Biosensing

---

1.1) Linear Biosensors

PCR Amplification of linear biosensors from G-blocks

All our linear biosensors (Table 1) had been ordered in whole as G-blocks from IDT. To ensure we had enough DNA for the later in-vitro transcription experiments, we performed a PCR amplification (Figure 1). The linear DNA was purified from the PCR reactions. After the purification, the DNA concentration in the tubes for all the linear construct was around 30-50 ng/µl, indicating that the PCR worked for all the constructs. However, when we ran the gel, we did not see any band for the mercury iSpinach construct (Hg iS). This show that the DNA could not be amplified, and the result is most likely an anomaly. Bands for all the other biosensors were present, however, the arsenic Broccoli biosensor (Ars Br), the arsenic iSpinach construct (Ars iS) and the lead Broccoli (Pb Br) all had very faint bands suggesting a low yield of DNA. We hypothesise that the low yield of DNA from PCR was because of secondary structures spontaneously formed by the DNA, as adding DMSO to the reaction mixture increased PCR efficiency. The extracted DNA was used for running a biosensor test.

1.2) Transcription Factors

For more information on the design of our transcription factor plasmids go to our Design page.

Construction of level 1 plasmids

After we had completed our level one assemblies for the mutated MerR (m_MerR), PbrR and ArsR transcription factor level one plasmids we ran a digestion on the plasmids using EcoR1 (Figure 2). The mutated MerR and the ArsR plasmids were around 3000bp, suggesting that the acceptor plasmid was ligated without the transcription factor insert. Only the PbrR was assembled correctly as we see a band around 3500bp which is the size band we expect if there is an insert in the plasmid. We then performed the m_MerR and ArsR assemblies again along with the MerR. We were unable to do this assembly along with the initial ones we did as we did not have the parts needed to make this assembly. We then ran a colony PCR on these three assemblies to check for the presence of our insert in the plasmids. In figure 3a we can see a smear at around 400bp which suggests that the level 1 JUMP plasmid contains the MerR insert. However, for the ArsR and mutated MerR, we saw no band (Figure 3b) showing that there was no insert present in the DNA and that the assembly failed. Due to time constraints, we could no longer pursue the ArsR and mutated MerR and try and get a working assembly. So, we only used the MerR and PbrR assemblies for protein expression analysis.

Protein Expression Analysis

To perform the protein expression analysis we transformed E. coli Bl21(DE3) with the level 1 JUMP containing the PbrR and MerR expression cassettes and performed SDS-PAGE analysis (Figure 4). The SDS-PAGE indicated that both expression cassettes produced their respective transcription factors. The presence of the band at 15 kDa in the PbrR lane indicates that it expresses PbrR as this is its predicted size. We also saw a band at 16 kDa in the MerR lane which indicates that it expresses MerR as this is its predicted size.

1.3) In vitro transcription

Once we had our linear construct and cells which expressed the transcription factors we then wanted to characterise our biosensors. To do this we performed our in-vitro transcription reactions (More information on the experiments page). We did not include the cell lysates containing the transcription factors as we wanted to see if the linear constructs would produce fluorescence without the transcriptional repression caused by the transcription factors (More information on the design page). However, none of our linear biosensors produced any fluorescence (Figure 5a) when compared to our positive control of GFP seen at the top of figure 5a. This would suggest either that there is no RNA transcription happening or that the RNA aptamer is unable to bind to the DFHBI fluorophore as it is no longer functional.

In normal light conditions after 2 hours, we noticed a slight colour change from clear to green in all our experimental in-vitro transcription reactions (Figure 5b 5c). This colour change cannot be observed under blue light and is therefore most likely a by-product of the DFHBI fluorophore. This would suggest that the fluorophore is still functional and that no RNA transcription is taking place.

To confirm this we performed a gel electrophoresis on the in-vitro transcription factor reactions (Figure 6) to see if we could detect any RNA being produced in the reactions. In the gel, we can clearly see there is DNA present, however, we could not see any RNA smear below the DNA, confirming that no RNA transcription is taking place.

To troubleshoot this problem we tried replacing all our reagents with reagents from other labs to see if they were preventing RNA transcription. However, we still could not produce fluorescence. The only reagent we could not replace from another lab was the DFHBI fluorophore. However, as we had no RNA transcription this would likely not be the cause of the problem. If we did not have time and cost restraints we would of ordered in fresh reagents to ensure that these were no the problem.

We also tried using a linear biosensor construct which is known to work. The construct was kindly given to us by Mengxi Li who works on cell-free biosensors similar to ours. However, even using her construct we were still unable to produce fluorescence. Mengxi also looked at our protocols and construct design and could see nothing wrong with them. This meant that we were unable to find the exact cause of our problem.

Bioremediation

---

2.1) Initial characterization

After synthesis and dehydration of our CMC-CA hydrogels, which are composed of sodium carboxymethylcellulose (CMC) and the citric acid (CA) crosslinker, we tested their properties and viability in water by doing swelling and degradation tests. For verifying the presence of citric acid-mediated cross-linking, we compared them to negative control ‘hydrogels’ made only with CMC. We did these tests on two different sizes and shapes of the CMC-CA hydrogel: 25 ml frog-shaped hydrogels and 1 ml box-shaped hydrogels (Figure 7).

When making the hydrogels we noticed that the control samples were thin and very fragile (easy to break when being handled) compared to the slightly thicker and more rigid CMC-CA hydrogels. The 25 ml frog-shaped hydrogels were quite fragile and fragmented while being removed from the silicon moulds probably due to the irregular mould shape and the large number of bubbles that emerged during the dehydration process.

Swelling Tests

After immersion in water for 1 hour, the swollen weights (W_S) of the hydrogels were measured and compared to the initial weight of the hydrogels (W₀) (Figure 8).

After 30 minutes of being underwater, the negative control ‘hydrogels’ had already completely disintegrated and dissolved (Figure 2A), so no measurements could be taken for them. This is because CMC is soluble in water and there was no citric acid added to the negative controls to prevent this. Therefore, this test also confirmed the presence of citric acid-mediated crosslinking in the CMC-CA hydrogels. This proves that the citric acid crosslinking is what makes the hydrogels insoluble. 

The 25ml frog-shaped CMC-CA hydrogels were able to take on water and showed a 32% increase in weight whilst the 1ml box-shaped CMC-CA hydrogels did not take on any water and had a 2% loss of weight. We think that either something went wrong in their construction or that their small size prevented them from taking on water.

Degradation tests

After the swelling tests, the hydrogels were re-dehydrated at 30-40°C overnight, then the final weights (W_F) of the hydrogels were measured and compared to the initial weight of the hydrogels (W₀) (Figure 9).

Both shaped showed significant damage (Figure 9A) and weight loss after re-dehydration as they both showed degradation degrees of around 77% (Figure 9B). This shows that the CMC-CA hydrogels could not be reused after being immersed in water and if they were to be used for a bioremediation device they could only be used once and then recycled. However, this result was expected as the hydrogels swell when submerged in water as the hydrophilic carboxyl groups interact with the water molecules, creating enough tension to maintain the water inside the polymeric mesh. However, as not all the carboxymethylcellulose or citric acid is crosslinked and incorporated into the network of the hydrogel proper. Therefore, when submerged in water the reagents that hadn’t reacted are washed out of the hydrogel. This causes a loss of mass or degradation.

2.2) Protein production

Production of CBD- and SB7-tagged Proteins and Non-Tagged sfGFP

After we successfully produced our level 1 JUMP assemblies for our CBD- and SB7-tagged proteins (SB7-sfGFP, CBD-Metallothionein (MT) and CBD-sfGFP) We also produced non-tagged sfGFP to use as controls for the fluorescence assays for assessing the performance of the CBD and SB7 tags. We then transformed them into E. coli BL21(DE3) for protein expression. After protein expression in the BL21(DE3) cell cultures, the cultures were lysed by sonication, and the lysates were run on an SDS-PAGE gel to confirm the presence of our CBD- and SB7-tagged proteins (Figure 10). This SDS-PAGE step also serves as a solubility test to confirm that all of our desired proteins are in the soluble fraction and can be properly incorporated into protein immobilisation methods.

As we saw bands at all the correct sizes on our SDS-PAGE (indicated by the red lines in Figure 10) it proved that all our cells expressed our level 1 JUMP assemblies.

2.3) Characterizing protein-hydrogel interaction

Characterising CBD Interaction with CMC-CA Hydrogels (and Comparison to Silica Immobilisation Method)

We tested the interaction of our CBD tag with our CMC-CA hydrogels by fluorescence assays of our lysates containing sfGFP constructs. We expected the fluorescence intensity of the CBD-sfGFP lysate to decrease after incubating the hydrogels in them as the protein tag could interfere with the sfGFP. We also compared the effectiveness of CBD-sfGFP binding to our hydrogels compared to SB7-sfGFP binding to Celite545 silica beads (20-100 um in diameter). Before the fluorescence assays were carried out, we measured the total protein concentration of our lysates, and then diluted all of them to 1.0 mg/ml total protein concentration for uniform measurement.

Lysate	Initial Fluorescence Intensity
Non-tagged sfGFP	40813
CBD-sfGFP	19223
SB7-sfGFP	13268

Table 2. Initial fluorescence intensities of lysates containing sfGFP constructs. All lysates were diluted to 1.0 mg/ml total protein concentration using the buffer (0.4 M Tris-HCl, pH 7.5) to allow for uniform measurement environments.

Table 2 shows that the non-tagged sfGFP lysate has a significantly higher initial fluorescence intensity than the CBD-sfGFP and SB7-sfGFP, suggesting that N-terminal CBD- and SB7-tagging of the sfGFP may affect its folding and thus, have negative impacts on protein function.

CBD Binding to CMC-CA Hydrogel Matrix

To determine level of CBD-sfGFP interaction with our cellulose-based solid materials (or SB7-sfGFP interaction with silica beads), we measured the proportion of the sfGFP constructs bound to the solids (or pellets) compared to the remaining cell lysate.

We took 20-30 mg fragments from the 1 ml box-shaped hydrogels (it was very difficult to cut out fragments of equal weight because we lacked precision cutting instruments). Due to the varying weights of our hydrogel fragments, we divided the data obtained for each of them by its weight and multiplied this by 20 to calculate the level of CBD binding to 20 mg of the hydrogel as follows:

%sfGFP Construct Bound = ((I₀-I_F) x 20/hydrogel weight)/I₀) x 100% 

Where I₀ denotes the initial fluorescence intensity of the lysate and I_F denotes the final fluorescence intensity of the supernatant after incubation of the cellulose/silica-based solids for 1 hour.

These fragments were incubated in 1 ml of lysate containing non-tagged sfGFP (control) or CBD-sfGFP for 1 hour, then they were removed. Figure 5 shows the fluorescence assay results.

Oddly, there was no significant difference in the levels of immobilisation between the non-tagged sfGFP control and our CBD-sfGFP construct, but it is worth noting that the hydrogels bound significant amounts of both sfGFP constructs (70-75%). This may suggest that the hydrogels have a significant level of non-selective absorption of proteins.

To confirm this superabsorbent property, we measured the total protein concentrations in the final supernatants and calculated the percentage of protein absorbed into the hydrogels from the lysates. This experiment revealed that a 20 mg CMC-CA hydrogel fragment absorbs, on average, 52.62% total protein content from a 1 ml lysate (with an initial total protein concentration of 1.0 mg/ml). This confirms the high levels of non-specific protein interactions with the hydrogels and so, we proceeded to alter the CBD binding test by substituting the CMC-CA hydrogel matrix with non-absorbent microcrystalline cellulose to better characterise selective CBD affinity to cellulose-based solid platforms.

CBD Binding to Microcrystalline Cellulose

20 mg of insoluble microcrystalline cellulose were incubated in 1 ml of lysate (of 1.0 mg/ml total protein concentration) containing non-tagged sfGFP (control) or CBD-sfGFP for 1 hour, then the supernatants were isolated from the pellets by centrifugation (Figure 12).

The percentage of CBD-sfGFP immobilised is 81% while that of non-tagged sfGFP control is only 10% (Figure 12A) and we also saw visibly more intense fluorescence in the lysate containing CBD-sfGFP (CBD) than the control (no CBD). This proves that our CBD tag selectively binds to cellulose and immobilises the protein fused to it. This can also be applied for the CMC-CA hydrogel matrix, because the CBD specifically binds to the beta-1,4-glycosydic linkage of the cellulose and CMC sugar backbone.

SB7 Binding to Silica Beads

20 mg of Celite545 silica beads were incubated in 1 ml of lysate containing non-tagged sfGFP (control) or SB7-sfGFP for 1 hour, then the supernatants were isolated from the pellets by centrifugation (Table 3).

Table 3. Percentages of non-tagged sfGFP or CBD-sfGFP bound to a 20 mg Celite545 silica beads incubated in 1 ml of their respective lysates (1.0 mg/ml total protein concentration).

The values were oddly significantly negative, but we would assume virtually 0% of both sfGFP constructs were immobilised to silica beads. Therefore, our SB7 tag failed to immobilise proteins to silica beads. This may be due to something wrong with the design of the SB7 tag sequence.

2.4) Metallothionein displaying 3C-Hydrogels

Zinc (II) and Nickel (II) Binding Capacity of Metallothionein-Displaying 3C Hydrogels

Our metallothionein-displaying 3C hydrogels were prepared by incubating 20-30 mg fragments of the (1 ml, box-shaped) CMC-CA hydrogels in 1 ml of the lysate (of 1.0 mg/ml total protein concentration) containing CBD-MT for 1 hour. For the negative controls, CMC-CA hydrogel fragments were incubated in 1 ml of lysate of the negative control BL21(DE3) hosting the pJUMP29 backbone plasmid with no insert.

The initial and final metal ion concentrations in the supernatants were measured by inductively coupled plasma mass spectrometry (ICP-MS). ICP-MS measures the concentration of a certain metal ion in solution. Therefore, the reduction in the number of unbound metal ions in the supernatant represents the number of metal ions sequestered in the 3C hydrogels (Figure 13).

We saw no significant difference between the control and the 3C hydrogel decorated with CBD-tagged metallothionein. The most probable reason for this would be that the hydrogel matrix plays a major role in metal ion sequestration, as we saw this effect in earlier experiments. Another reason may be that the CBD tag is affecting the metallothionein folding and therefore reducing its metal binding capacity, we also saw this effect in earlier results.

If we were to pursue this further, we would use purified proteins instead of crude lysates as they are too impure and contain metalloenzymes and metal-chelating proteins which could affect our results. We could also try to optimise the expression and purification of CBD-tagged proteins to ensure correct folding and so, allow effective solid-phase immobilisation while maintaining protein function.

Directed Evolution

Determination of the Maximum Inhibitory Concentration

We established the Maximum Inhibitory Concentration (MIC) of untransformed BL21(DE3) and TOP10 cells by plating on agar plates with a range of AgNO₃ concentrations, starting with 8 mg/L until 16 mg/L, where we observed a fall-off in growth of BL21(DE3) cells (See figure 14)

With this, we had determined that the MIC of AgNO₃ for BL21(DE3) cells was 16mg/L, as there was a growth fall off afterwards.

Screening metallothionein mutants

Next we used error-prone PCR to generate mutants of M. edulis, M. galloprovincialis, D. rerio, and S. cerevisiae MTs. We made transcriptional units with the error prone PCR product, and found the following colony counts (See figure 15).

MTs of almost all species except for S. cerevisiae and M. galloprovincialis grew as lawns at 16 mg/L and 17 mg/L AgNO₃. We were not able to do any screening, however we were able to propose a new selection method for MT function. Since no mutants were screened, we did not use the directed evolution mutants in the rest of our bioremediation device.

Our iGEM Team also took part in the iDEC Competition (International Directed Evolution Competition). To read more in depth about our directed evolution exploration, check out our wiki.

PET Bioconversion

---

3.1) Assembly

Overview of all planned constructs results

17 of 22 of planned Lv.1 JUMP plasmids were successfully constructed (See Figure 16). The enzyme solubility or activity were assessed. All the successful PETase-related constructs were immobilised, and the activity of them was assessed. We chose the best composite part [C-terminal L2NC-linker-tagged FAST-PETase] for the proof-of-concept.

Lv.1 JUMP plasmid construction

After JUMP assembly, successful assemblies were selected from Blue-White Screening and further verified by colony PCR. The successful assemblies are listed below in Table 4:

Part Name	Shorthand name	Link to parts repository
Untagged Dou-PETase	[Dou-PETase]	BBa_K4390088
N-terminal Car9-tagged Dou-PETase	Car9-linker-[Dou_PETase]	BBa_K4390085
Untagged Tri-PETase	[Tri_PETase]	BBa_K4390089
N-terminal L2NC-linker-tagged Tri-PETase	L2NC-linker-[Tri_PETase]	BBa_K4390116
N-terminal Car9-tagged Tri-PETase	Car9-linker-[Tri_PETase]	BBa_K4390086
C-terminal L2NC-tagged Tri-PETase	[Tri_PETase]-L2NC	BBa_K4390082
C-terminal L2NC-linker-tagged Tri-PETase	[Tri_PETase]-linker-L2NC	BBa_K4390083
C-terminal Car9-tagged Tri-PETase	[Tri_PETase]-linker-Car9	BBa_K4390081
Untagged FAST-PETase	[FAST_PETase]	BBa_K4390090
N-terminal L2NC-linker-tagged FAST-PETase	L2NC-linker-[FAST_PETase]	BBa_K4390114
N-terminal Car9-tagged FAST-PETase	Car9-linker-[FAST_PETase]	BBa_K4390084
C-terminal L2NC-tagged FAST-PETase	[FAST_PETase]-L2NC	BBa_K4390076
C-terminal L2NC-linker-tagged FAST-PETase	[FAST_PETase]-linker-L2NC	BBa_K4390077
C-terminal Car9-tagged FAST-PETase	[FAST_PETase]-linker-Car9	BBa_K4390075
Untagged MHETase	[MHETase]	BBa_K4390091
N-terminal L2NC-linker-tagged MHETase	L2NC-linker-[MHETase]	BBa_K4390118
N-terminal Car9-tagged MHETase	Car9-linker-[MHETase]	BBa_K4390087

Table 4. Successful JUMP assemblies related to PETase immobilisation on silica beads.

The 4 assemblies with N-terminal L2NC silica tag failed (Figures 17 and 18). We tried to move the L2NC DNA fragment from plasmid containing C-terminal-L2NC sequence to new plasmids for constructing N-terminal-L2NC silica tag. However, the PCR of the C-terminal-L2NC silica tag failed.

3.2) Protein expression and solubility test

The Lv.1 plasmids were transformed into E. coli SHuffle strains for protein expression. We then ran an SDS-PAGE to see if our proteins were subsequently expressed (Figure 19). This SDS-PAGE step also serves as a solubility test to confirm that all our desired proteins are in the soluble fraction and can be properly incorporated into protein immobilisation methods (Figure 20).

From Figures 19 and 20, bands with correct weights were observed (labelled with a red line). This confirmed that we had successfully expressed untagged PETase, Car9-PETase, L2NC-linker-PETase, untagged MHETase, Car9-MHETase and L2NC-linker-MHETase.

3.3) Immobilisation and Activity assessment

PETase-silica tag fusion protein Activity Test

We assessed the PETase mutants' activity based on para-nitrophenol-butyrate (pNPB) assay, since pNPB can be hydrolysed by PETase into para-nitrophenol (pNP) with maximum absorbance at 415 nm. This is a preliminary assay to determine the activity of PETase, although pNPB has structural differences to the polyethylene terephthalate which is the real substrate of PETase.

First, we assessed the FAST-PETase constructs activity towards pNPB, comparing to Dou-PETase (Figure 21). Dou-PETase is the old part from iGEM repository (BBa_K3946023).

From the first set of pNPB assay (Figure 21), we observed that the protein sample containing untagged FAST-PETase showed the highest activity towards pNPB under the reaction condition over protein samples containing other constructs and empty control. All protein samples containing FAST-PETase-silica tag fusion proteins, except [L2NC-linker-FAST_PETase], showed activity towards pNPB higher than the empty control. The FAST-PETase-silica-tag fusion proteins showed similar activities as the untagged Dou-PETase. The empty control still had measurable activity towards pNPB, which supposed the E. coli SHuffle strain constitutively expresses certain Esterases that can hydrolyse pNPB. Generally, C-terminal tagged FAST-PETase showed higher activity than the N-terminal tagged ones. We supposed this may be due to elongated N-terminal flexible structure may interact and inhibit the FAST-PETase activity towards pNPB.

We found the 96 wells plate and plate reader were not suitable to assess the PETase activity after they immobilized on the actual silica beads. Therefore, we changed the reaction device to 1.5ml Eppendorf tube and spectrometer. Data for Tri-PETase and FAST-PETase were measured across different days (Figure 22). We observed an inconsistency in the empty control activity towards pNPB across different days. Therefore, we calculated the fold-change of individual protein sample activity towards the empty control in the same batch to reduce the inconsistency when comparing the data.

In general, all protein constructs showed higher activity towards pNPB comparing to the empty control, which means PETase-silica tag fusion proteins should still be soluble and active. All protein samples containing Tri-PETase or FAST-PETase constructs showed higher fold change of activity towards pNPB compared to the [Dou_PETase], the old part, under the reaction conditions. The [Car9-linker-Tri_PETase] showed the best activity when we calculated the fold-change in activity towards pNPB. Both Tri-PETase and FAST-PETase showed increased activity when they fused with silica tags.

PETase Immobilization

We incubated the protein sample containing PETase mutants with the silica beads (Celite 545) after activity assessment.. The protein concentration was measured by Bradford assay before and after incubation.

From the immobilization efficiency, the protein samples containing [FAST_PETase-linker-Car9] showed the highest immobilization efficiency over all the constructs (100%). And the protein samples containing [FAST_PETase-linker-L2NC] showed the lowest immobilization efficiency (11.93%). The Tri-PETase with silica tag shows low immobilization efficiency comparing to the FAST_PETase with silica tags after we corrected the data with empty control respectively.

The [C-terminal L2NC-tagged FAST-PETase] and [C-terminal L2NC-linker-tagged FAST-PETase] showed the immobilization efficiency (18.65% and 11.93%, respectively), lower than the empty control’s in the same batch. One possible reason might be the positively charged L2NC would interact with negatively charged PETase surface and inhibit the immobilization function.

We observed that the protein samples containing no silica-tag fusion proteins and untagged PETase both showed immobilization behaviour to the silica beads when they were not attached with any silica tags. Two main reasons may cause the results:

1. The container we used to incubate protein with silica beads was 1.5ml Eppendorf tube, and when using it, the certain portion of silica beads always precipitated at the bottom during incubation. The proteins may be trapped in the silica beads precipitated at the bottom. And the unspecific interaction between immobilized PETase and the surrounding proteins may also contribute to the immobilization rate.
2. The surface of the silica support is negatively charged at pH greater than three due to the deprotonated silanol groups. And PETase has a highly polarized surface charge, creating a dipole across the molecule and resulting in an overall isoelectric point (pI) of 9.6. Consequently, there is chance that untagged PETase can immobilized on the silica beads without the facilitation of silica tags.

Immobilized PETase activity assessment

We assessed the immobilized PETase activity based on pNPB assay. Same as before, we calculated the fold-change of immobilized protein sample activity over the empty control in the same batch (See Figure 24).

The [FAST_PETase-linker-L2NC] shows the highest activity towards pNPB after immobilizing on the Celite545 (Figure 24) (16.62-fold higher than empty control), but its immobilization rate was the lowest (Figure 25). Referring to the relatively low immobilization efficiency of [FAST_PETase-linker-L2NC], we assumed that the protein loading on each silica bead should be low. If not, the enzymes may crowd together and inhibit each other’s activity. The activity of all Tri-PETase-silica tag fusion protein was lower than the empty control, and the activity of most FAST-PETase-silica tag fusion protein was at the similar level as the empty control. The data indicated that the protein loading per silica bead should be well defined to maintain the enzyme activity after immobilization.

To assess the reusability of [FAST_PETase-linker-L2NC], we tested the same sample’s activity again on pNPB assay after washing out the remaining substrate and product from last reaction (See Figure 25).  Consequently, the remaining activity of [FAST_PETase-linker-L2NC] was 62.96%.

PET biodegradation

We chose the immobilized [FAST_PETase-linker-L2NC] construct to continue the PET fragment degradation experiment. The heat-treated PET plastic sample was provided by Dr. Joanna Sadler. We increase the pH of reaction system since the product of terephthalic acid (TPA) would lower the pH of reaction environment. We assessed the weight loss of PET plastic fragments after reacting with immobilized PETase.

After the reaction, we took out all the PET fragments from the reaction system and dehydrated them in 37°C incubator for 2 days. We observed the fissure in the plastic fragment (red square in Figure 26), and the less irregular edges of the plastic fragment (red circle in Figure 26). The results showed that the immobilized FAST-PETase can slightly degrade the PET plastic. The weight of PET fragment before reaction was 51.8 mg, and after reaction was 51.6mg. There was no significant weight loss of PET fragment, and it may result from multiple factors:

1. The optimized working temperature of FAST-PETase is 50°C, which is 13°C higher than our condition;
2. The amount of [FAST_PETase-linker-L2NC] loaded on each silica beads is low, at most 107.34µg;
3. Although the immobilized [FAST_PETase-linker-L2NC] can still react with pNPB, the C-terminus silica tag may interfere the interaction between FAST-PETase and actual PET polymer in reaction. Therefore, we supposed further assessment based on N-terminal tagged FAST-PETase should be conducted with PET sample after optimized immobilization. Considering the higher inhibition effect of N-terminal silica tag (Figure 20), we can try to substitute the N-terminal unstructured peptides (M1 to A27) with the silica tag.
4. Since the PET fragment is insoluble, PETase may react better with it in relatively static environment, instead of continuously shaking. Also, proteins with silica tag may immobilise in the mesopores in the silica beads, and they can’t directly contact with the insoluble PET polymer.