Results

Our goal was to set a new milestone in cancer treatment, by demonstrating that the unicellular microalga Chlamydomonas reinhardtii (Chlamydomonas) is capable of producing functional therapeutic antibodies. As a green alga, Chlamydomonas offers unique sustainability advantages for biopharmaceutical protein production: it harnesses light energy through photosynthesis, eliminating the need for expensive organic carbon sources while actively reducing CO₂ emissions during cultivation. Beyond its ecological benefits, cultivation is straightforward, cost-effective, and potentially scalable when grown in inexpensive media. However, since the use of Chlamydomonas for large-scale protein production is still under active research, we focused our efforts on proving the production and functionality of therapeutic single-chain antibodies in Chlamydomonas.

---

Getting started

Design & Modular Cloning

In order to design our first constructs, we first had to select the coding sequences. We decided to use the sequence of the single-chain variable fragment (scFv) of the Cetuximab monoclonal antibody provided by the iGEM 2015 team Apollo [1] (CetFv), as well as a humanized sequence of the Cetuximab scFv described by Banisadr et al. in 2018 [2] (HumcetFv). We reverse-translated these sequences with optimal Chlamydomonas codon usage and inserted native introns from RBCS2 approximately every 150-200 nucleotides to ensure efficient transgene expression[3]. Finally, we added BbsI restriction sites to enable Modular Cloning.

After obtaining the two sequences from TWIST, we designed our first Level 2 constructs to answer the following questions:

1. Can we produce and secrete scFvs in Chlamydomonas?

2. Can scFv production and maturation be tracked via fluorescence in Chlamydomonas?

We also designed Level 2 constructs containing an 8xHis tag for purification.

The constructs shown in Fig. 1 set the groundwork for our research. To address our first question of whether we can produce and secrete scFvs in Chlamydomonas, we transformed Level 2 constructs 9-cCA-CetFv-SP20, 3xHA (SH) and 10-cCA-HumCetFv-SH (see Fig. 1) into Chlamydomonas, grew individual transformants in liquid culture, precipitated proteins in the culture medium with acetone and analyzed them by SDS-PAGE and Western blotting using an anti-HA antibody. This demonstrated for the first time the production of scFv in Chlamydomonas, establishing the basis for further experiments.

As shown in Fig. 2, a clear HA-signal at approximately 55 kDa was detected, which corresponds to a larger than predicted molecular mass of 23 kDa. This shift can be explained by glycosylation of the SP20 module during secretion, which increases the apparent molecular mass of the protein and supports efficient secretion. So, we are confident that the detected proteins in Fig. 2 indeed are our scFvs, confirming successful scFv production in Chlamydomonas. The bands indicated below could be due to partially glycosylated or degraded scFv.

Screening via Fluorescence

While screening for transformants producing our scFvs by Western blotting, we also aimed to detect the humanized Cetuximab scFv fused with mTFP via fluorescence microscopy to address our second question. Successful detection would provide the first visualization of a scFv produced in Chlamydomonas.

For this experiment, transformants containing constructs 6-8 (see Fig. 1) were grown in a 24-well plate alongside a negative control without expression of any fluorescent protein and positive controls accumulating mTFP in the cytosol or chloroplast.

As shown in Fig. 3, only very little fluorescence leaks into the mTFP channel in the negative control. However, chlorophyll autofluorescence was clearly visible at 440 nm. The fluorescence from mTFP in the chloroplast colocalizes well with the chlorophyll autofluorescence. In contrast, the fluorescence of mTFP in the cytosol does not colocalize with chlorophyll autofluorescence. Fluorescence signals from secreted mTFP as well as HumcetFv fused with mTFP were detected in the cytosolic area in structures that could correspond to the ER/Golgi.

The results in Fig. 3 confirm, for the first time, successful detection of our scFv fused with mTFP, demonstrating the production of secreted scFv in Chlamydomonas. Although both proteins are secreted, we were able to capture them while in transit through the secretory pathway, most likely within the ER and the Golgi apparatus.

Secretion Signals

Lastly, we addressed our third question regarding the efficiency of different secretion signals for our scFvs by directly comparing the secretion signals from carbonic anhydrase (cCA), arylsulfatase (ARS2) and gamete lytic enzyme (GLE) in a single Western blot. In most cases, the cCA secretion signal is reported to function reliably and ensures efficient protein secretion. However, ARS2 and GLE can serve as alternatives if secretion levels with cCA are too low, e.g. MUT-PETase[4]. Should secretion efficiency be higher with ARS2 or GLE, it would be necessary to design new Level 2 constructs accordingly.

We selected positive transformants expressing scFv from constructs 9-14 (see Fig. 1). Liquid cultures were grown to a density of 2*10⁵ cells/ml for seven days and proteins in the culture medium were then precipitated with acetone and analyzed by SDS-PAGE and Western blotting.

As shown in Fig. 4, HumcetFv expression coupled with the cCA secretion signal resulted in markedly higher secretion levels compared to its ARS2 and GLE counterparts. In contrast, the CetFv construct displays slightly higher expression with GLE and occasionally with ARS2. Nevertheless, we decided to continue using cCA for CetFv, as the difference is not substantial and maintaining the same secretion signal allows for better comparability with HumcetFv.

These data conclude the initial construction and testing of our Cetuximab scFVs. In summary, we successfully showed secretion of scFv both by immunoblotting and by fluorescence microscopy and identified cCA as the most suitable secretion signal. Having laid this groundwork, we next aimed to explore secretion under different parameters such as culture volume or time. In parallel, we sought to refine existing protocols to reduce the workload and minimize potential hazards from vile reagents.

---

Optimization of Methods

After identifying cCA as the most effective secretion signal for our scFv constructs, we began planning the next series of experiments. One of our primary goals was to improve the trichloroacetic acid (TCA) protein precipitation protocol, as TCA is highly hazardous and difficult to handle safely. In addition, we aimed to investigate the secretion of our scFvs by varying parameters such as liquid culture volume and timing of protein harvest.

TCA vs Acetone Precipitation

Trichloroacetic acid (TCA) is a highly corrosive and hazardous chemical to work with. While TCA precipitation is routinely used in the Department of Biotechnology and Systems Biology, we were eager to improve the protocol by skipping the highly corrosive TCA-step, and precipitated with the less hazardous acetone, therefore reducing hazards, and saving time.

Chlamydomonas cultures of HumcetFv were grown at a density of 2*10⁵ cell/ml for six days. One batch was precipitated with acetone, the other with TCA. The ratio of protein concentration was adjusted by loading 20 µl of the acetone-precipitated sample onto the SDS gel and 4.1 µl of the TCA-precipitated sample (corresponding to a ratio of ~1:4.29). The remaining steps of SDS-PAGE and Western blotting were conducted as always.

As shown in Fig. 5, both TCA and acetone precipitation of HumcetFv from culture medium yielded comparable signal strengths, indicating that acetone precipitation is as efficient as TCA precipitation. Based on these results, we discontinued TCA protein precipitation and used acetone precipitation instead as the standard protocol.

Volume test

Our long-term goal in the iGEM competition was to establish scFv production in large-scale bioreactors. If successful, this would allow for mass production of scFvs without the need for numerous Erlenmeyer flasks, thereby approaching industry standards. To begin with, we needed to analyze the production of our scFvs in different liquid culture volumes to determine whether protein yield changes with culture size.

Liquid cultures of strains producing CetFv and HumcetFv, construct 9 and 10, were grown at a density of 2*10⁵ cells/ml for six days in the following volumes of media (in mL): 6, 10, 20, 50, 100, 200, 400 & 1000. After cultivation, the culture media were precipitated with acetone and analyzed by SDS-PAGE and Western blotting.

As presented in Fig. 6, the immunoblot signals of CetFv appear to be saturated, yet the signal of the band below appears to reduce in intensity, indicating a diminishing in scFv accumulation. This suggests that CetFv might be well suited for scaling up and could serve as a suitable candidate for bioreactor experiments. In contrast, the HumcetFv signal decreased with increasing culture volume, suggesting lower secretion efficiency in larger cultures. Based on these initial scale-up results, we proceeded by planning and operating several types of photobioreactors (see Outlook).

Secretion-Time Test

The secretion-time test provides insights into the temporal dynamics of scFv secretion in Chlamydomonas enabling the secretion time optimization of production protocols. Understanding when and how efficiently our two scFv variants accumulate in the culture medium is essential for maximizing protein yield and establishing standardized harvesting procedures.

The Western Blot analysis of CetFv and HumcetFv in Fig. 7 reveals distinct secretion profiles for the two scFvs. CetFv shows a gradual increase in protein accumulation starting from day one, with relatively consistent levels maintained through day 6. In contrast, HumcetFv demonstrates secretion beginning on day one, reaching high levels at day 2 that remained stable throughout the entire week. Notably, HumcetFv accumulated to much higher levels compared to CetFv, suggesting superior production/secretion efficiency or enhanced protein stability in the culture medium. The early and sustained secretion of HumcetFv indicates that it can be harvested as early as day 3 without compromising yield, potentially reducing production time. The same applies to CetFv. This time analysis establishes the foundation for optimized production schedules.

---

Proof of concept

As we gained experience and confidence in laboratory work, we began to develop ideas on how to move forward. Our primary objective was to demonstrate the interaction between our Cetuximab scFvs and the EGFR (Epidermal Growth Factor Receptor). In addition, the Department of Toxicology offered us the opportunity to test effects of our scFvs on cancer cell lines. To perform this assay, we required large amounts of purified protein and therefore scheduled weekly protein purification sessions to meet this demand. Furthermore, we anticipated designing scFv variants in which the C-terminal protein tags could be removed by a protease, addressing potential requirements for future commercial applications.

HRV Cleavage Site Design

To commercialize our therapeutic scFvs, it is essential that no additional tags such as SP20, His- or HA-tags remain on the final scFv. To achieve that, we designed a new B5 part containing a PreScission protease recognition sequence (human rhinovirus (HRV) type 14), which enables the removal of all C-terminal tags. However, due to synthesis problems with the repetitive SP20 tag, we were only able to order the B5 part with mNeonGreen (mNG). In order to obtain the B5 part with the cleavage recognition site but without mNeonGreen, we planned to remove the mNG sequence by PCR.

After the arrival of the HRV part including mNeonGreen, we assembled Level 2 constructs using Modular Cloning, according to the scheme shown in Fig. 8. Following correct assembly, we proceeded with UVM4 Chlamydomonas transformation. After two weeks of growth on TAP agar plates, we picked single colonies for screening of transformants.

From Western Blot analysis shown in Fig. 9, multiple strong bands were detected for both CetFv and HumcetFv above 70 kDa, which corresponds to the expected molecular mass of ~56.4 kDa including glycosylation. These results confirm the successful expression of both constructs. In addition, we performed fluorescence microscopy with the CetFv strain, analogous to our screening of mTFP fused scFv variants, to assess whether the observed fluorescence signals correlate with our Western Blot results.

As shown in Fig. 10, both Western Blot (Fig. 9A) and fluorescence microscopy imaging identify the same positive transformants. However, microscopy imaging detected one transformant that did not show a signal in the Western blot, whereas the Western blot revealed transformants 6 and 9, which were not detected by fluorescence microscopy. These discrepancies may result from gene silencing/activation events.

Simultaneously, we removed the mNeonGreen sequence from the B5 part by site-directed mutagenesis. For this purpose, primers were designed to anneal directly adjacent to the mNG sequence, thereby excluding it from the PCR product. After amplification, the PCR product was purified, and the original plasmid was digested with DpnI to eliminate template DNA.

As shown in Fig. 11, we successfully amplified DNA at the ~2.5 kb mark, matching the size of 2464 bp plasmid lacking the mNG coding sequence. Following Sanger sequencing confirmation of the Level 0 B5 part lacking the mNeonGreen coding sequence, we proceeded to assemble new HRV constructs and transformed them into Chlamydomonas UVM4. However, because of time limitations, we were unable to screen the new HRV transformants.

Protein Yield Estimation and Concentration

As mentioned before, the Department of Toxicology invited us to test our scFvs on cancer cell lines. For this, however, we required sufficient amounts of purified protein. Therefore, we established a routine of weekly protein purifications using the strain harboring construct 16, which produce and secrete the HumcetFv scFv with a C-terminal 8xHis tag. This process relied on immobilized metal affinity chromatography (IMAC), in which the liquid culture supernatant containing our scFv was applied to a column loaded with Nickel-NTA resin. Histidine residues of the 8xHis tag specifically bind to the nickel, while protein lacking the histidine repeats are washed away. Elution of the bound protein is then achieved by applying a high concentration of imidazole, which competes with histidine for nickel binding, thereby releasing the protein. Purification also opened the opportunity to calculate protein yield, an important metric when evaluating the potential of Chlamydomonas for future commercialization.

Approximately 1-3 L of liquid culture of cells producing HumcetFv, construct 10, were grown for seven days. After harvesting, proteins in the culture medium were precipitated with ammonium sulfate and subjected to dialysis for buffer exchange. Following one day after dialysis, the proteins were centrifuged at 25,000 g for 20 min to remove larger particles, to prevent clogging the column. Each eluate was mixed with 4xSDS and DTT buffer and then analyzed by an SDS-PAGE along with a BSA-standard for yield estimation. SDS-gels were then stained with Coomassie.

After Coomassie staining, the gel in Fig. 12A was analyzed with ImageJ to generate a calibration curve with the BSA standard to calculate protein yield. The highest yield obtained was ~49 µg/L, while the average across multiple purifications was ~32 µg/L. These values are in line with typical yields obtained with Chlamydomonas. Although these yields appear low compared to industrial expression systems, it should be noted that UVM4 strain cultures in TAP medium can only reach cell densities of approximately 2*10⁷ cells/ml. This intrinsic limitation is the reason why we routinely required liters of Chlamydomonas culture every week to provide sufficient protein for toxicology assays. Elution fractions containing our scFvs were concentrated using Amicon filters, resulting in higher protein concentrations suitable for downstream applications.

---

Pulldown Interaction-Assay

Demonstrating the activity of our Cetuximab scFvs towards the EGFR was our primary goal. Proving its interaction would not only validate our scFvs functionality but also highlight the potential of Chlamydomonas as a future contender for biopharmaceutical protein production. For this, we went for a pulldown interaction assay inspired by the work of Kiefer et al., 2022 where the authors proved the interactions between recombinant hACE2 to the SARS-CoV-2 spike protein produced in Chlamydomonas. In their experiment, hACE2, containing a deca-histidine tag, was immobilized on Nickel-NTA beads, enabling the spike protein to bind hACE2. Since the spike protein had no direct affinity for the Nickel beads, it remained bound through its interaction with hACE2 during washing steps.

In our experiment, we incubated purchased recombinant EGFR harboring a C-terminal histidine tag with ammonium sulfate concentrated Chlamydomonas culture supernatant containing Cetuximab scFv (without His tag, construct 9 and 10). 66 µl Nickel-NTA were equilibrated with PBS before adding the protein mixture. After incubation under constant inversion, the supernatant was removed and the beads washed with PBS and 5 mM imidazole. Elution of bound proteins was then conducted with PBS and 500 mM imidazole, all samples were precipitated with acetone prior to SDS-PAGE and Western blot analysis.

As shown in Fig. 13, both CetFv and HumcetFv displayed weak signals in supernatant and washing fractions, while strong bands for EGFR and the respective scFv were detected in elution fractions. In control samples without EGFR, no HA signal was detected in the elution fractions, clearly demonstrating that our scFv specifically binds to EGFR.

Although Luciferase should not bind EGFR, a very weak signal was detected for luciferase in the elution fractions. This effect is most likely caused by unspecific binding of luciferase to the nickel-NTA matrix, as it was also seen in the EGFR-free control in Fig. 13. The second negative control, PETase, showed as expected no signal in wash and elution fractions with or without EGFR.

This result provides proof of the interaction of Chlamydomonas-made Cetuximab scFVs with its target protein EGFR and marks a major success of our project!

---

Resazurin-Assay

After confirming the binding capacity of the scFv to the EGFR in the pulldown assay, we wanted to test if the scFvs can bind the EGFR of cancer cells in vitro. This assay is designed to evaluate the cytotoxic effects of various treatments on cell viability using a resazurin-based metabolic activity indicator. Resazurin, a deep blue compound, undergoes reduction to pink resorufin in the presence of metabolically active cells, providing a quantitative measure of cellular viability. The colorectal cancer cell lines HT29 and SW48 were treated with the isolated scFv (HumcetFv) and commercial Cetuximab, which serves as a positive control to enable comparison with an established inhibitor. Cell culture media and PBS + imidazol function as negative controls to establish baseline viability, while saponin (10%) and 5-fluorouracil (5-FU) at 7.5 nM serve as additional positive controls, representing known cytotoxic agents. This experimental design allows for direct assessment of the isolated scFv's efficacy against established inhibitors while ensuring proper validation through appropriate positive and negative controls.

The results of the resazurin reduction assay indicate that the therapeutic scFvs purified and isolated from Chlamydomonas culture media did not demonstrate the expected inhibitory effects on cell viability in the tested cancer cell lines. While the positive and negative controls functioned as expected, validating the assay's reliability, neither the isolated scFv nor the commercial Cetuximab exhibited cytotoxic activity under the tested concentrations. These findings prevent definitive conclusions regarding the scFv's therapeutic efficacy on the cancer cell lines, as the lack of activity observed with cetuximab suggests that the concentrations used may have been insufficient to elicit measurable cytotoxic responses.

---

Outlook

More single chain Fragments

After sufficient testing of the scFv of Cetuximab, we initiated a new iteration of construct design and testing of newly engineered scFvs from Matuzumab, Panitumumab and Trastuzumab. The aim was to evaluate both their production and their interaction with the respective target receptors, thereby establishing Chlamydomonas as a suitable host for the production of various scFvs.

Both Matuzumab and Panitumumab also target EGFR[5] [6], making them highly suitable candidates for our study. Trastuzumab was included, due to its well-established role in breast cancer therapy. However, it targets the HER2 receptor.

For each antibody, the sequences of the heavy and light variable chains were adopted[7] [8] [9] and connected by a flexible linker (GGGS₃)[10]. Codon usage, intron insertion and inclusion of BbsI restriction sites were performed in the same manner as for the Cetuximab scFvs.

The following Level 2 constructs were designed:

However, due to time constraints, we could not test the interaction with their respective receptor. This is something we look forward to addressing in the future. Nonetheless, we can confirm production of Trastuzumab single-chain fragments derived from Chlamydomonas, as shown in Fig. 16.

Scale Up

In our efforts to produce enough scFv so that Chlamydomonas could be used as a viable alternative platform to produce therapeutic antibodies, we conducted research into bioreactors as systems for the growth and harvest of therapeutic proteins. One such system is a 10 L fermenter shown in Fig. 18, previously used for the fermentation of filamentous fungi. This bioreactor had not previously been known to be capable of growing Chlamydomonas, especially strains such as UVM4 that we used, which lacks a cell wall and is therefore more susceptible to cell damage. The bioreactor showed potential in growing Chlamydomonas to stationary phase in tests with the wild-type strain. Unfortunately, constant contamination that occurred during inoculation of the bioreactor, coupled with technical difficulties with the automatic water-cooling system, prevented further experimentation.

The initial positive results of the 10L bioreactor prompted us to go for more sophisticated bioreactor systems such as the BIOSTREAM 2 – Benchtop flatpanel bioreactor for 1.8 L. This specialized bioreactor was our first planned approach into using a bioreactor designed to be used with Chlamydomonas. This bioreactor features several advantages specifically tailored for photosynthetic microorganisms, including optimized light distribution through its flat-panel design, precise pH and dissolved oxygen control, and gentle mixing that minimizes shear stress on fragile cell wall-deficient strains like UVM4.

The BIOSTREAM 2 system addresses many of the limitations encountered with the conventional 10 L fermenter. Its transparent flat-panel design ensures uniform light penetration throughout the culture media, critical for maintaining photosynthetic activity and optimal growth conditions. Additionally, the system's sterile sampling ports and automated monitoring capabilities significantly reduce contamination risks that plagued our previous experiments. We allowed the cultures to grow for six days before beginning sample collection on days 7-9 to test whether scFv accumulation occurs in the 1.8L bioreactor system. Both TAP (Tris-Acetate-Phosphate) and HMP (High Magnesium Phosphate) media were evaluated to assess scFv production under different nutritional conditions at this larger scale. This sampling strategy was designed to capture the accumulation phase when sufficient biomass and protein expression would be expected based on our previous small-scale experiments.

Western blot analysis of culture supernatants revealed successful scFv production in both media conditions, with cells grown in TAP medium showing consistently higher scFv production compared to cells grown in HMP medium across all time points (Fig. 20). These results could be due to the slower growth rate in the minimal medium HMP coupled with the bigger cell size when Chlamydomonas grows in HMP. The results demonstrate that the specialized bioreactor system can maintain stable protein production while scaling up culture volumes, bringing us closer to achieving therapeutically relevant quantities of scFv antibodies from Chlamydomonas expression systems.

In our final weeks of the iGEM competition, we had the opportunity to make one last attempt to increase cell density and protein yield by using the promising CellDEG membrane bioreactor. According to the manufacturer, this cultivation system enables high biomass accumulation by continuously enriching photoautotrophic cultures bubble-free CO₂ supply up to 10%, thereby reducing oxygen stress. Besides constant CO₂ supply, the system also supports high light intensities of over 1000 µmol photons m^-2s^-1, ensuring that the entire cultures receive sufficient illumination.

We acclimated Chlamydomonas cultures of 9-cCA-CetFv-SH, 10-cCA-HumCetFv-SH (see Fig. 1) and the wild-type strain cc1690 (used as a control) in optimized phototrophic production medium (6xP) for four days (Freudenberg et al. 2021). After incubation, eight HD100 Cultivators were each filled with 100 ml of either strain 9 or 10, while the last cultivator was filled with the cc1690 strain. All cultures were adjusted to an optical density (OD) of 0.3 at 680, 720, and 750 nm. Samples from each of the three strains were collected daily for Western Blot analysis and for determination of cell density for eight days.

As shown in Fig. 22, despite a slower initial growth phase, the CetFv & HumcetFv strains surpassed the maximum cell density typically achieved with TAP medium after approximately 80 hours of cultivation. Following this point, cultures exhibited a rapid increase in biomass, reaching a maximum cell density of 1,44*10⁸ cells per mL. After eight days, the cell density obtained in the CellDEG system was almost ten times higher than the maximum density typically observed under standard TAP cultivation. This increase in biomass raised our expectations for a corresponding improvement in protein yield.

As shown in Fig 23, protein yield is noticeably higher in cultures grown in the CellDEG bioreactor, even after three days of incubation. This demonstrates that the system is an effective tool for increasing both biomass and protein production. Although the exact fold increase in protein yield could not be quantified due to time constraints, these results strongly suggest that the CellDEG bioreactor holds great potential for future commercialization of therapeutic single-chain fragment production in Chlamydomonas.

In addition to developing new methods and systems to increase the yield of therapeutic scFvs, we also conducted research into patent landscapes and commercialization pathways for translating our experimental results into a market-ready product. Our patent research revealed that the Chlamydomonas strain UVM4 is subject to intellectual property restrictions. We identified an alternative strain carrying the same gene knockout that enables high transient expression of recombinant proteins from the Chlamydomonas CLiP library, designated CLiPUVM. This strain allows continued production of high amounts of recombinant protein without patent restrictions. As a proof of concept, we transformed this alternative strain with our cetFv and humcetFv constructs to assess whether comparable levels of recombinant protein accumulation could be achieved.

The Western blot analysis demonstrates successful scFv production in the CLiPUVM strain for both constructs enabling the production of cetFv and HumcetFv.

---

Conclusions

This leads us to the final conclusion of our iGEM project. In summary, we were not only able to demonstrate the production, secretion and purification of pharmaceutical Cetuximab scFvs in Chlamydomonas, but also proved their functionality, as shown in Fig. 13, through interaction with the epidermal growth factor receptor (EGFR). Additionally, we conceptualized the use PreScission protease for cleavage of C-terminal protein tags with successfully produced scFvs such as Trastuzumab (Fig. 17), and operated various types of bioreactors for scale up purposes (Figs. 19, 20).

We believe that our results represent noteworthy progress in research and an important step toward the sustainable and cost-effective production of biopharmaceuticals using microalgae. While we are very satisfied with the outcomes, many questions remain unanswered. For example: Does our Cetuximab scFv have anti-cancer effects resembling those of the original monoclonal Cetuximab antibody? Do all scFvs coming from Chlamydomonas function effectively? These are just a few of the many questions that need to be addressed in future studies.

For most of us, this marks the end of our iGEM journey. We are extremely grateful for the opportunity to engage in actual scientific research as young and ambitious students. We would also like to sincerely thank the Department of Toxicology and the Department of Molecular Biotechnology and Systems Biology for allowing us to work alongside them and for their continuous support. Finally, we are deeply grateful to everyone, including sponsors, who supported us throughout the iGEM project.