Overview

As seen in our extensive Design, our project is intricately woven with engineering principles in biology. From the use and characterisation of modified strains that can take up TPA (Terephthalic Acid) and EG (Ethylene Glycol) to the Assembly and Expression of Gene Constructs associated with the Sandalwood Oil Pathway, our experimental core is firmly embedded within the Design, Build, Test, Learn Framework. Using this framework, we navigate the complexities and challenges associated with our experimental goals.

Primarily, designing through a thorough literature review, we narrowed down the specificity of our P. putida KT2440 and its ability to sustain on PET monomers. Guided by expert insights and informed by our human practices research, we selected P. putida TA7-EG as our primary chassis. We then worked on characterising this strain's performance in different media formulations where TPA and EG served as the sole nutrient sources to understand its capabilities in these conditions fully.

To optimize sandalwood oil production, initially, we evaluated the pCDFDuet system, which was previously studied in E.coli and S. cerevisiae but encountered low santalene abundance in our P. putida chassis. This led us to redesign our approach, crafting custom constructs for SaSSy-FPPS and Cytochrome P450-CPR. We assembled our gene fragments using Gibson Assembly for both and integrated these designs into pSEVA631 and pSEVA241 plasmids through either conventional ligation or assembly. This completed plasmid was transferred into our chassis via electroporation. Recognising the unique challenges of expressing P450-CPR in bacteria, we developed multiple strategies, including truncated P450 variants and P450-CPR fusion proteins. Meticulously documented on our parts page, these constructs bring our molecular biology efforts to be open and shared with the global research community

We harnessed advanced analytical techniques and mathematical modelling to refine our bioengineering approach in our final engineering cycles. Using GC-MS analysis, we quantified the production of santalol and santalene, our key sandalwood oil components, validating our genetic constructs and design strategies. Simultaneously, through our Drylab work, we leveraged the mathematical modelling technique FBA (Flux Balance Analysis) to simulate the biochemical pathways of our modified P. putida TA7-EG in silico and provided valuable insights into optimal flux distributions, guiding targeted genetic modifications and culture condition adjustments. The FBA results inform our current wet-lab experiment and offer a roadmap for future teams, providing a clear next step for maximising the efficiency of the engineered system.

Iteration/ Engineering Cycle 1

Strain Making vs Strain Obtaining

Introduction

This initial phase of the project set the stage for our first engineering cycle, where we grappled with the critical decision of strain development versus obtaining existing one, a process deeply informed by our Integrated Human Practices (iHP) to ensure alignment with real-world applications. We carefully weighed the pros and cons of creating custom strains versus obtaining existing ones. During the ideation, we found a paper in which the Pseudomonas putida KT2440 could grow on ethylene glycol, one of the PET monomers, through laboratory evolution. Through the literature review, we learned that Pseudomonas putida KT2440 has a β-ketoadipate pathway, which has an aromatic acid catabolite of protocatechuate (PCA) ^[1], but the P. putida can’t convert the TPA into PCA naturally. To convert the TPA into PCA, an enzyme cascade must transform the tphII operon from the Comamonas sp., the multi-subunit enzyme TphAII, which converts TA to 1,2-dihydroxy-3,5-cyclohexadiene-1,4-dicarboxylate (DCD), and the enzyme TphBII, which converts DCD to PCA, are encoded in the Pseudomonas putida KT2440^[2].

We obtained the wild-type strain from Dr Phale from IIT Bombay to start with our approach. To obtain more insight into this idea, we discussed engineering the strain with Dr Oliver Brandenburg to enable terephthalic acid and ethylene glycol uptake. He reviewed our proposed timeline and concluded that engineering a new organism from scratch alongside developing the santalol pathway was too ambitious within the given timeframe. Instead, he recommended leveraging his existing engineered strains of Pseudomonas putida KT2440, which could take up both TPA and EG monomers, TA7-EG strain.

The DBTL CYCLE

Incorporating this adapted design into our project, we sought to characterize the growth of TA7-EG strain on terephthalic acid (TPA), ethylene glycol (EG), and a combination of both substrates in minimal M9 media. However, during the course of our experiments, we encountered several challenges related to the composition of the M9 media and the extended growth time of the strain.

One of the primary difficulties we faced was growing the strain on the M9 media. A major obstacle was the precipitation of calcium salts in the media, caused by both CaCl₂ and MgSO₄. This issue was further exacerbated by the pH changes resulting from adding TPA and EG to the media. Another challenge was the low solubility of TPA at neutral pH, which required careful optimization to ensure that sufficient amounts of the substrate were available for bacterial growth without significantly altering the media pH and causing calcium salt precipitation.

To address this issue, we fine-tuned the concentrations of the components of the media and monitored the pH at all times to balance TPA dissolution and ensure no precipitate formation. We also made sure to add the components just before further experiments to ensure the whole media was prepared each time a new and to avoid pH and precipitate issues. With all these challenges faced, we were able to successfully grow the P. putida TA7-EG strain on the different substrates, one such is shown below.

Fig.1.: The successful culturing of P. putida TA7-EG in M9 media with TPA (shown in image).

The second challenge we encountered during the characterization of the P. putida TA7-EG strain was its slow growth rate on PET monomers. Upon receiving the strain from Dr Oliver, we observed that the bacteria required 24-36 hours to achieve substantial growth on terephthalic acid (TPA) and ethylene glycol (EG), which was not optimal for real-world applications. We consulted with our primary PI, Dr Vijay and Dr Yashraj, to address this issue and designed an experimental plan to improve the strain's growth rate.

The approach we built involved creating a series of dilutions of the bacteria in fresh media, supplemented with varying glucose concentrations alongside the PET monomers. Based on our PI's suggestion, we started with a medium containing 5% glucose in addition to TPA and EG and gradually reduced the glucose concentration in subsequent dilutions until the media contained only TPA, EG, or a combination of both. This adaptive evolution strategy aimed to progressively acclimatise the strain to grow solely on the PET monomers.

After following the above plan, we tested the growth curves of the adapted strain in different media compositions (TPA, EG, and TPA+EG) by measuring the optical density (OD) using a microplate reader. The results showed that the adapted strain reached the logarithmic growth phase faster than the original TA7-EG strain, which took over 2 days. Notably, the highest growth rate was comparable in the medium containing TPA and EG, except EG, which was slower.

Fig.2.: Growth Curve of TA7-EG in M9 minimal media, with 20mM of each TPA and EG.

This result enabled us to learn that the adaptive evolution approach successfully improved the growth characteristics of the P.putida TA7-EG strain on PET monomers, bringing us one step closer to developing an efficient bacterial platform for PET bioconversion.

Iteration/ Engineering Cycle 2

SaSSy-FPPS Expression Testing

Introduction

Our second engineering cycle focused on testing the expression of SaSSy-FPPS enzymes, which catalyse the conversion of IPP/DMAPP into santalene, the first component of our sandalwood oil (see our Design Page for more details). Dr Bohlmann previously studied the expression of these enzymes in E. coli using the pCDFDuet expression vector. In our discussions with him, he mentioned that while his exact construct might not be optimally expressed in our P.putida TA7-EG strain and may require further modifications, he kindly agreed to provide us with the pCDFDuet containing SaSSy-FPPS for testing in our chassis.

The DBTL CYCLE

To test the initial pCDFDuet construct’s expression in our chassis. We transformed the pCDFDuet with SaSSy-FPPS into our TA7-EG strain via electroporation. After successful transformation, we induced the expression of SaSSy and FPPS with 1 mM IPTG and assessed the expression through Coomassie staining of purified protein and chemical analysis using GC-MS. The SaSSy enzyme, which already had a His-tag upstream from the original construct in the Bohlmann paper^[3], allowed us to purify the protein using Ni-NTA columns with His-trap resin. Despite some purification issues, the Coomassie-stained gel revealed a thick band at 65 kDa, corresponding to the expected size of the SaSSy protein.

Fig.3.: Electroporation of TA7-EG.

Fig.4.: Coomassie gel for pCDFDuet showing the band at 65kDa for SaSSy, Lane 1 - Ladder, Lane 2 - Whole cell lysate, Lane 3 - Purified protein.

GC-MS analysis was performed using our institute's Agilent 8890gc system 597 7 GC/MSD with a suitable column. We observed a peak at an m/z ratio of 204, which corresponds to the known ratio of santalene as reported in previous literature. However, the abundance of this compound within the sample was relatively low, confirming Dr Bohlmann's suggestion that the construct might not work optimally in our chassis.

Fig.5.: GCMS of pCDFDuet expressed in TA7-EG strain, showing spectrum for 204 m/z corresponding to santalene in the left one; at the right side there is a control with untransformed sample.

To address this issue, we devised several construct designs that could potentially improve the expression of SaSSy-FPPS in our chassis. These designs were based on principles such as bicistronic expression and dual expression systems, aiming to optimise the expression of SaSSy-FPPS in different settings. For more details, go to our Parts page.

Fig.6.: The araBAD_SaSSy_FPPS insert.

Due to time constraints and the availability of parts from IDT, we initially worked with our bicistronic SaSSy-FPPS construct under the control of an arabinose promoter, BBa_K5181014. The complete part was received as three separate fragments built using the NEBuilder HiFi DNA Assembly kit. Following assembly, we PCR-amplified the entire construct to obtain a high yield and to verify successful assembly and PCR.

Fig.7.: PCR of SaSSy_FPPS confirming the assembly.

After digestion, the gene insert and vector were ligated and transformed into our TA7-EG strain via electroporation. The transformed cells were plated on gentamicin-containing plates. To test the presence of the plasmid following successful transformation, we isolated the plasmid and performed a confirmatory digest using EcoRI and HindIII, expecting bands of 3 and 4 kb. As shown in the image below, both single (EcoRI) and double digests confirmed the presence of our plasmids in the TA7-EG strain.

Fig.8.: Single digested SaSSy_FPPS.

Fig.9.: Double Digested SaSSy_FPPS.

We also tested the expression of this construct by inducing various concentrations of arabinose (0.2%, 0.5%, 0.7%, 1%, 1.5% and 2%). We found the best results from the 0.7% and 1.5% inductions. The proteins were purified using the His Trap NI NTA column, and the expression of these proteins was confirmed by the bands observed in the following Coomassie-stained gel. To analyse their functionality, we tested the production of santalene using our designed construct through GCMS.

Fig.10.: Coomassie gel for the SaSSy_FPPS.

Fig.11.: The Left shows the presence of the santalene with a sharp peak at 204 m/z resembling the molecular weight; on the right side of the image, there is an untransformed sample showing no peak at the same retention time for 0.7% induction.

Fig.12.: A peak of 204 m/z presence strongly suggests the santalene expression in 1.5% arabinose induction in the left one; on the right side, the control shows no peak at the same retention time.

Through these iterative processes, we learned that our designed SaSSy-FPPS construct (BBa_K5181014) is functionally expressed in our TA7-EG strain when TPA and EG are the sole carbon sources. The functionality of the enzymes was confirmed by GC-MS analysis. To our knowledge, this is the first demonstration of santalene production in P.putida using PET monomers as the carbon source. While this is a promising result for us undergrad students, it presents new directions for further testing.

We still need to investigate the modularity of our parts by changing promoters and RBS sites to find optimal expression conditions. Building upon our current method, we can compare it with our other designed part, BBa_K5181018, which expresses SaSSy and FPPS under different promoters as part of the pRGPDuo vector and is suitable for separately tuned expression. This can be particularly useful for the independent expression of FPPS, as it is an important step in synthesising other terpenoids.

Iteration/ Engineering Cycle 3

Assembly and Expression of P450-CPR

Introduction

For our third and most relevant engineering cycle, our goal was to design the most suitable constructs for the expression of our P450-CPR enzymes. The P450-CPR enzymes catalyze the conversion of santalene to santalol, the final major component of sandalwood oil. The expression of these enzymes in prokaryotic systems remains a profound challenge that is still being tackled by many scientists worldwide. Several challenges are faced in expressing the eukaryotic (plant) membrane-bound cytochrome P450 enzyme in prokaryotic systems like Pseudomonas putida, such as differences in membrane composition, incorrect anchoring of the enzymes, and lack of optimal expression conditions in bacterial hosts.

However, our answer to this question involved combining experimental and computational efforts, extensively described on our Design and Modelling page. To overcome these challenges, we adopted transmembrane domain truncation and fusion with membrane-anchoring proteins to enhance the expression and solubility of membrane-bound P450s in our bacterial system.

The DBTL CYCLE

To address the design challenge of expressing Santalum album CYP736A167 (P450) in bacteria, we performed homology studies between this enzyme and P450 CAM, a naturally occurring P450 enzyme in P. putida and saw their core domains match. Following the truncation strategy, we removed the 27 amino acid hydrophobic extension in the N-terminal end of the Santalum album P450, considered to be the transmembrane domain, and similarly truncated the transmembrane domain of the native CPR1, the redox partner of P450. To ensure the proximity of the truncated P450 and CPR1 for FADH shuttling, we sought inspiration from a self-sufficient bacterial p450 from Bacillus cereus. Structural alignment revealed that these P450’s had significant overlaps, and we also identified a 36 amino acid long loop (BC_linker) connecting the C-terminal of the oxidase domain to the N-terminal of the reductase domain in pP450 from Bacillus. This corresponds to the same area where the transmembrane region of our CPR1 connects to the core enzyme body. We used this BC_linker to join the truncated saCYP736A167 and saCPR1, forming our fusion protein.

Fig.13.: The alphafold structure of our P450-CPR part with the linker after design.

To test our design, we built a composite part BBa_K5181015 with tet promoter and truncated fusion P450-CPR, an optimized fusion protein sequence for expression in P. putida. This part is also modular with respect to the promoters and RBS sites.

Fig.14.: The P450-CPR part sequence.

This part arrived as three separate fragments, which were Gibson assembled using the NEBuilder DNA HiFi Assembly Kit. After assembly, the whole gene insert was PCR amplified for confirmation and further downstream ligation into a suitable vector.

Fig.15.: 1% agarose gel, showing bands for 4151bp, the expected size of our designed P450-CPR.

Despite multiple attempts, we encountered difficulties in successfully ligating the vector and insert using both conventional ligation and Gibson assembly methods. After digesting the vectors and inserts with EcoRI and HindIII, convetional ligation was carried out. At the same time, Gibson assembly was performed using the same kit but following the 2-fragment protocol for the vector backbone and gene. These ligation trials were conducted with various vector backbones, including pSEVA 241, 631, and 424. However, the transformations either failed to yield colonies after electroporation or resulted in transformed colonies that, when picked and isolated for plasmids, showed only smears when run on a gel, indicating unsuccessful ligations. More can be found on our results page.

In conclusion, our attempts to assemble the vector backbone with the gene insert through ligation proved challenging. To identify the underlying cause, we sequenced the first and third fragments involved in binding with the vector, focusing on the regions containing restriction sites and overlaps. The sequencing results revealed a perfect alignment of the first fragment sequence with the expected sequence at 130 base pairs, indicating the correctness of the fragment itself. However, we could not confirm the primer binding site due to its involvement in sequencing.

Fig.16.: Sequencing chromatogram of the first fragment of BBa_K5181015.

Fig.17.: Global alignment (Needleman-Wuntsch) of our sequenced and actual ordered fragments.

The sequencing of the third fragment unveiled numerous overlapping peaks throughout the sequenced region, suggesting potential issues related to the sequence, primer binding, or the presence of a homopolymer region. These issues can hinder successful ligation, even if the fragment is amplifying during PCR.

Fig.18.: Sequencing chromatogram of the third fragment BBa_K5181015.

Based on these observations, we hypothesise that the fragment may amplify correctly during PCR but fail to ligate properly due to sequence or structural problems. To address these challenges and continue our research, we propose a two-pronged approach of doing further ligation tests to investigate the ligation efficiency and expressing our designed part BBa_K5181017, which incorporates the truncated P450 and CPR genes under the control of tac promoters.

By acknowledging limitations, proposing testable hypotheses, and outlining a plan for future experiments, we aim to continue our research efforts and make progress towards the successful expression of the P450-CPR in our chassis.

This, however, did not hinder our efforts in exploring the functions of the designed P450-CPR genes. Our computational team sought to understand how these genes would function if expressed. Through our modeling work, we explored the docking of santalene and its derivatives with our designed P450-CPR genes and investigated their binding effectiveness by analyzing the binding energies.

As shown in the table below, the results from our docking simulations provide valuable insights into the binding energies of various substrates with the designed P450-CPR constructs. The mean binding energy of rank 1 cluster for α-santalene, a key intermediate in the santalene biosynthesis pathway, was found to be -7.37 kcal/mol. This binding energy suggests a favourable interaction between α-santalene and the P450-CPR enzyme, indicating the potential for effective conversion to the desired product, α-santalol. Similarly, the binding energies for other santalene derivatives, such as β-santalene and epi-β-santalene, were also promising, ranging from -7.64 to -7.41 kcal/mol.

Sr. No.	Substrate molecule	Mean binding energy of rank 1 cluster in Kcal/mol	Product molecule	Mean binding energy of rank 1 cluster in Kcal/mol	Binding enegy difference in Kcal/mol
1	α-santalene	-7.37	α-santalol	-7.26	0.11
2	β-santalene	-7.64	β-santalol	-7.4	0.24
3	epi-β-santalene	-7.41	epi-β-santalol	-7.24	0.17
4	α-exo-bergamotene	-7.68	α-exo-bergamotol	-7.24	0.44

The image below shows one docking pose illustrating the interaction between α-santalene and the P450-CPR enzyme.

Fig.19.: Interacting mesh of P450-CPR with alpha santalene.

These computational insights provide valuable support for the potential functionality of our designed P450-CPR genes, even as we continue our efforts to overcome the experimental challenges encountered in their expression. By combining experimental and computational approaches, we aim to gain a comprehensive understanding of the P450-CPR system and its role in the conversion of santalene and its derivatives to the desired products. For more detailed information on our modeling approaches and additional results, please refer to our Modelling page.

Iteration/ Engineering Cycle 4

FBA and DXS-DXR

Our final engineering cycle focused on optimising the santalene and santalol production pathway by investigating the role of DXS (1-deoxy-D-xylulose-5-phosphate synthase) and DXR (1-deoxy-D-xylulose 5-phosphate reductoisomerase), the first two enzymes in the MEP (methylerythritol phosphate) pathway. The MEP pathway is crucial for synthesising IPP (isopentenyl pyrophosphate) and DMAPP (dimethylallyl pyrophosphate), the precursors for santalene production. Our modelling team analysed the active pathways under our experimental conditions and discovered that DXS and DXR are critical enzymes that must be expressed to increase IPP/DMAPP production. These findings align with the results previously reported by Dr Jordi^[4].

To better understand the metabolic network, we designed and compared FBA (Flux Balance Analysis) models based on previously published models, namely iJP815, PpuQY1140, and iJN1463, as listed in the table below. We modified each model by incorporating reactions relevant to our project, such as TPA (terephthalic acid), EG (ethylene glycol) uptake, and the santalol synthesis pathway. The FBA approach also allowed us to perform in silico tests to determine the optimal TPA and EG uptake rates for maximizing santalol production using Pareto-optimization. Refer to our modelling page for more details.

Model name	No. of Metabolites	No. of reactions	No.of genes	Year of publication
iJP815	892	948	818	2008
PpuQY1140	1104	1171	1140	2017
iJN1463	2160	2936	1462	2019

Based on our modelling results and the advice from Dr Jordi, we obtained plasmids containing DXS and DXS-DXR genes. Dr Jordi suggested that co-expressing DXS-DXR might be the most effective strategy to overexpress the MEP pathway and increase our santalene and santalol yield, which aligns with our modelling predictions. We successfully revived the bacterial stocks received from Dr Jordi, as shown in the image below.

Fig.20.: Revived DXS, DXS-DXR plasmid bacterial stocks.

However, due to time constraints and the late arrival of the plasmids, we were unable to co-express the DXS-DXR genes along with our SaSSy-FPPS and P450-CPR genes within the given timeframe. Despite this setback, our modelling efforts provided valuable insights into the potential optimization of the santalene and santalol production pathway.

This engineering cycle highlights the importance of integrating computational modelling with experimental design to guide metabolic pathway optimization. By identifying critical enzymes and predicting optimal substrate uptake rates, we can make informed decisions to enhance the production of our target compounds. Although we faced challenges in the experimental implementation, our modelling results lay the foundation for future work on optimizing the MEP pathway and ultimately increasing santalene and santalol yields.

References

Harwood, C. S., & Parales, R. E. (1996). The β-Ketoadipate Pathway and the Biology of Self-Identity. Annual Review of Microbiology, 50(1), 553-590. https://doi.org/10.1146/annurev.micro.50.1.553
Brandenberg, O. F., Schubert, O. T., & Kruglyak, L. (2022). Towards synthetic PETtrophy: Engineering Pseudomonas putida for concurrent polyethylene terephthalate (PET) monomer metabolism and PET hydrolase expression. Microbial Cell Factories, 21, 119. https://doi.org/10.1186/s12934-022-01846-w
Jones, C. G., Moniodis, J., Zulak, K. G., Scaffidi, A., Plummer, J. A., Ghisalberti, E. L., Barbour, E. L., & Bohlmann, J. (2011). Sandalwood Fragrance Biosynthesis Involves Sesquiterpene Synthases of Both the Terpene Synthase (TPS)-a and TPS-b Subfamilies, including Santalene Synthases. The Journal of Biological Chemistry, 286(20), 17445–17454. https://doi.org/10.1074/jbc.M111.231787
Hernandez-Arranz, S., Perez-Gil, J., Marshall-Sabey, D., & Rodriguez-Concepcion, M. (2019). Engineering Pseudomonas putida for isoprenoid production by manipulating endogenous and shunt pathways supplying precursors. Microbial Cell Factories, 18(1), 152. https://doi.org/10.1186/s12934-019-1204-z

Engineering Success

Contents

Overview

Strain Making vs Strain Obtaining

Introduction

The DBTL CYCLE

SaSSy-FPPS Expression Testing

Introduction

The DBTL CYCLE

Assembly and Expression of P450-CPR

Introduction

The DBTL CYCLE

FBA and DXS-DXR