ENGINEERING SUCCESS
𧬠Parts
Genetic Component Engineering
Our parts engineering focused on developing and optimizing genetic constructs for expression in Rhodococcus. Through multiple iterations, we addressed challenges in construct amplification, assembly methods, and promoter characterization.
Iteration 1: Q5 Amplification + Homology with Fusogenic Region
π DESIGN
The construct was designed for amplification using Q5 high-fidelity polymerase, with homology arms flanked by a region for fusion PCR. PCR primers for ROCI 1 homology region were designed for accuracy, minimal error rate, and compatibility with downstream cloning. ROCI 1 region was discovered in paper [1]. Primers for their amplification also added 20bp homology regions with the extremities of linear construct. A deeper insight on LoxPort design can be found on part BBa_2566C7GE documentation.
π¨ BUILD
PCR amplification was performed using Q5 polymerase, generating the desired homology regions. PCR was also performed to amplify the linear construct.
π§ͺ TEST
PCR confirmed successful amplification of the desired fragments only for homology regions, showing bands of the expected size. No PCR results were obtained for the linear construct, neither with high-GC enhancer buffer addition.
π LEARN
The observed inefficiency suggested that even with Q5βs fidelity, amplification failed, probably due to a mixture of construct length and high GC content. We therefore decided to amplify smaller fragments.
Iteration 2: Division into Multiple Parts + Fusogenic Region
π DESIGN
To manage construct size and complexity, we designed primers to divide the construct into multiple smaller fragments. The principle was that smaller fragments would be easier to amplify and assemble.
π¨ BUILD
Fragments were generated by PCR and prepared for fusion PCR.
π§ͺ TEST
Although individual fragments amplified correctly, assembling them into a single construct showed inefficient and incomplete results.
π LEARN
The results showed that dividing the construct introduced additional junctions that increased assembly difficulty. This highlighted the tradeoff between fragment manageability and fusion PCR efficiency. Q5 polymerase did not seem to be able to amplify the whole construct. We therefore decided to amplify the construct in a plasmid.
Iteration 3: Assembly in pGEM
π DESIGN
To amplify the native linear construct, we decided to clone it into pGEM vector. We needed to perform A-tailing and ligation to obtain a plasmid to transform into E.coli.
π¨ BUILD
The linear synthesis result was A-tailed following NEBβs klenow protocol. It was then ligated in pGEM vector, with T tailing. Then, transformation in Escherichia coli DH5-Ξ± was performed with ligation results.
π§ͺ TEST
DNA extraction from colonies yielded no plasmid DNA.
π LEARN
No successful transformation was obtained. Also, not much linear construct was left: a working amplification was needed to proceed with the experiments, so we tried PCR with a different polymerase.
Iteration 4: Extaq Amplification
π DESIGN
An alternative strategy used Extaq polymerase for amplification, anticipating robust amplification even for challenging sequences, followed by fusion PCR with previously obtained plasmids.
π¨ BUILD
PCR with Extaq successfully generated the desired fragments, on which fusion PCR with homology regions was performed.
π§ͺ TEST
While PCR products were clear and matched expected sizes, attempts to assemble construct and homology regions fragments consistently failed.
π LEARN
The results revealed that successful amplification did not guarantee successful fusion PCR. Sequence context, particularly high GC content and secondary structure, likely interfered with amplification, together with construct length. This suggested the need for sequence optimization and to have the construct directly inserted in a plasmid.
Iteration 5: Twist Bioscience Synthesis - Reduced GC + Medium-Copy Plasmid
π DESIGN
To overcome previous challenges, the construct was redesigned with reduced GC content, already fused homology regions and was synthesized directly in a plasmid by Twist Bioscience. LoxPort construct (BBa_2566C7GE) was the result of this optimization. It was placed in a medium-copy backbone to limit toxicity, while allowing sufficient yield.
π¨ BUILD
The optimized construct was synthesized and delivered in plasmid form, bypassing the challenges of PCR-based amplification and in vitro assembly.
π§ͺ TEST
The synthesized plasmid worked as expected, remaining stable in the chassis and yielding reliable DNA for downstream applications.
π LEARN
This success confirmed that high GC content and sequence complexity were key barriers in previous cycles. The iteration demonstrated that codon optimization and backbone choice are critical parameters, feeding directly into future design strategies for complex constructs.
Iteration 6: Construct Integration in pKSAC45
π DESIGN
LoxPort construct (BBa_2566C7GE) was designed to be integrated into the pKSAC45 backbone, with the goal of performing double recombination inside Rhodococcus. The choice of pKSAC45 [2] was based on its compatibility with the target organism and its ability to carry the desired genetic elements for stable integration.
π¨ BUILD
Utilizing SphI and EcoRI enzymes, the construct was successfully cloned into pKSAC45. This modified plasmid was then introduced into E. coli for amplification and plasmid preparation.
After plasmid DNA extraction failure of pKSac+LoxPort, PCR was attempted to see whether recombination happened.
π§ͺ TEST
Transformation in E. coli yielded colonies resistant to kanamycin, indicating that the plasmid was likely present and functional. Unfortunately, the following miniprep attempts consistently failed to recover sufficient DNA. Only one attempt showed plasmid extraction, but sequencing proved that it was a self-ligated form of pKSac45.
PCR reactions have been attempted to test whether recombination occurred and if it was site-specific or not.
π LEARN
The discrepancy between antibiotic resistance and failed plasmid extraction suggested that the plasmid might have been integrated in Escherichia coli DH5-Ξ± genome. This revealed limitations in using pKSAC45 as a backbone for amplification in this chassis. A redesigned approach may involve using a medium copy plasmid backbone better suited for E. coli propagation. Also, linear plasmid tansformation approach may be a possibility, as described by De Lorenzo et al. (2018) [1]
Successful Promoter Constructs (p2, pB2, pB3)
Constitutive Promoter Characterization
π DESIGN
A set of constitutive promoters was selected to evaluate their relative strengths in driving expression of sfGFP in Rhodococcus. Reporter system was based on pTip-QT1 plasmid. The reporter gene sfGFP BBa_25BE2EBO was based on part BBa_J428326, and domesticated by removal of SapI and BsrGI restriction sites for cloning inside the MCS.
Promoter set was developed starting from the paper from Round et al. 2019 [3]. We chose to study p2 and p10 promoters, since they showed a stronger constitutive activity than pNit in Rhodococcus jostii RHA1. We applied the same optimization strategy used in the paper for p10 promoter (shortening of tested sequence) on both p10 and p2 promoters, creating M2 and M6 from p10 (as performed in the paper), as well as pB2 and pB3 as novel promoters form p2.
Furthermore, we analyzed pLac, whose constitutive activity in Rhodococcus was identified in 2015 by Hetzler et al [4].
Each promoter was designed to be amplified by PCR. Primers added lg-10 RBS from pTip [2] right downstream of the promoter, and added BsrGI and NcoI restriction sistes, for substituting the resulting promoter+RBS fragment to native pTip promoter and RBS.
Positive control was performed with pNit-QT1 plasmid, carrying the widely used in Rhodococcus pNit constitutive promoter, on which MCS sfGFP was cloned, following the same approach as pNit-QT1 cloning with NcoI and NotI restriction enzymes.
π¨ BUILD
sfGFP was amplified, domesticated and flanked by NcoI and NotI restriction sites by PCR with sfGFP_f (BBa_25TME224) and sfGFP_r (BBa_25RAR337) primers . It was cloned in MCS of pTip-QT1 and pNit-QT1 plasmids, using NcoI and NotI enzymes.
Promoters were amplified by PCR with specific primers (which added a standard RBS and flanked the sequence with BsrGI and NcoI restriction sites) and cloned in pTip+sfGFP reporter plasmid, replacing native promoter and RBS.
Reporter plasmids were first transformed into E. coli by heat-shock, selected on ampicillin plates, extracted and verified by plasmid purification and digestion. Validated constructs were then transformed by electroporation into Rhodococcus opacus PD630 and plated on tetracycline-containing plates.
π§ͺ TEST
PCR amplification gave poor results for p10 and its derivative M2 and M6, so they were discarded.
Colonies obtained from Rhodococcus opacus PD630 transformations have been re-streaked and re-cultured to ensure stability. For each promoter construct, four colonies were randomly selected, grown in liquid medium to ODβββ = 0.5, and 3 for each promoter were tested in technical triplicates for fluorescence. Promoters p2, pB2, pB3 an pLac produced measurable expression. A detailed protocol can be found at experiments page.
π LEARN
The results demonstrated that while several constitutive promoters drove detectable sfGFP expression, differences showed due to technical limitations. Specifically, p10 amplification failed because primers from the literature were not compatible with our PCR conditions. pB3 promoter showed weaker activity than expected, probably because the optimization strategy used removed important regions, like -10. These failures suggest the need for primer redesign and improved annealing temperatures. The cycle highlighted both successful constitutive promoter characterization and optimization, as well as critical redesign steps for future iterations.
Anyways, promoters in the library showed consistent fluorescence among replicates and different strength between each other, opening up to a possibility to study more promoters using techniques established here.
Parts Engineering: Key Takeaways
- Sequence optimization is critical: High GC content and secondary structures were major barriers overcome through professional synthesis with codon optimization
- Backbone selection matters: Different backbones show varying compatibility with chassis organisms and cloning methods
- Fragment strategy affects efficiency: Smaller fragments improve amplification but can decrease assembly efficiency
- Promoter characterization success: Successfully validated multiple constitutive promoters (p2, pB2) for Rhodococcus expression
- Primer validation is essential: Literature primers require experimental validation for new PCR conditions
π» Software:
Iteration 1 - Excluding currency metabolites
π DESIGN
Once our KEGG network was available, we started looking for some paths. Sadly, they were full of unrealistic shortcuts through currency metabolites. Inspired by the work of Kim et al [5], we created a list of currency metabolites to exclude from the graph.
π¨ BUILD
We created a list of 58 metabolites to remove from the graph, available on our GitLab repository in the folder data/
π§ͺ TEST
The median of the average path lengths increased from around 3 to around 8. Similar results were obtained also by Kim et al.
π LEARN
Introducing a small bias by excluding some arbitrary currency metabolites allows to create a realistic metabolic representation.
Iteration 2 - Biological Meaningful Paths
π DESIGN
Even though removing currency metabolites greatly improved the topology of the network, pathways proposed were still not meaningful. Other subtle shortcuts were still present. We then decided to implement a penalty system to assign weights to edges.
π¨ BUILD
Penalty is composed as a sum of the following:
- Annotation level: KEGG white edge (+0), KEGG white edge (+2), RetroPath2.0 prediction (+4)
- Chemical dissimilarity using Tanimoto distance (+3 for chemical dissimilarity >=99%)
- Molecular weight change penalty: huge gains or losses in molecular weight are penalized (0-3)
- RetroPath2.0 Score, only for predicted reactions (0-4)
π§ͺ TEST
Several empirical tests for known paths confirmed the value of this system.
π LEARN
Introding a small a certain degree of bias is necessary to obtain biologically relevant paths. Further improvements might be represented by a "path penalty", considering more than one reaction at a time, or by integrating the concept of "module", like KEGG Modules.
π§ Hardware: Device Development
Iteration 1 β Optimization for lipase solubilization
π DESIGN
The initial aim was to achieve complete solubilization of lipase while maintaining its catalytic activity. Preliminary tests indicated that lipase exhibited partial insolubility and aggregation under standard buffer conditions; consequently, a systematic investigation of both chemical and physical factors influencing solubility was planned.
π¨ BUILD
Different buffer systems were evaluated, including sodium phosphate and Tris-HCl at different concentrations and pH values: none of these conditions provided full solubilization.
Subsequent trials incorporated several additives, such as glycerol, Tween 20, Triton X-100, DMSO, and isopropanol, alongside various physical treatments (shaking, orbital waves, vortexing). Despite these modifications, lipase remained only partially soluble, and aggregation persisted.
Further optimization involved testing bovine serum albumin (BSA) at different concentrations as a potential stabilizing agent.
π§ͺ TEST
A buffer formulation consisting of 1.5% BSA in 1 M Tris-HCl (pH 8) enabled complete solubilization of lipase without detectable aggregation: the enzyme remained stable under incubation and ready for subsequent enzymatic assays.
π LEARN
The findings demonstrated that the presence of BSA plays a crucial role in preventing aggregation and enhancing enzyme stability, likely by shielding hydrophobic regions and improving dispersion. The optimized buffer composition was established as the standard condition for all further experiments, providing a reproducible and reliable foundation for analyses.
Iteration 2 β Lipase activity assay
π DESIGN
After achieving full solubilization, the next objective was to quantify and optimize lipase activity using p-nitrophenyl palmitate (pNPP) as the substrate: the goal was to identify the enzyme concentration providing a stable, linear and sensitive colorimetric response.
π¨ BUILD
Activity assays were conducted using a Varioskan multimode plate reader set at 410 nm, employing a concentration range of 10–350 U/mL of lipase: each reaction was performed under identical conditions, with appropriate controls lacking enzyme or substrate.
π§ͺ TEST
At concentrations below 100 U/mL, the enzymatic activity was minimal, whereas concentrations above 200 U/mL resulted in rapid saturation of the absorbance signal.
An optimal performance was observed at 125 U/mL, yielding a strong and linear response within the measurable range.
π LEARN
This optimization confirmed 125 U/mL as the ideal working concentration for lipase under the established assay conditions: excessive enzyme loading was shown to compromise linearity through signal saturation.
These insights provide key parameters for integrating lipase into the complete enzymatic cascade and for guiding the refinement of enzyme ratios in subsequent calibration and device assembly stages.
Iteration 3 β Calibration curve with palmitic acid
π DESIGN
The design of this cycle focuses on establishing a robust calibration curve for the biosensor using palmitic acid as a model compound. This includes defining the range of TAGs concentrations to be tested, selecting appropriate enzymatic assay conditions, and setting the parameters for data collection to ensure high precision and reproducibility.
π¨ BUILD
This phase would involve preparing the calibration samples with different concentrations of palmitic acid, setting up the enzymatic reactions under the defined conditions, and configuring the detection system for quantitative signal acquisition.
π§ͺ TEST
To be continued!
π LEARN
To be continued!
Iteration 4 β Paper-based biosensor optimization
π DESIGN
The aim of this cycle is to optimize the performance of the paper-based biosensor. The design phase includes outlining different enzyme immobilization strategies (e.g., covalent binding, physical adsorption) and identifying possible improvements in signal detection, such as alternative colorimetric or optical readouts.
π¨ BUILD
This phase would consist of fabricating paper-based prototypes incorporating the selected immobilization approaches and preparing the setups for comparative evaluation of the different sensing formats.
π§ͺ TEST
To be continued!
π LEARN
To be continued!
π Education: Outreach & Learning Programs
Iteration 1 - University lessons
π DESIGN
We designed interactive lessons For university students enrolled in the elective courseViral and Microbial Biotechnology (link al corso) during their Bachelor’s degree in Biotechnology aimed at keeping students engaged, encouraging participation, testing their knowledge, and reinforcing concepts through explanations.
π¨ BUILD
The lessons included interactive elements such as quizzes (prepared with Quizziz) and hands-on plasmid construction exercises while explaining our project.
Project presentation at University QUIZ for Univesity lectureπ§ͺ TEST
This strategy effectively captured students’ attention. However, sometimes quiz solutions did not work as intended, as interruptions during explanations caused a loss of focus.
π LEARN
The software used for the quizzes proved cumbersome. In the future, we would structure this part more carefully beforehand or include physical materials that allow students tosee and interact with the concepts directly, enhancing their understanding and engagement.
Iteration 2 - High school students
π DESIGN
After facing several challenges while teaching university students, we realized we needed a new strategy to explain the field of synthetic biology in a simpler and more engaging way. Since high school students are less specialized than university students, we had to find an approach that could make complex concepts accessible without losing their attention. The best solution we found — since quizzes wouldn’t be as effective — was to create a board game!
π¨ BUILD
Before preparing the lessons, we focused on designing the board game itself. Our goal was to clearly represent lab activities and the steps required to bring synthetic biology to life. We started by playtesting the game within our team to refine its dynamics, and then involved PhD students from our lab to ensure that the scientific content was accurate and relevant. Later, the lessons were structured to introduce students to the theoretical concepts underlying the game’s dynamics, so they could better understand the scientific principles behind each action.
Synthetic biology lecture for High Schoolπ§ͺ TEST
We tested the game with fifth-year high school students during synthetic biology classes. After a short introduction to the fundamentals — revisiting core concepts such as the central dogma and DNA — we explained the game’s rules. Each session lasted about two hours: the first 30 minutes were dedicated to explanations, followed by about one and a half hours of gameplay.
π LEARN
The game includes a large amount of information — after all, representing lab life in a board game is no easy task!It is designed to be played for at least an hour, mirroring the reality that even a single laboratory procedure can take hours to complete. During the first lessons, we learned how to organize playtesting sessions more efficiently by dividing each class into three groups of three students. This setup allowed us to run complete game sessions smoothly and ensure that all students could fully engage with the material. This interactive approach proved far more effective than the one used with university students.
Iteration 3 - Feedback form
π DESIGN
After noticing some issues in previous university lessons, we decided to introduce the feedback form for high school students to collect suggestions, improve future sessions and make them more engaging. We therefore designed a feedback formspecifically aimed at evaluating how well our lessons were received. The questionnaire was implemented only for the high school lessons.
π¨ BUILD
form was created using Google Forms and was designed to assess students’ interest and understanding of the topic before and after the lesson. It was divided into four sections:
- Section A – Interest and Knowledge: gauging prior knowledge and initial interest.
- Section B – Methods and Activities: evaluating how the board game helped understanding.
- Section C – Overall Experience: whether the lesson was enjoyable and what could be improved.
- Section D – Future Improvements: suggestions for further enhancement.
π§ͺ TEST
The questionnaire was administered immediately after the high school lesson, with participation being completely voluntary.
π LEARN
Students provided very useful feedback. Out of around 40 participants, we received 21 completed responses, which allowed us to confirm the excellent results we had observed during the sessions. A summary of the results is included in the annexed data.
Students satisfactionLab workshops
Iteration 1 - Teacher training course
π DESIGN
We designed the first set of experimental activities for laboratory training courses aimed at high school teachers. Since the available equipment was more limited than in our usual work environment, we needed to plan a quick, simple, and reliable project that could both explain the principles of synthetic biology and teach teachers new skills to bring back to their classrooms.
π¨ BUILD
The lessons were divided into three parts: bioinformatics, synthetic biology, and experimental activity. For the bioinformatics module, we created a ChimeraX-based session to visualize and compare the amino acid sequences of GFP and its derivatives, showing how specific mutations can shift fluorescence emission spectra.
In the synthetic biology section, we introduced key concepts and assembly techniques, using a quiz to engage teachers before the practical part.
The experimental activity focused on applying Gibson Assembly to construct a plasmid containing GFP and mCherry, as well as on designing primers for specific PCR reactions. Each part of the experiment was pre-prepared by the team, so teachers could focus on performing the core procedures in the lab.
π§ͺ TEST
The full 10-hour session was divided into 2.5 hours of bioinformatics, 2.5 hours of synthetic biology, and 5 hours of experimental work. The experimental section presented some challenges — the Gibson Assembly had not been pre-tested, resulting in transformed cells expressing only GFP rather than the complete assembled construct. The primers we designed for amplifying STRs from human genomic DNA worked, but the gel resolution was low.
π LEARN
We learned that the theoretical aspects of synthetic biology can be effectively adapted for teachers and, by extension, their students. However, the experimental part requires optimization. Gibson Assembly should be fine-tuned to ensure reliable results even when performed by less experienced users, and gel electrophoresis conditions must be adjusted depending on the desired resolution to improve visualization of PCR products.
Iteration 2 - Complementary Oprional Project (COP)
π DESIGN
We designed a30-hour POC (Complementary Operational Programme) for fourth-year high school students, aimed at introducing motivated and aware participants to synthetic biology and the main steps required to apply it in the lab. Building on the structure of the previous teacher training course, we prepared a program combining lessons in bioinformatics, synthetic biology, and hands-on experimental work.
π¨ BUILD
The teacher training course served as a pilot for this program. The lessons will follow a similar structure, featuring simplified introductory bioinformatics sessions compared to the teacher version, and a fully lab-based experimental plan for the synthetic biology module. Given the 30-hour duration, students will also have the opportunity to experience the preparation and setup phase of the experiments, which had been omitted from the teacher course due to time constraints. To ensure the experimental module works smoothly, we will test Gibson Assembly protocols in advance, optimizing all steps for classroom conditions. If the assembly does not succeed as expected, we have a backup plan: we ordered a plasmid from IDT that allows expression of mCherry under the control of a different antibiotic than the GFP already included in the educational kit, so students will still be able to observe the desired results.
π§ͺ TEST
To be continued in January!
π LEARN
To be continued in January!
Surveys
Iteration 1 - Assessment of public knowledge on bioremediation and biotechnology
π DESIGN
We created a survey to assess the public's knowledge and interest in our project, bioremediation, and our chassis—our preferred future application field.
π¨ BUILD
The survey included specific questions about bioremediation knowledge and public perception. We deliberately varied terminology to identify potential biases, such as reactions to terms like "GMO" and "bacteria."
π§ͺ TEST
The survey was distributed through social media channels to evaluate public perception. We analyzed the responses and created data visualizations to identify key trends.
Survey results plotsπ LEARN
The survey revealed significant gaps in public knowledge about bioremediation. We identified strong negative biases, particularly regarding bacteria and GMOs, which are often perceived as dangerous or undesirable due to societal misconceptions and lack of accurate information.
Iteration 2 - Improving public perception of bioremediation and GMOs through social media
π DESIGN
Based on the survey findings, we identified a need to address misinformation and challenge common biases. We decided to create an Instagram reel addressing the most controversial questions and concerns raised by respondents.
π¨ BUILD
We analyzed the survey results to identify critical misconceptions. We then produced a direct-response video where team members took turns addressing these key questions and concerns in an accessible, conversational format.
π§ͺ TEST
We evaluated the effectiveness of our educational approach by tracking audience engagement metrics, including views, likes, shares, comments, and overall interaction patterns. To see what we did go to to our Instagram page!
π LEARN
The video received 594 views and 25 likes. While engagement was moderate, we recognized that our distribution strategy could have been more effective. Since the original survey was distributed through private channels and the response video was posted on a different platform (our Instagram profile), many survey respondents may not have seen our educational content. Despite this limitation, the direct-response format proved effective for addressing misconceptions and challenging biases rooted in insufficient public knowledge about biotechnology and bioremediation.
π£ Science Communication
Iteration 1 β First period
π DESIGN
We designed our science communication strategy to share engaging content on social media. The model was based on successful scientific dissemination accounts, creating mostly carousel like posts.
π¨ BUILD
The communication content was created and integrated into our Instagram profile and posted on a biweekly basis. It served as a testing ground, where the outreach strategy could be tested and observed in action.
π§ͺ TEST
The effectiveness of our communication strategy was assessed through audience interactions, including likes, shares, comments, and engagement trends.
π LEARN
We observed low engagement and a slow growth in follower, questioning whether the type and amount of content that we were creating was most fitting for our audience. These learnings were used to redesign and refine our communication strategy for improved effectiveness in future posts.
Iteration 2 β Second period
π DESIGN
We contacted several science influencers asking for advice and guidance in the production of new types of content that could be more engaging for our audience. Stefano Bertacchi showed interest in the project and became our official advisor for communication.
π¨ BUILD
With the help of Stefano, we created new types of content that integrated also memes and new trends, expanding our platforms to tik tok and LinkedIn. We started posting more frequently and posts were also re-shared on Stefanoβs socials to expose them to a bigger audience.
π§ͺ TEST
The effectiveness of our communication strategy was assessed through audience interactions, including likes, shares, comments, and engagement trends and growth in followers.
π LEARN
We observed that by posting more content our page reached a broader audience, and therefore more interactions, likes and comments. This was a positive result, but we were noticing an imbalance between the actual scientific content and the more light-hearted content, so we had to create something new.
Iteration 3 β Third period
π DESIGN
We observed what other teams were posting, and we saw that we werenβt using Instagram stories at their full potential. We decided to create a weekly column called βThursday Science Pillsβ where we talked about the most interesting articles of the week.
π¨ BUILD
We created a set of four stories containing the title of the article and a small description of the contents.
π§ͺ TEST
The effectiveness of our new content was assessed through audience interactions and likes.
π LEARN
We saw our page grow both in followings, interactions and likes and establishing a balance between scientific and funny contents.
References
βΌ- 1. DeLorenzo DM, Rottinghaus AG, Henson WR, Moon TS. Molecular Toolkit for Gene Expression Control and Genome Modification inRhodococcus opacus PD630. ACS Synthetic Biology. 2018;7(2):727-738. doi: https://doi.org/10.1021/acssynbio.7b00416
- 2. Nakashima N, Tamura T. Isolation and Characterization of a Rolling-Circle-Type Plasmid from Rhodococcus erythropolis and Application of the Plasmid to Multiple-Recombinant-Protein Expression. Applied and Environmental Microbiology. 2004;70(9):5557-5568. doi: https://doi.org/10.1128/aem.70.9.5557-5568.2004
- 3. Round JW, Roccor R, Eltis LD. A biocatalyst for sustainable wax ester production: re-wiring lipid accumulation inRhodococcusto yield high-value oleochemicals. Green Chemistry. 2019;21(23):6468-6482. doi: https://doi.org/10.1039/c9gc03228b
- 4. Hetzler S, Bröker D, Steinbüchel A. Saccharification of Cellulose by Recombinant Rhodococcus opacus PD630 Strains. Applied and Environmental Microbiology. 2013;79(17):5159-5166. doi: https://doi.org/10.1128/aem.01214-13