Re-purposing Chlorophyll into a product of value.
Selectively binding and removing chlorophyll from canola oil gives our system a huge advantage over the current industry standard, acid-activated clays. With acid-activated clays, chlorophyll cannot be recovered and is discarded as a waste product. The yOIL system allows chlorophyll to be recovered and repurposed into a valuable product. By enzymatically degrading chlorophyll into pheophorbide, we can create an end product valuable not only in photodynamic research, but also to prevent crop losses from fungal infestations. Ultimately, we sought to create new value from waste produced at the oil processing level that feeds back into the canola industry at the farmer level, thus decreasing losses and increasing profit.
Converting chlorophyll a and b into pheophorbide a
In the canola oil processing industry, acid-activated clays are used to extract chlorophyll from canola oil to remove its bitter taste and prevent premature spoilage. However, the chlorophyll is discarded by processors because it is tightly bound to the clay particles. By using our chlorophyll extraction emulsion system we can capture and isolate the chlorophyll found in canola oil.
With the tools to capture chlorophyll, there was a new opportunity to turn this former waste product into a commercially valuable product. Our review of the current commercial uses of chlorophyll revealed that it is mainly used as a nutraceutical. With this knowledge, we wanted to dig deeper. To find an application for our system that goes beyond nutraceuticals and addresses a current issue. The conversion of recovered chlorophyll to a marketable product was imperative to the economic viability of our project as a whole and has the potential to create a market for what is now considered a waste product.
Figure 1. Chlorophyll Degradation Pathway. The natural chlorophyll degradation pathway found within plant species. The above shows enzyme catalyzed reactions that in sequence convert chlorophyll b into 7-hydroxymethyl chlorophyll a into chlorophyll a into pheophytin a into pheophorbide a. Colored boxes surround enzyme names to indicate their pairing in a theoretical "two step in-vitro degradation" process.
In examining the natural degradation pathway of chlorophyll we identified a catabolite of interest: Pheophorbide. Pheophorbide has been shown in recent research to be a strong photosensitizer producing cytotoxic effects on cancer cells and fungi when light activated. Particularly, we are interested in its potential as a new anti-fungal treatment that may be used to circumvent current antibiotic resistance in fungal populations.
In order to produce pheophorbide a we required the use of four enzymes: chlorophyll b reductase (CBR), 7-hydroxymethyl chlorophyll a reductase (HCAR), Mg2+ dechelatase (SGR), and pheophytinase (PPH). In delving into the various catalysis effects of these proteins, it was found that SGR not only removes Mg2+ from chlorophyll a, but also 7-hydroxymethyl chlorophyll a to a lesser degree, which produces 7-hydroxymethyl pheophytin a. The production of 7-hydroxymethyl pheophytin a not only is the result of competitive inhibition (reducing our desired reaction's efficiency), but also creates a sink of compound that our enzymes will be unable to use to make our desired product. In order to circumvent these losses, we have developed a two step in-vitro degradation process that separates the SGR protein from 7-hydroxymethyl chlorophyll a. This comes in the form of feeding CBR and HCAR chlorophyll a and b captured by our emulsion system and allowing the reaction to reach relative "completion" in the first step. The second step uses the predicted "chlorophyll a" dominated solution as an input for the SGR and PPH enzymes to produce pheophorbide a, which can then be applied as a marketable product.
Universal Design of Pathway Products
While designing our parts (Bba_K3114024, Bba_K3114025, Bba_K3114026, Bba_K3114027) to recombinantly express native plant and algae genes in Escherichia coli, we decided to stick to a single expression system. We used BL21(DE3) E. coli cells as a chassis for our protein production, allowing for our parts to be controlled by IPTG induction, therefore we used a T7 promoter (BBa_I719005) in our constructs. To reduce the effect of leaky expression in our system we used a double terminator (BBa_B0014). To increase the efficiency of translation based on ribosome sequence recognition in E. coli we used a strong ribosome binding site (BBa_B0030). In our background research, we found little work had been done on expressing our proteins of interest (CBR, HCAR, SGR, PPH) in E. coli. We were able, however, to find sequences which have been proven to have successful recombinant expression of CBR, HCAR, SGR, and PPH (Horie, Ito, Kusaba, Tanaka, & Tanaka, 2009; Meguro, Ito, Takabayashi, Tanaka, & Tanaka, 2011; Matsuda, Shimoda, Tanaka, & Ito, 2016; Guyer, Salinger, Krügel, & Hörtensteiner, 2017). Three of our four genes came from Arabidopsis thaliana (CBR, HCAR, PPH) and the fourth came from Chlamydomonas reinhardtii (SGR).
Figure 2. Construct design for recombinant expression of CBR, HCAR, SGR, and PPH in E. coli. T7 Promoter sequence: (BBa_I719005). RBS (ribosome binding site) sequence: (BBa_B0030). GOI (gene of interest): location of CBR, HCAR, SGR, PPH sequences. ICARUS (universal spacer for 6xHis-tag purification of large proteins with strong electrostatic interactions): (BBa_K3114014). Bidirectional terminator sequence: (BBa_B0014). The ICARUS part sequence contains the 6xHis-tag and the double terminator. This construct design is representative of the following parts: Bba_K3114024, Bba_K3114025, Bba_K3114026, Bba_K3114027. These parts were designed to recombinantly express CBR, HCAR, SGR, PPH, respectively; in E. coli.
Of particular interest, in our preliminary literature review, we found that the 7-hydroxymethyl chlorophyll a reductase (HCAR) enzyme recombinantly expressed in E. coli was catalytically inactive after being purified using a 6x His-tag, even though the protein was produced and functional from the whole cell lysate (Meguro et al., 2011). Further research was done to analyze the structure of our proteins and only HCAR had an available protein data-bank (.pdb) file. It was found that the resulting homo-hexamer of the folded HCAR protein has a highly electronegative binding pocket; as per our electrostatic interaction modelling. It was our assumption that due to this high electronegativity the 6xHis-tag that was added to the protein from the aforementioned paper, likely folded into the binding pocket, resulting in enzyme function inhibition. With this in mind we decided to try to design a spacer that would allow for non-disrupted functionality of both the protein and the 6x His-tag, allowing for the protein to be both purified and functional.
ICARUS Spacer Solution to Purification Problem
Initially, in attempting to create a suitable spacer, we attempted using standard methods of alternating alanine and glycine residues. However, the high GC content of these amino acids prevented the possibility of gene synthesis. To avoid this problem, we tested several combinations of alternating alanine residues with other amino acids that have "simple" side chains, including threonine, serine, leucine, isoleucine, valine. By using homology modelling we thought the alternation of alanine and serine amino acids would be safest for our purposes. Further thinking about the ways in which proteins fold, we determined that it would be best to stabilize the spacer linkage between the large protein and the 6xHis-tag by attempting to create a "helix-turn-helix-turn-helix" structure. We hypothesized that this motif would allow for increased stability and space between the two moieties, while also enabling us to add extra features within the turns. We implemented a thrombin proteolytic site in turn #1 and a string of negatively charged aspartic acids in turn #2. The proteolytic site allows for removal of the synthetic sequence from the protein using thrombin digestion post Nickel-NTA column purification. The negatively charged region we hypothesized would cause repulsion between the problematic binding pocket and the spacer itself, making potential folding of the positively charged 6xHis-tag into the negatively charged pocket less attractive. By using Rosetta comparative and ab initio structural modelling, we determined that our designed spacer would likely form the desired helix-turn-helix-turn-helix structure.
Figure 3A and 3B. Hypothesized (3A-left) and Modelled (3B-right) ICARUS Structure The left image depicts our hypothesized structure of ICARUS containing a "helix-turn-helix-turn-helix" structure. The right image is the result of ab initio structural prediction modelling.
Applying ICARUS Beyond HCAR
Inspired by the potential for ICARUS to be applied beyond the HCAR case study, we sought to create in-silico crystal structures of our other three proteins we intended on purifying with a 6xHis-tag (CBR, SGR, PPH), as there are no available pdb files for these proteins. To do this our modelling team returned to Rosetta comparative and ab initio structural modelling to create predicted molecular structures of the proteins with ICARUS attached.
Figure 4A, 4B, 4C, 4D. Predicted Structural Modelling for CBR (4A-top left), HCAR (4B-top right), SGR (4C-bottom left), and PPH (4D-bottom right). CBR, HCAR, and PPH models were created via comparative modelling done through Rosetta. The SGR model was produced by ab initio predictive structural modelling. HCAR is known to form a hexameric structure, above portrays a monomer of this complex. CBR was predicted to form a tetrameric structure. SGR and PPH were predicted to have monomeric structures.
These models further solidified the hypothesized utility and structure of the ICARUS spacer with our chlorophyll catabolite enzymes, giving us confidence to use it as a universal spacer to increase the potential efficacy of our purification methods in the lab. Therefore the spacer was implemented into all four of the genetic constructs, each containing an enzyme involved in the chlorophyll degradation pathway. Learn more about how Rosetta comparative and ab initio structural modelling was employed.
Pheophytinase (PPH) was of particular interest, as we found in literature that while the purification of this protein was attempted from recombinant expression in E. coli it was unable to be purified using a maltose binding protein (MBP) fusion or with a 6xHis-tag (Guyer, Salinger, Krügel, & Hörtensteiner, 2017). The design of our parts and creation of ICARUS made us excited to get into the lab and confirm our system's functionality.
Successful 6xHis-tag driven purification of HCAR and PPH
To confirm purification of our successfully cloned and recombinantly expressed proteins (HCAR and PPH) we visualised proteins via SDS-PAGE gels stained with Coomassie blue and performed ECL Western Blot analysis using an anti-his-tag MAb (primary antibody from mouse) and an anti-mouse IgG conjugated with HRP (secondary antibody).
Figure 5. 10% SDS-PAGE HCAR and PPH Purification Confirmation. 10% SDS-PAGE was run at 100V for 15 minutes and then 180V for 35 minutes, then stained using Coomassie Blue. Lanes read left to right contain Color Prestained Protein Standard, Broad Range (11–245 kDa) as a ladder (NEB), PPH - post loading fraction, PPH - elution fraction 1, PPH elution fraction 2, HCAR - post loading fraction, HCAR - elution fraction 1, HCAR elution fraction 2, pSB1A3 (plasmid control in BL21) - elution fraction 1, and pSB1A3 (plasmid control in BL21) - elution fraction 2. NEB Ladder is shown on the left. HCAR is 58 kDa and PPH is 55.8kDa. This SDS-PAGE Confirmation is replicate 4/5, other replicate images are available here.
Figure 6. Western Blot HCAR Purification Confirmation. 10% SDS-PAGE was run and transferred to a PVDF membrane. An "anti-his-tag MAb" primary antibody was used with an "anti-mouse IgG conjugated with HRP" secondary antibody. ECL was used to visualise. Lanes from left to right on the SDS-PAGE were ladder (Color Prestained Protein Standard, Broad Range (11–245 kDa) (NEB)), ~60 kDa protein (positive control) in two lanes, pSB1A3 (plasmid control in BL21) - elution fraction 3, HCAR - elution fraction 1, HCAR - elution fraction 2, HCAR - elution fraction 3. HCAR is 58 kDa.
As seen above, it is clear that with our ICARUS spacer we were able to purify HCAR and PPH proteins using Nickel-NTA column chromatography. With specific reference to the PPH protein this purification is an accomplishment we had not seen previously done in any published literature. We reproduced five separate 10% SDS-PAGE gels showing that the HCAR and PPH proteins were successfully isolated and purified. Presence of HCAR and PPH proteins was demonstrated by the 58 kDa and 55.8 kDa bands, respectively, when compared to the control containing an empty vector (pSB1A3-BL21).
To gain further confidence in having purified our desired proteins with the 6xHis-tag, we performed a Western blot using a 6xHis-tag mouse antibody and a ~60 kDa positive control, as depicted above showing the presence of our protein HCAR at 58 kDa. This observation confirms it is our recombinant protein with the fused 6xHis-tag that is visualised in our SDS-PAGE confirmation gels.
The isolated proteins underwent ammonium sulfate precipitation and were re-solubilized in our reaction buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% Triton X-100) for storage. This reaction buffer is based off of enzyme assay buffers used to characterize HCAR and PPH, found in (Meguro et al., 2011) and (Guyer et al., 2017), respectively. As our substrates were unable to be ordered from a chemical distributor we had to chemically produce them in order to characterize our recombinant proteins. Chlorophyll extraction from spinach and subsequent pheophytin production is described in the protocols section here. HCAR's substrate, 7-hydroxymethyl chlorophyll a, was unable to be produced as we required NaBH4, a highly reactive and explosive compound (Ito, Ohtsuka, & Tanaka, 1996). Due to these circumstances, we were unfortunately unable to further characterize HCAR.
Figure 7. Predicted Structural Modelling of Pheophytinase. Comparative modelling done via Rosetta was used to generate a predicted structural model for PPH.
Pheophytinase (PPH) is arguably the most important enzyme in our defined pathway as it directly produces our desired compound: pheophorbide a. As a result, the characterization of this protein was the most important to us. In order to do this, we needed a reliable method to confirm the enzymatic conversion of pheophytin into pheophorbide.
Visualisation of pheophytin and pheophorbide via Thin-Layer Chromatography
Our aim was to differentially separate pheophytin and pheophorbide to qualitatively characterize pheophytinase activity. We chose to use Thin Layer Chromatography (TLC), as pheophytin and pheophorbide have very similar absorbance spectras at 665 nm (Guyer et al., 2017). As a result of this, spectrophotometry would not be useful for observing relative decreases in pheophytin and increases in pheophorbide. Chlorophyll derivatives have previously been differentially separated based on polarity using thin layer chromatography, therefore we had a basis for pursuing a TLC approach to characterization (Jeffrey, 1981).
In our initial trials, we used flexible cellulose sheets as our TLC plate and hexane as our solvent since it has weak elution strength and is nonpolar. However, we faced challenges as the pigments chlorophyll, pheophytin, and pheophorbide a did not elute fully and were not distinctly separated from each other (Figure A1). To better distinguish these derivatives, we changed our solvent to 100% acetone (Figure A2). This resulted in better elution of the compounds from the spotting line.
We then ran our compounds on silica plates to eliminate the possibility of high affinity and adsorption to the cellulose plates. Using the silica plates we observed the fluorescence of our compounds for the first time via UV visualisation (Figure A3). However, while we could visualise the fluorescence of our compounds they were not differentially separated, so we decided to again change our solvent system. We used a 70:30 methanol:hexane solvent mixture, which showed better differential separation of pheophorbide and pheophytin. This solvent combination eluted pheophytin higher than pheophorbide, and allowed us to use the brighter, neon-like pink fluorescence of pheophorbide as a marker (Figure A4). We have determined after several TLC trials that pheophytin fluorescence consistently appears as a red tinge, whereas pheophorbide a fluorescence appears as a distinctive neon-like pink. Additionally, when running pheophorbide with our reaction buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% Triton X-100), it shows a characteristic pink curve that is not present in pheophytin with the same buffer.
Having developed a reliable way to differentially visualise pheophytin and pheophorbide we began to carry out our pheophytinase (PPH) reactions and ran them on our TLC plates using a 70:30 (methanol:hexane) solvent system. Reactions were run using pheophytin produced via acidification of chlorophyll extraction solution as detailed here. Figure A3 showed that the protein was inactive, as the characteristic pink curve was not present, however we suspected that the ammonium sulfate precipitation may have denatured our proteins. After this result we decided to produce new PPH stocks, with which we ran new reactions using cell lysate and Nickel-NTA column purified protein elution fractions.
After using our new cell lysates and protein stocks for our reactions and performing TLC analysis, the characteristic pheophorbide a curve and fluorescence was observed for the first time from the pheophytinase reactions. However, the pheophorbide a + pheophytin + reaction buffer control did not show the same curve which suggested the absence of pheophorbide. To verify this, more pheophorbide a was added to the pheophorbide a + pheophytin + reaction buffer mixture and was run with another set of cell lysate and protein reactions (Figure 8). Ultimately, we were able to obtain qualitative results, in triplicate, characterizing functional pheophytinase activity. It was found that the cell lysate reaction created more pheophorbide a when compared to the Nickel-NTA column purified protein elutions. Replicate TLC plate pictures can be found in the Appendix as Figure A5 and A6.
Figure 8A and 8B. Thin Layer Chromatography Analysis of Pheophytinase Reactions. Samples were eluted using a methanol:hexane (70:30) solvent system on a silica plate. Lanes (read left to right) contain 24 hour reactions using Nickel-NTA purified PPH elution fraction 3, 2, 1, whole cell lysate (containing recombinant PPH) fraction 2, 1, pheophytin + reaction buffer solution, pheophytin + pheophorbide a + reaction buffer. Reaction buffer: 25 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% Triton X-100. This is one of three replicates. The left (8A) figure is the raw image taken using an LG G6 Camera whereas the right (8B) is the same image with modified brightness:contrast ratio using the imaging software FIJI.
Recombinant pheophytinase enzymatically converts pheophytin into pheophorbide
Through thin layer chromatography analysis of pheophytin, pheophorbide, and reactions involving our recombinant PPH protein, we have been able to qualitatively show that pheophytinase is able to enzymatically convert pheophytin into pheophorbide. Additionally, it was determined that pheophorbide has a characteristic neon-pink curve that is distinctive from pheophytin, a pale red, when using a methanol:hexane (70:30) solvent system on a silica plate. Overall, we were able to recombinantly express and purify PPH, while also demonstrating it had the ability to produce pheophorbide given pheophytin.
Purification of large plant proteins using ICARUS
ICARUS, our "universal" spacer, implemented into our four chlorophyll degradation enzyme constructs is thought to have successfully folded and performed as hypothesized. ICARUS was modeled in-silico via Rosetta comparative and ab initio folding by itself and attached to CBR, HCAR, SGR, and PPH. In all five of these models, ICARUS folded according to our predicted structure. In the lab, both HCAR (58 kDa) and PPH (55.8) were confirmed to have successfully been purified via 10% SDS-PAGE, stained with Coomassie blue (Figure 5) and Western Blot analysis, using a 6xHis-tag mouse antibody (Figure 6) confirmed purification of HCAR. Further, as described above, PPH was shown to be catalytically active. The use of ICARUS in our constructs lead to the purification of PPH, an accomplishment not seen in previously published literature.
Pheophorbide a Production Scale-Up
To address the challenge of scaling up pheophorbide production using enzymatic degradation, CRISPR could be used to integrate the chlorophyll degradation enzymes as an operon into yeast. This would allow us to feed in chlorophyll and produce pheophorbide a in large quantities through batch fermentation, which is an established method for industrial-scale biosynthetic production. To improve this further, a multi-enzyme complex scaffold could be made within the yeast cells. The multi-enzyme complex increases the concentrations of produced chlorophyll catabolites being fed to the enzymes while preventing their accumulation to toxic levels. They also allow compound formation to have lower rates of intermediate losses.
Additionally, pheophytin a and b, along with chlorophyll a and b, are the most prevalent pigments found within green oil. Therefore, PPH could be used in the future to capture and process pheophytin in an emulsion system to produce pheophorbide that can be extracted directly from the oil processing step of our overall system. This avoids the need for in vitro degradation of chlorophyll a and b, making it more economical. This can be done in tandem with our existing chlorophyll extraction emulsion system.
Making ICARUS More Universal
To more fully characterize ICARUS we would like to directly compare the purification efficiency between proteins with ICARUS, and without, to concretely validate its contribution. While we attached ICARUS to the C-terminal of our proteins, the sequence can also be flipped for functionality at the N-terminal for proteins with active sites closer to the C-terminal. ICARUS could also be tested with a broader range of proteins and be applied outside of 6xHis-tag purification. For example, ICARUS can be used to fuse a maltose binding protein to verify if it can be used with other purification methods. Our spacer is more than just a sequence, we hope that it can be built upon to improve its functionality and/or stability. ICARUS' potential varies depending on the application and we hope future teams can develop its characterization in this way.
CBR, HCAR, SGR Characterization
In order to realise our vision of a two-step in-vitro degradation pathway, we would need to characterize CBR, HCAR, and SGR recombinant proteins. CBR can be characterized by tracking the change in absorbance at 430 nm (chlorophyll b) and 663 nm (7-hydroxymethyl chlorophyll a) to quantify their relative decrease and increase, respectively (Meguro et al., 2011). Observing absorbances of the reaction at one minute intervals to a total of 15 minutes would be done to analyze the enzyme kinetics. High-Performance Liquid Chromatography (HPLC) analysis could also be done to quantify the amount of product made from the reaction. HCAR characterization can be conducted through HPLC analysis in order to track the small change in compound structure via spectral analysis (Meguro et al., 2011). SGR characterization can be done by using a trackable parallel reaction that uses up Mg2+ as a substrate. This would need to be done as chlorophyll a and pheophytin a share the same structure other than the presence or absence of the Mg2+. Additionally, HPLC analysis can be done for all four reactions.
With our team being the first to purify functional pheophytinase, our construct can be used by other researchers to purify the enzyme and determine its crystal structure. This will allow for more characterization of pheophytinase by providing insights on its substrate-binding sites and catalytic mechanism. In the future, this would enable for the construction of better recombinant pheophytinase to increase our product output.
Guyer, L., Salinger, K., Krügel, U., & Hörtensteiner, S. (2017). Catalytic and structural properties of pheophytinase, the phytol esterase involved in chlorophyll breakdown. Journal of Experimental Botany, 69(4), 879–889. doi: 10.1093/jxb/erx326
Horie, Y., Ito, H., Kusaba, M., Tanaka, R., and Tanaka, A. (2009). Participation of Chlorophyll b Reductase in the Initial Step of the Degradation of Light-harvesting Chlorophyll a/b-Protein Complexes in Arabidopsis. J Biol Chem, 284(26), 17449-17456. doi: 10.1074/jbc.M109.008912
Ito, H., Ohtsuka T., and Tanaka, A. (1996) Conversion of chlorophyll b to chlorophyll a via 7-hydroxymethyl chlorophyll. J Biol Chem, 271(3), 1475-1479. doi: 10.1074/jbc.271.3.1475
Jeffrey, S. W. (1981) An improved thin‐layer chromatographic technique for marine phytoplankton pigments. ASLO, 26(1), 191-197. doi: 10.4319/lo.1981.26.1.0191
Matsuda, K., Shimoda, Y., Tanaka, A., and Ito, H. (2016) Chlorophyll a is a favorable substrate for Chlamydomonas Mg-dechelatase encoded by STAY-GREEN. Plant Physiology and Biochemistry, 109(2016), 365-373. doi: 10.1016/j.plaphy.2016.10.020
Meguro, M., Ito, H., Takabayashi, A., Tanaka, R., & Tanaka, A. (2011). Identification of the 7-Hydroxymethyl Chlorophyll a Reductase of the Chlorophyll Cycle in Arabidopsis. The Plant Cell, 23(9), 3442–3453. doi: 10.1105/tpc.111.089714