-hydroxylase (H3H) into the fungal luciferin, 3-hydroxyhispidin. Fungal luciferin is then oxidized by luciferase (Luz), producing a high-energy intermediate that emits photons when degraded into caffeylpyruvate (fungal oxyluciferin). The cycle starts anew when caffeylpyruvate hydrolase (CPH) recycles oxyluciferin back into caffeic acid, generating pyruvic acid as a byproduct. Additionally, HispS requires post-translational modification from 4′-phosphopantetheinyl transferase (NpgA) in order to function.

The original study that successfully expressed fungal bioluminescence in a heterologous chassis utilized genes for the caffeic acid cycle from the fungi Neonothopanus nambi and the NpgA gene from Aspergillus nidulans. This approach was expanded upon by Mitiouchkina et al. in 2020, demonstrating successful heterologous expression of bioluminescence in plants.⁶ Plants naturally produce caffeic acid as part of the lignin and flavonoid synthesis pathways. This allows for the creation of light-emitting plants without the addition of external substrates through expression of fungal bioluminescence genes.

Chassis Selection

Once we selected a system that is efficient at producing light and doesn't require electrical energy nor heavy metals, we needed to find other ways of making it “green”. One promising method for the creation of sustainable biological systems is phototroph engineering. Phototrophs are able to carry out complex biological processes while fixing carbon dioxide, all while being powered by the sun! One prominent choice of phototrophic chassis is higher plants, although they possess several drawbacks that would make a bioluminescent plant lighting system disadvantageous.

Plant drawbacks:

• Although chloroplast engineering is possible, propagation and cell culture are a concern
• Occupy fertile area
• May require constant water supply
• GMO plants are very restricted in many countries

Therefore we examined other options. Phototrophic microorganisms bring with them unique advantages. Growth in liquid media allows for added versatility by adapting to a container’s shape, and easy containment of said liquid media allows for improved secondary containment for better biosafety and a wider range of applications. After deciding upon phototrophic microorganisms as a chassis, we determined our final goal of having a versatile light source in the form of a growable luminous liquid that can be incorporated in any environment as a sustainable light source, artistic medium, or urban design feature.

Microalgae and cyanobacteria were our finalists. Although we considered the advantages of expressing eukaryotic genes in eukaryotic microalgae, we opted for cyanobacteria for two main reasons. First, they have a shorter doubling time than common microalgal chassis.⁷ Second, we aim to control our bioluminescent system so that it only glows when needed. Synechocystis sp. PCC 6803, a species of cyanobacteria, are model organisms for the study of circadian rhythms and the engineering of systems under circadian control. By demonstrating that this system works in cyanobacteria, we can continue to try to control the system with confidence through circadian promoters.

Synechocystis sp. PCC 6803 is an ubiquitous chassis in cyanobacterial synthetic biology, earning the nickname “green E. coli”. We decided upon this organism as our phototrophic, prokayotic chassis, making this the first attempt at expressing the entire fungal bioluminescence pathway in prokaryotes. Upon success, we can start trying more interesting and challenging chassis such as saltwater strains.

We obtained a non-motile glucose-tolerant strain from the Hallam Lab in UBC Vancouver. However, we quickly ran into a problem. Synechocystis sp. PCC 6803 doesn’t produce caffeic acid which is the precursor of fungal luciferin necessary for the production of bioluminescence. We found a research group that was able to engineer this strain to synthesize caffeic acid.⁸

We considered this a great finding as our system recycles caffeic acid, and it's possible that it will start accumulating caffeic acid, feeding the bioluminescence reaction without the need for external substrates. We managed to contact this group from the University of Copenhagen and they kindly sent their strain overseas to us!

Gene Selection

Inspired by the previous attempts of heterologously expressing fungal bioluminescence, we decided to take the same strategy to increase our chances of success. Sequences for the cDNA of the 4 enzymes (HispS, H3H, Luz, and CPH) from N. nambi and NpgA from A. nidulans were obtained from NCBI.

We explored possible methods of improving the bioluminescence system. Some of our members have experience with ecology, and suggested we compare literature regarding the relative light intensities of various bioluminescent fungi. We were informed by mycologists that the fungus that produces the most visible light was Panellus stipticus, which can be found in Canada as well. All bioluminescent fungi share common bioluminescence genes in a highly-conserved gene cluster, only differing in certain mutations, duplications, and the presence of p450-coding genes that may contribute to the process.⁵ There’s no certainty at the moment regarding the factors responsible for its increased brightness, but we decided to express the luciferase gene from P. stipticus, as the enzyme’s activity plays a significant role in the intensity of the light emitted.²

On the side, other members dug into the literature to see what other people have accomplished in improving the system. In the presence of oxygen, fungal luciferase is able to oxidize artificial luciferin analogs, producing different wavelengths and intensities of light.¹⁰ Another potential method of altering wavelength is through site-directed mutagenesis of the luciferase gene. One paper explored the effects of site-directed mutagenesis, and determined that most point mutations resulted in a loss of enzyme function. However, a V49A mutation resulted in a 1.7 times increase in luminescence intensity, demonstrating the potential for improvement of the bioluminescent system through mutagenesis. This paper also investigated the structure of luciferase, speculating the existence of an N-terminal transmembrane domain. The existence of this region was supported by deletions of up to 37 N-terminal amino acid residues having no effect on luciferase activity. Inspired by the results of this paper, we decided to express N. nambi luciferase mutants possessing the V49A substitution and 33 N-terminal amino acid deletion in addition to the wild type N. nambi and P. stipticus luciferase genes.

Due to how recently the fungal bioluminescence pathway was elucidated, the enzymes are not well-characterized. For example, fungal luciferase belongs to a novel, undescribed protein family lacking any known homologs. By expressing different variants of the luciferase gene along with the rest of the bioluminescence pathway in a novel chassis, we hoped to expand upon the understanding of heterologous expression of fungal bioluminescence.

MoClo

After determining the genes to be expressed in Synechocystis sp. PCC 6803, we designed our plasmid constructs. We settled on expression of all fungal bioluminescence genes in single multi-gene constructs for ease of transformation and expression in Synechocystis sp. PCC 6803. Additionally, we utilized a variety of promoters and terminators to minimize the chances of homologous recombination in our constructs. When considering our plasmid assembly process, we compared the advantages and disadvantages of traditional BioBrick cloning with Type IIS golden gate cloning. In comparison to traditional BioBrick cloning, golden gate cloning offers increased modularity and convenience for large assemblies due to its use of Type IIS enzymes that cut outside their recognized restriction sites, thus allowing for custom design of fusion sites for large one-pot, one-step reactions of many inserts simultaneously. The large size of our multi-gene constructs, the large number of inserts in said constructs, the variety of plasmid designs, and the future plans for expression with circadian promoters were all factors that contributed to our decision to use golden gate cloning. This decision was further reinforced by the convenience of the CyanoGate Kit, a plasmid kit with a large variety of promoters, terminators, and backbones for the engineering of cyanobacteria through golden gate cloning. CyanoGate specifically utilizes a MoClo standard, which establishes a specific syntax for the custom fusion sites of assembling MoClo parts together. Individual basic biological parts in plasmids flanked by Type IIS restriction sites are classified as level 0 parts, assembly of multiple level 0 parts into a single functioning transcriptional unit produces a level 1 part, and assembly of multiple level 1 transcriptional units into a multi-gene construct results in a level 2 part. This assembly process also required vector backbones for level 0 and level 1 assemblies from the MoClo Toolkit.

The cDNA sequences acquired from NCBI were then codon-optimized using IDT’s integrated software for expression in Synechocystis sp. PCC 6803, and illegal restriction sites for Type IIS assemblies were removed through silent mutations using Benchling.

Circadian Control

The modularity of MoClo allows for the easy integration of circadian promoters into our designed plasmid constructs. Inducing the expression of bioluminescence through circadian control would allow for greater manipulation of bioluminescence gene expression to save on resources and possibly integrate a “charging system” for when light is not available by storing energy and releasing it later. Although we took these possibilities into consideration in our design process, they were far outside the scope of our project, and were thus not attempted.

After determining our genes, assembly method, and plasmid designs, we modified the sequences of our codon-optimized genes to include 5’ and 3’ flanks for compatibility with golden gate assembly. These flanks included BbsI restriction sites and fusion sites for assembly of linear fragments into a level 0 plasmid. Furthermore, they also included BsaI restriction sites and fusion sites for assembly of level 0 plasmids into level 1 transcriptional units. Following the addition of these flanks, the genes were synthesized by IDT.

All CyanoGate and MoClo parts were formatted as level 0 plasmids compatible for assembly with our synthesized fragments. Following the extraction of these parts, we assembled our linear fragments from IDT in the pAGM9121 universal level 0 acceptor vector from the MoClo Toolkit utilizing the Type IIS restriction endonucleases BbsI. The assembled level 0 plasmids contained BsaI restriction sites and the MoClo standard fusion sites for coding sequences that were incorporated into the IDT-synthesized fragments.

These level 0 plasmids were assembled into level 1 transcriptional units with BsaI, combining promoters, 5’ UTRs, and terminators from CyanoGate with backbones from the MoClo Toolkit, and our level 0 gene constructs. The level 1 acceptor backbones utilized produce sticky ends that are complementary to each other when the level 1 parts are digested with BbsI in a level 2 assembly. These sticky ends are designed for up to 7 transcriptional units in a level 2 construct, and a dummy insert is included to produce sticky ends for assembly of our 5 gene constructs. We produced multi-gene constructs with each variation of luciferase. The transcriptional units were assembled in backbones in the order of Luz, CPH, H3H, NpgA, HispS. This was chosen based on the conformation of the genes found in N. nambi's chromosome. We decided to go for a promoter driving the expression of each gene to help us troubleshoot in the future and also to overexpress as much of the protein as possible. This system of gene expression was inspired by previous studies with heterologous expression in cyanobacteria.

Every step of our build process was screened and sequence-verified through antibiotic selection, blue-white screening, colony PCR, and analytical digestions.

Our first goal is to demonstrate this system is feasible in cyanobacteria. So far there have been no attempts to express HispS, NpgA, nor CPH in prokaryotes while Luz and H3H have successfully been expressed with demonstrated enzymatic activity.^{9, 10}

Due to time constraints we decided to not transform cyanobacteria. This process can take weeks due to their slow growth. Meanwhile, we used our plasmid-carrying E. coli to carry out some tests.

The first test was to see if there was any bioluminescence produced at all. No measurable bioluminescence was observed in transformed DH5a E. coli grown in varying concentrations of caffeic acid-supplemented media. However, BL21 E. coli are generally used for measuring gene expression, and DH5a E. coli are not as suitable for heterologous gene expression. We also tried adding varied amounts of caffeic acid and liquid media cultures in a black 96-well plate donated by Nyoka Design Labs.

No light was observed when looked in a black box (as our plate reader only analyses fluorescence). We also conducted this experiment with low growth of our level 2 transformed E. coli, because as mentioned above, time constraints didn't allow for further growth. Additionally, the E. coli were incubated at 37 ^oC, and fungal luciferase is known to inactivate at temperatures surpassing 30 ^oC which would greatly explain our results considering we did not allow the cells to grow enough under 30 ^oC.

Extraction and visualization of our final construct from transformant colonies through gel electrophoresis revealed a plasmid construct greater than 10 kilobases (the length could not be precisely determined as the plasmid construct was larger than all available DNA ladders).

E. coli level 2 transformant colonies inoculated in liquid media demonstrated slower growth than liquid media inoculated with level 1 transformants and wild type E. coli, suggesting that the E. coli transformed with all of the bioluminescence genes potentially experienced metabolic constraints as a result of expressing the bioluminescence genes regulated by strong, constitutive promoters.

This is a potential indication of bioluminescence gene expression, but further analysis and expression tests are necessary for verification of expression.

Our level 1 constructs have no apparent effect on E. coli growth while our level 2 and level T seem to reduce the doubling time of transformed strains. This may indicate that the expression of these enzymes combined might be intervening with some process in the physiology of E. coli. This may not be the case for cyanobacteria but it's certainly worth consideration in the future.

Future iterations of the engineering design cycle have numerous opportunities to expand upon our engineered biological system by exploring aspects of engineering bioluminescence outside the scope of this year’s project. Potential areas to explore include the incorporation of circadian promoters, the expression of fungal bioluminescence genes spread across multiple smaller plasmid constructs rather than one large construct, and the transformation of caffeic acid-producing Synechocystis sp. PCC 6803 with fungal bioluminescence genes through natural transformation or triparental mating.

Further analytical techniques could be utilized for characterizing the expression of fungal bioluminescent genes in E. coli or Synechocystis sp. PCC 6803.

We tried conducting an SDS-PAGE analysis but had no results in our WT control and due to time constraints the experiment was not repeated. Performing this test would allow us to see if the enzymes are being successfully translated. SDS-PAGE analysis would allow us to compare proteins present in WT E. coli, the level 1 transformants expressing all the 5 single enzymes individually (as well as the other 3 versions of Luz), and the level 2 transformants. We would expect a strong band representing each additional enzyme in its corresponding level 1 strain as well as a combination of all of them in the Level 2 strain.

Protein extraction and fractionation by column chromatography and substrate assays would further validate protein expression. For measuring the expression of heterologous bioluminescent genes in E. coli, future iterations of the engineering cycle could utilize BL21 E. coli rather than DH5a E. coli as we would get much better expression of these exogenous enzymes.

We conducted assays with caffeic acid as well and we noted no bioluminescence. As mentioned before, this could very well be lack of enough cell density, a good chassis for overexpression of heterologous genes, and the fact that we never attempted these experiments after leaving the cells to grow at temperatures under 30^oC for sufficient time. In our next iteration of the engineering cycle, we will design better experiments for E. coli and have these in mind for future attempts with cyanobacteria.

To check for correct expression and troubleshooting we need to conduct metabolite analysis. HPLC and LC-MS analysis would allow for effective detection of the bioluminescence pathway precursors, intermediates, and products. Our greatest concern regarding expression of bioluminescence is proper expression of HispS due to its size, complexity, and post-translational modification. By adding caffeic acid we would expect the presence of certain intermediates in hispidin synthesis as well as hispidin in the media. This test can be done with our level 1 strain for HispS as well as our level 2 strains. This method would allow us to see if our enzymes are functional and allow us to try switching promoters or adding more copies for higher efficiency in the system.

The modularity of golden gate cloning and the successful cloning of all parts allows for easy modifications to the system by altering promoters, coding sequences, or terminators for expression of different genes and alteration of gene expression.

Another aspect of the project that could be expanded upon is mutants of the bioluminescence genes. Site-directed mutagenesis could be utilized to produce different light wavelengths and intensities, or investigate the activity of enzymes in the bioluminescence pathway.