The 2024 UAlberta iGEM teams project, ReneWool, aimed to address the ongoing issue of textile waste accumulation both locally and globally utilizing synthetic biology. Our objective was to engineer a strain of Eschericha coli (E. coli) capable of degrading textile waste into its constituent amino acid and carbohydrate components which are then bioconverted into recombinant spider silk.

The two main areas of our lab work focused on:

1. The development of a constitutive expression vector for production of our degradation enzymes and the development of an inducible expression vector for spider silk production.
2. Optimizing the E. coli chassis used in our degradation/production system.⁴

Overview of Experiments

Our experimentation process followed five main focus points, each aimed at developing our system and its functionality.

In phase 1 of experimentations, we cloned our keratin gene into the commercial pET-22b(+) vector to produce pure protein used in the determination of michaelis-menten parameters, to test binding interaction with our keratinase (KerDZ), and validate metabolic parameters predicted by our dry lab’s metabolic flux models. We performed a growth curve to determine if the keratin-containing plasmid imposed a significant metabolic burden on three different E. coli strains transformed with the plasmid (E. coli DH5ɑ, Rosetta-gami™[Hereby referred to as Rosetta-gami] 2(DE3), and BL21) by measuring optical density (OD) at 600 nm. Overall the cells did not exhibit a metabolic burden with DH5ɑ transformants observed to have the highest biomass and the fastest growth rate of the transformed strains, followed by Rosetta-gami, indicating these two strains are optimal chassis. We also attempted to do multiple protein purification strategies for downstream experimentation which ultimately were faced with difficulties and we were unable to do any further analysis.

Phase 2 of our project shifted focus to determining the strength of four different ribosome binding sites (RBS) we identified for potential use in our expression plasmids, and the effect they had on E. coli growth in conjunction with the inducible pTac promoter. We cloned RBS1/RBS2/RBS3/RBS4 downstream of the inducible Tac-promoter and utilized pJUMP24-1A as the backbone vector. To quantify bacterial growth, we measured the optical density (OD) of the culture at 600 nm. Growth curves were performed for the following E. coli strains: DH5ɑ, K12, Rosetta-gami 2(DE3), and BL21. Furthermore, we utilized a fluorescent plate reader to measure the relative fluorescence of the sfGFP reporter gene cloned downstream of the four RBS’s to quantify their relative strength. The results from phase 2 of experimentation concluded that our constructs showed optimal growth and expression of sfGFP with RBS1 at 37^°C.

Expanding on the results we gathered from phase 2, phase 3 of experimentation led us to clone spider silk as a transcription fusion with sfGFP in our construct under the control of RBS1 (pJUMP24-1A_pTac_RBS1_SpiderSilk_sfGFP). We again opted to quantify bacterial growth via OD600 and completed a growth curve for E. coli DH5ɑ transformed with the construct, allowing us to visualize the metabolic burden on the cells. We observed slower growth and somewhat inhibited biomass production in the strain transformed with pJUMP24-1A_pTac_RBS1_SpiderSilk_sfGFP in comparison to the untransformed DH5ɑ indicating that the construct construct imposed a metabolic burden on cells.

Considering our textile waste degradation and bioconversation system utilizes two plasmids cloned into the same organism, it was crucial for us to develop another separate construct utilizing pJUMP28-1A as a backbone with a constitutional promoter (J23119) and our optimized RBS1. Our ultimate goal was to clone the genes for our degradation enzymes into a constitutive expression vector. Phase 4 of experimentation began with the assembly of the pJUMP28-1A_J23119_RBS1 construct. We aimed to examine the effect the J23119_RBS1 upstream regulatory site (URS) imposed on microbial growth (metabolic burden) in different E. coli host strains; this was quantified utilizing a growth curve generated from OD600 measurements for DH5ɑ, Rosetta-gami 2(DE3), and BL21. The overall results showed that of the transformed strains, DH5ɑ achieved the highest OD and thus biomass. The results of this experiment suggested that the J23119_RBS1 URS did not impose a significant metabolic burden on cells. We completed an additional growth curve in E. coli DH5ɑ to determine the optimal growth media for the transformed DH5ɑ strain comparing LB (86.2 mM NaCl) and Lennox LB (43.1 mM NaCl) and found no significant difference between the two media recipes.

Utilizing the results we obtained from phase 4, phase 5 of experimentation began with cloning a keratin-degrading enzyme (kerDZ) downstream of our J23119_RBS1 URS (pJUMP28-1A_J23119_RBS1_ KerDZ). We then transformed E. coli DH5ɑ, Rosetta-gamiTM 2(DE3), and BL21 with this construct and completed a growth curve in LB (86.2 mM NaCl) and Lennox LB (43.1 mM NaCl) and generated a growth curve from OD₆₀₀ measurements to determine the ideal host strain and conditions for this constructs maintenance and expression. We observed that the transformed DH5ɑ and BL21 strains exhibited nearly identical growth, making them ideal candidates for an optimized chassis; salt concentration posed no significant effect on growth.

Overall, our goal is to genetically engineer an E. coli strain to contain two constructs: a textile-waste degrading plasmid using a constitutional promoter controlling the expression of keratinase and cellulase, and an inducible expression vector controlling the production of spider silk. This design would allow us to degrade the keratin/cellulose constituents in mixed fabric waste continuously while present, and produce spider silk when the promoter is activated by a product of the enzymatic degradation activity. Our greatest achievement throughout the course of this summer was identifying, cloning, and testing a strong ribosome binding site in conjunction with different coding sequences that we could utilize in our expression constructs, allowing us to enhance protein expression which is essential for our project's functionality. Furthermore, we found success in identifying two compatible host vectors, designing a preliminary two-plasmid expression system with the modulation of gene expression we desired, and testing the metabolic burden on different E. coli chassis; these developments are crucial in establishing the groundwork for the future development of this project. At the bench level, our team struggled the most and faced many failures in terms of our cloning process (plasmid linearizations for gibson assembly, PCR amplifications, assembly via T4 ligase, cloning with Gibson assembly) which caused significant obstacles towards the progress of our project. More detail on these shortcomings and how we overcame them can be found in the notebook section of our wiki. Examining the project more broadly, our project requires significant improvements to our protein expression and purification strategies as we struggled for weeks with our keratin-31 expression and purification and were never able to do any in vitro protein analysis for our keratinase or spider silk genes. This lack of results is potentially due to the non-optimized promoters we utilized in our constructs, which are crucial regulatory sequences to gene expression and also inefficient protein extraction strategies.

Phase 1: Keratin-31 Production in pET-22b(+)

The successful assembly of our pET-22b(+) construct containing Keratin-31 (krt31) was verified using Plasmidsaurus’s whole plasmid sequencing service. An annotated map of the construct is attached below (Figure 1) along with the FASTA file we received from Plasmidsaurus.

We aimed to test whether our keratin-containing plasmid imposed any metabolic stress on the E. coli strains by measuring their growth. We transformed our sequence verified pET22-b(+)_T7_krt31 into three strains of E. coli: DH5ɑ, BL21, and Rosetta-gami 2(DE3) to compare the growth of each strain with our plasmid. Figure 2 below highlights the growth of each strain over a 10 hour period, where optical density (OD) at 600nm was taken every 2 hours. From Figure 2 we concluded that when comparing each transformed strains to the empty control strains, DH5ɑ and Rosetta-Gami perform worse than their control counterpart, whereas BL21 grows better than its control. The highest OD of the transformed strains is observed after 10 hours where DH5ɑ has an optical density of 0.421; BL21 and Rosetta-gami have OD measurements of 0308 and 0.350, respectively. With this in mind, we then observed that each strain still exhibited adequate growth, with little statistical significance between strains. Thus, we can conclude that DH5ɑ would be the best strain for plasmid propagation and Rosetta-gami would likely be the best strain to utilize for protein production.

After determining that our plasmid had no inhibitory effect on microbial growth, we then attempted to purify the keratin protein and test Krt31 expression. Rosetta-gami and BL21 cells were lysed by sonication, applying 10-second shocks, followed by 10-second rests, for thirteen cycles. From here, we performed Immobilized metal affinity chromatography (IMAC)with optimized buffers to purify our protein. Our team then ran the lysate through an IMAC column where our histidine tagged proteins were subsequently separated. The eluted fractions containing our protein of interest were analyzed on a 10% SDS-PAGE gel and visualized via coomassie staining. Figure 3 shows the protein purification results. Krt31, our protein of interest, has a molecular weight of 46kDa.⁵ Lane 2 (pETK31 R. gammi cell pellet) shows a slight band around 46kDa, likely corresponding to keratin. Seeing Keratin in lane 2 is expected, as it is an insoluble protein and would be found in the bacterial pellet¹. Given the appearance of the band and the molecular weight, we can conclude that our protein was successfully purified, but further optimization is required to ensure further purity and concentration in future experiments. Disregard lanes relating to KerDZ as they are not relevant to this section of results and no protein was observed to be purified. Our overall conclusion from the attempted protein purification of Krt31 was that the protein is extremely insoluble and was likely in inclusion bodies. This result informs our future work as we need to be considerate of how accessible purified keratin protein is to any enzyme we subject it to as our ultimate goal was to produce pure keratin to test keratinase enzyme degradation and efficiency.

Phase 2: Determining the optimal RBS for our expression vectors

All four of our constructs' cloning success was verified using Plasmidsaurus’s whole plasmid sequencing service. The sequencing results we received were annotated and presented as the plasmid maps seen below in Figure 4.

To test these constructs, we performed a growth curve analysis at 30^oC and 37^oC in four different strains of E. coli (DH5aplha, Rosetta-gami, BL21, and K12) to assess the metabolic stress via growth rate and OD₆₀₀, and observe which temperature would be optimal for growth (Figure 5 & Figure 6). This experiment aimed to compare the metabolic burden of the four constructs, differentiated by the RBS they possessed, by evaluating the growth of various E. coli strains transformed with the construct. The optimal RBS was theoretically predicted to produce the strongest sfGFP signal, while minimizing any detrimental effects on bacterial growth. Our growth curve results showed a significant increase in growth at 37^oC for each bacterial strain over the strains grown at 30o C. Cultures were incubated at both temperatures for 10 hours, with optical density (OD) measurements at 600nm taken every 2 hours. Slightly inhibited growth of the transformed strains in comparison to the untransformed controls was observed across all experiments; this was expected as there is a certain metabolic stress imposed on microbial cells when grown in antibiotic-containing media while maintaining a plasmid. RBS1 showed relatively better performance in terms of growth overall while RBS4 performed the worst overall. All transformed E. coli strains exhibited only slightly inhibited growth in comparison to the untransformed control with the exception K12, which indicated significantly inhibited growth when transformed with the constructs, especially when grown at 30^oC. This result indicated that our construct has some significant metabolic burden placed on K12 strain, and therefore it would not be optimal for use in our system. Based on these observations, we concluded that there was little metabolic stress imposed by our constructs for DH5ɑ, BL21, or Rosetta-gami and thus we would exclude K12 from future iterations of experimentation.

After determining the overall metabolic stress imposed on the different strains when transformed with the four constructs, we then moved to quantify the expression of our fluorescent reporter gene, sfGFP, in E. coli DH5ɑ. To determine expression of sfGFP we performed a fluorometric characterization experiment. E. coli strain DH5ɑ was induced with 3mM IPTG for 3 hours or with 1mM IPTG overnight and subsequently lysed via French Press. Fluorescence was then measured for ribosome binding sites 1,2,3 and 4 using a fluorescent plate reader. The results from the 3mM IPTG induction indicated that RBS1 showed a relatively strong fluorescence signal of 10.33, compared to RBS’s 2,3 and 4 signals of -0.66, -4.66, and -2.66, respectively (Figure 7) Furthermore, we observed stronger fluorescence in the 1mM IPTG induction where RBS1 showed an even stronger fluorescence signal of 105.36, compared to RBS’s 2,3 and 4 signals of -27.96, -20.3, and -7.3, respectively. From this, we then concluded that RBS1 produces sfGFP at a 1665.15% higher rate compared to RBS2. This result further confirms that RBS1 is the best ribosome binding site for our system when considering both growth and overall protein production.

Phase 3: Spider Silk under RBS1 control

After we found that RBS1 was the ideal transcriptional regulator that imposed the least metabolic stress on E. coli growth while resulting in the most optimal protein expression, we opted to clone our spider silk gene in a transcription fusion with sfGFP to allow us to directly quantify spider silk expression. Unfortunately due to time restrictions, we were only able to complete a growth curve to visualize the metabolic stress on the strains of E. coli we were attempting to optimize for a chassis. The results from our growth curve indicated that there was a significant difference between the untransformed control cells and the cells transformed with the pTac_RBS 1_Spider silk_sfGFP_pJUMP 24, indicating that the construct imposed a significant metabolic stress related to the cells ability to grow (Figure 9). Each culture was grown for 10 hours at 37^oC and the optical density (OD) was taken every 2 hours at 600nm. The data showed that although both strains grew following typical logarithmic growth, the untransformed cells significantly outperformed those that were transformed. We saw that transformed cells' exponential phase of growth lasted until around the 6 hour time point where we can see a beginning of plateauing into the stationary phase. When we compared that with the untransformed cell growth, we saw a continuation of the exponential phase until the end of our experiment (10 hour time point). More specifically, at the 10 hour time point we observed a 48.03% reduction in growth of our transformed DH5a. We concluded that there must be some metabolic strain that our construct placed on the bacteria. Other literature has stated that in E. coli, silk protein synthesis upregulates stress response proteins, and therefore hinder growth significantly.² Furthermore, strong promoter-RBS combinations on plasmids have been demonstrated to overwhelm host cellular machinery and lead to inhibited growth. Despite this, we still observed growth, though significantly diminished, and thus we determined that spider silk production would still be possible, but at a lower output than desired and that its growth rate does not affect biomass accumulation. Further optimization and experimentation is required to fully determine the optimal chassis and URS for spider silk production under an inducible promoter that does not inhibit biomass accumulation significantly.

Phase 4: RBS 1 with a constitutive promoter

In phase 4 we tested the effect of RBS1 in a constitutive promoter (J23119) in comparison to our previous experiments with pTac, an inducible promoter. We developed a new construct (J23119_RBS1_pJUMP28) using the pJUMP 28-1A vector provided in the iGEM distribution kit instead of the previous pJUMP24 vector. As previously stated, our final genetically engineered organism must contain two distinct expression plasmids, thus demonstrating the need for two different backbones. After we observed that our previous constructs have the capacity to impose metabolic stress on cells, we elected to perform a growth curve at 37^oC of this construct in 3 bacterial strains; DH5ɑ, Rosetta-gami, and BL21 (Figure 10). Our construct was not tested with a K12 E. coli strain or at 30^oC as previously described results indicated poor growth from both conditions. As expected, the untransformed DH5ɑ and Rosetta-gami exhibited the best growth, whereas of the transformed strains, DH5ɑ exhibited the highest biomass accumulation, followed by Rosetta-gami and BL21 which exhibited more similar growth. There was a significant inhibition of biomass accumulation in the Rosetta-gami transformants and both the transformed and untransformed BL21 exhibited the lowest overall biomass, thus we opted to move forward in our experimentation with DH5ɑ.

We performed an additional growth curve for DH5ɑ in both regular LB and Lennox LB to test salt stress (Figure 11). Lennox LB (43.1 mM NaCl) has 50% lower concentration of salt in the media when compared to regular LB (86.2 mM NaCl), which should relieve any potential salt stress³ associated with the antibiotic selective marker used in our construct.

Testing our J23119_RBS1_pJUMP28 construct in the E. coli strains Rosetta-gami, DH5ɑ, and BL21 allowed us to qualitatively determine if there was any metabolic difference between the growth of these strains when compared to each other. As seen in Figure 10, cells transformed with our constructs under perform when compared to the empty cell controls within the same strain. BL21 transformed was the only strain to outcompete its untransformed control, but exhibited worst overall growth when compared to transformed DH5a and Rosetta-gami. Although DH5ɑ and Rosetta-gami grew the best of the transformed construct strains, with an OD value of 0.414 and 0.338, respectively, at the end of the time points, their OD values were significantly lower than their empty control counterparts. This informed us that our construct placed some metabolic burden on DH5ɑ and Rosetta-gami. We proceeded to use DH5ɑ to determine if salt concentrations had any metabolic effect on our strains transformed with our plasmid. Figure 11 highlights the effect of salt concentrations on transformed DH5ɑ grown at at 37^oC. Our second experiment tested two salt conditions, growth with regular LB, the most commonly utilized E. coli growth media, compared to Lennox LB which has a 50% lower sodium content. We concluded from this growth curve that there is no real significant difference between the conditions. Overall these results indicated that our construct could successfully be transformed into these strains tested, with minimal inhibition on growth. We conclude that the J23119_RBS1 URS tested here would be useful for subsequent construct design.

Phase 5: RBS 1 with a constitutive promoter for KerDZ expression

Phase 5 of our project began as a continuation of phase 4, where we determined our new construct containing a constitutional promoter (J23119_RBS1_pJUMP28) did not impose a significant metabolic burden on the E. coli chassis and was potentially suitable for cloning our degradation enzymes; thus we created the J23119_RBS1_KerDZ_pJUMP28 construct and began testing the metabolic burden imposed on the cells. Once our construct was assembled with our keratinase (KerDZ), we transformed this plasmid into the DH5ɑ, BL21, and Rosetta-gami E. coli strains and performed a growth curve analysis. Cultures were grown for 10 hours and optical density (OD) was taken every 2 hours at 600nm, we again sought to compare the microbial growth between regular LB and Lennox LB as the culture media.

The growth curves of each E. coli strain in regular LB and Lennox LB allowed us to observe the metabolic burden on growth due to the construct, by highlighting the overall growth rate and biomass accumulation. Additionally, we again sought to determine whether there was a significant difference between growth in low salt (43.1 mM) or normal salt (86.2 mM) LB media. We observed that our cells transformed with the J23119_RBS 1_KerDZ_pJUMP 28 construct grow most effectively in DH5ɑ and BL21, notably at 10 hours we saw a 51.85% and 23.67% increase in growth, respectively, in comparison to Rosetta-gami at the same time point. Although we observed adequate growth from both BL21 and DH5ɑ, when compared to the untransformed cells, DH5ɑ exhibited less growth, whereas BL21 grew more than its control. These results are consistent with what we observed in phase 4 of experimentation where BL21 grew better than its control and DH5ɑ the contrary. Therefore, we determined that either BL21 or DH5ɑ would be a suitable chassis for KerDZ production. We did not observe any significantly large inhibitions of growth between any strain.

We then performed this experiment again with cells incubating in Lennox LB (86.2mM NaCl). Even in low salt media DH5ɑ transformed with the construct achieved the highest overall growth, followed by BL21, and finally Rosetta-gami, which observed the lowest overall growth. These results were consistent with our previous growth curves completed in phase 4; BL21 grew better than its untransformed control, while DH5ɑ grew less than its control. These tests confirm previous results and allow us to conclude that either BL21 or DH5ɑ would give us adequate growth with our constructs. We then compared Figure 12 and Figure 13 to determine if there was any significant difference in biomass accumulation and growth rate between the different salt concentrations. Comparing the OD measurements at the final 10 hour time point for each strain between both NaCl concentrations, we determined that these cells grew better in regular LB. DH5ɑ at 10 hours of growth had an average OD measurement of 0.38 in regular LB but a measurement of 0.269 in Lennox LB, meaning we saw a 29% increase in growth for regular LB. Similarly, BL21 showed a 16.6% increase in growth in regular LB at the 10th hour. In contrast, Rosetta-gami, when comparing both regular and Lennox LB, saw a 0.25% reduction in growth in the regular LB at hour 10. These results indicate that there is very little if any salt stress happening in the cells for regular LB compared to Lennox. We can also conclude that there is very little effect that salt concentration has on the overall growth of the E. coli strains, as there was no real significant difference between the variables themselves. Future experimentation with this construct would focus on optimizing protein expression and purification, allowing us to determine the keratin degrading capability of our recombinantly produced KerDZ enzyme.