Update results.md

docs/project/results.md

@@ -2,9 +2,9 @@
 
 ## Selection of of the E3 ligase
 
-After evaluating several well-known human E3 ligases that could be engineered, we decided to work on evolving SIAH1 and SIAH2 (collectively referred to as SIAH1/2 in this text), which are primarily associated with cellular stress response, such as hypoxia [^E3_1][^E3_2]. Both belong to the RING family of E3 ligases and share 86% sequence identity with nearly identical substrate-binding domains[^E3_3]. Their small sizes—282 amino acids for SIAH1 and 324 for SIAH2—are optimal for phage-assisted continuous evolution (PACE), as this allows for efficient expression in E. coli and packaging into M13 bacteriophages. Moreover, this small size compared to other E3 ligases also reduces the theoretical library size which is usually advantageous in directed evolution experiments. In general, neither SIAH1 nor SIAH2 are individually characterised to the extent we would like them to be but together they generate a full picture. SIAH2 has already been used successfully in _E. coli_ ubiquitination assays without any partner proteins except E1 and E2, suggesting that additional regulatory proteins aren’t required for its ubiquitination activity [^E3_4]. This streamlines the evolution processes by reducing potential points of failure in the selection system and keeps the plasmid sizes within an acceptable range. Given how similar SIAH1 is to SIAH2 in structure and function, we anticipate it will behave similarly in _E. coli_ as well. While these studies are missing for SIAH1, the binding of SIAH1 to specific peptide sequences has been well-studied via X-ray crystallography [^E3_5][^E3_6], which is missing for SIAH2. Together, the data available for SIAH1/2 offer a solid foundation for targeting new sequences and evolving either protein to recognize non-canonical substrates.
+After evaluating several well-known human E3 ligases that could be engineered, we decided to work on evolving SIAH1 and SIAH2 (collectively referred to as SIAH1/2 in this text), which are primarily associated with cellular stress response, such as hypoxia[^E3_1][^E3_2]. Both belong to the RING family of E3 ligases and share 86% sequence identity with nearly identical substrate-binding domains[^E3_3]. Their small sizes—282 amino acids for SIAH1 and 324 for SIAH2—are optimal for phage-assisted continuous evolution (PACE), as this allows for efficient expression in E. coli and packaging into M13 bacteriophages. Moreover, this small size compared to other E3 ligases also reduces the theoretical library size which is usually advantageous in directed evolution experiments. In general, neither SIAH1 nor SIAH2 are individually characterised to the extent we would like them to be but together they generate a full picture. SIAH2 has already been used successfully in _E. coli_ ubiquitination assays without any partner proteins except E1 and E2, suggesting that additional regulatory proteins aren’t required for its ubiquitination activity[^E3_4]. This streamlines the evolution processes by reducing potential points of failure in the selection system and keeps the plasmid sizes within an acceptable range. Given how similar SIAH1 is to SIAH2 in structure and function, we anticipate it will behave similarly in _E. coli_ as well. While these studies are missing for SIAH1, the binding of SIAH1 to specific peptide sequences has been well-studied via X-ray crystallography[^E3_5][^E3_6], which is missing for SIAH2. Together, the data available for SIAH1/2 offer a solid foundation for targeting new sequences and evolving either protein to recognize non-canonical substrates.
 
-The SIAH family recognizes its targets through a PXAXVXP degron motif [^E3_7], where conserved residues Pro, Ala, Val, and Pro face the SIAH binding pocket (Figure 1). Specificity is mainly determined by the Ala and Val residues in positions 3 and 5, as these pockets are too small to accommodate larger side chains [^E3_8]. Additionally, the Pro residue at position 7 interacts with Trp178 in SIAH, contributing further specificity. Among canonical substrates, the VXP sequence is highly conserved [^E3_6][^E3_9], while other residues show more variability, making them prime candidates to alter SIAH1/2 specificity for these positions.
+The SIAH family recognizes its targets through a PXAXVXP degron motif[^E3_7], where conserved residues Pro, Ala, Val, and Pro face the SIAH binding pocket (Figure 1). Specificity is mainly determined by the Ala and Val residues in positions 3 and 5, as these pockets are too small to accommodate larger side chains[^E3_8]. Additionally, the Pro residue at position 7 interacts with Trp178 in SIAH, contributing further specificity. Among canonical substrates, the VXP sequence is highly conserved[^E3_6][^E3_9], while other residues show more variability, making them prime candidates to alter SIAH1/2 specificity for these positions.
 
 <figure markdown>
 ![Figure_E3](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/SIAH1_peptidebinding.png)
@@ -22,7 +22,7 @@ The selection of canonical SIAH1/2 targets, needed to test the evolution logic,
 4. The size of the canonical substrate should not exceed 300 amino acids in size, as this could also compromise our synthetic evolutionary system
 5. The structure of the protein and interaction with the selected E3 ligase (SIAH1/2) should be characterised.
 
-Based on these five criteria, we searched published ubiquitination databases [^substr_2][^substr_3] for canonical substrates. EGLN1, EGLN3 and ɑ-Synuclein were selected as canonical substrates from the mentioned databases as they matched our criteria the best. 
+Based on these five criteria, we searched published ubiquitination databases[^substr_2][^substr_3] for canonical substrates. EGLN1, EGLN3 and ɑ-Synuclein were selected as canonical substrates from the mentioned databases as they matched our criteria the best. 
 
 | Canonical substrate  | Protein size (aa) | Degron | Ubiquitination site | Canonical E3 ligase |
 | ------------- | ------------- | ------------- | ------------- | ------------- |
@@ -34,7 +34,7 @@ Based on these five criteria, we searched published ubiquitination databases [^s
 
 ɑ-Synuclein is well known for causing Parkinson's disease (PD) and multiple systems atrophy (MSA). Therefore, optimization of SIAH1/2 binding to ɑ-Synuclein could be an attractive alternative evolution strategy in case reprogramming SIAH1/2 specificity does not work as planned.
 
-NLRP3 was selected as an evolutionary target, for its mentioned clinical relevance in hepatic and neurodegenerative diseases. Furthermore, NLRP3 is an attractive target as it contains two VXP motifs (at positions 200 and 707), located in disordered regions on the surface of NLRP3 (Figure 2). This motif seems to be highly conserved among canonical SIAH1/2 substrates. Therefore, we assume our target choice is limited to proteins naturally displaying this motif on its surface, while other residues involved in binding could be changed. This would allow an evolution of SIAH1/2 to adapt for recognizing different VXP surrounding residues while retaining the conserved VXP binding motif. Furthermore, lysins are found within the VXP surrounding residues, which are needed for ubiquitination. Remarkably, one of these lysins (K689) is natively polyubiquitinated and leads to the canonical degradation of NLRP3 [^substr_1], hence suggesting that polyubiquitination of K689 is a viable strategy for inducing NLRP3 degradation. With 1036 amino acids and its potential for oligomerization, NLRP3 could constrain our evolutionary system. Thus two peptide fragments of NLRP3 containing the VXP motif as well as surrounding residues were incorporated in the selection system.
+NLRP3 was selected as an evolutionary target, for its mentioned clinical relevance in hepatic and neurodegenerative diseases. Furthermore, NLRP3 is an attractive target as it contains two VXP motifs (at positions 200 and 707), located in disordered regions on the surface of NLRP3 (Figure 2). This motif seems to be highly conserved among canonical SIAH1/2 substrates. Therefore, we assume our target choice is limited to proteins naturally displaying this motif on its surface, while other residues involved in binding could be changed. This would allow an evolution of SIAH1/2 to adapt for recognizing different VXP surrounding residues while retaining the conserved VXP binding motif. Furthermore, lysins are found within the VXP surrounding residues, which are needed for ubiquitination. Remarkably, one of these lysins (K689) is natively polyubiquitinated and leads to the canonical degradation of NLRP3[^substr_1], hence suggesting that polyubiquitination of K689 is a viable strategy for inducing NLRP3 degradation. With 1036 amino acids and its potential for oligomerization, NLRP3 could constrain our evolutionary system. Thus two peptide fragments of NLRP3 containing the VXP motif as well as surrounding residues were incorporated in the selection system.
 
 <figure markdown>
 ![Figure_nlrp3](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/NLRP3_figure.png)
@@ -47,7 +47,7 @@ NLRP3 was selected as an evolutionary target, for its mentioned clinical relevan
 
 To evolve the SIAH1/2 E3 ubiquitin ligases using the PACE system, we needed to link the ligase’s activity directly to phage propagation. To achieve this, we utilised a modified T7 bacteriophage RNA polymerase (RNAP) that had been split into two halves. Normally, this split RNAP is inactive unless both halves are brought close together within the cell, forming a complete, functional complex. We designed a system where the RNAP halves would only assemble if the E3 ligase successfully ubiquitinated its target.
 
-In this setup, one half of the RNAP is fused to ubiquitin, while the other half is fused to a substrate that can be ubiquitinated. If SIAH1/2 (E3) ubiquitinates the substrate, the two RNAP halves are brought close enough to combine, forming a complete RNAP capable of transcribing genes. We took advantage of this by placing the gene _gIII_ (coding for pIII, a protein crucial for phage propagation) under the control of a T7 promoter, which only the assembled RNAP can activate. As a result, phages can only propagate if SIAH1/2 effectively ubiquitinates the target, tying the phage’s propagation to the ligase’s activity (Figure 3a). Additionally, we also designed a system that could select against specific off-target effects (Figure 3b). To this end, we would add an additional genetic circuit which ‘punished’ E3 ligases that recognised unwanted targets. Instead of the substrate towards which we want to evolve, we fuse the unwanted protein (dubbed the mock substrate) to the C-terminal  RNAP subunit. This could be the original canonical substrate of the E3 or a protein recognized as an off-target in later stages of the SIAH1/2 evolution. If this mock substrate is recognized by a SIAH1/2 variant it would also trigger the formation of a functioning RNAP through the ubiquitination of the mock substrate. Crucially, the C-terminal RNAP fused to the mock substrate differs from the one present in the positive selection as it does not bind this T7 promoter sequence, but a slightly altered one. This ensures that one can perform the positive and negative selection at the same time. The altered T7 promoter sequence (*PT7; Figure X) is then recognized by the negative selection RNAP which leads to the transcription of a mock _gIII_. Phages that incorporate the encoded protein are essentially unable to infect new cells, leading to the washout of phages carrying SIAH1/2 variants that recognize unwanted substrates.
+In this setup, one half of the RNAP is fused to ubiquitin, while the other half is fused to a substrate that can be ubiquitinated. If SIAH1/2 (E3) ubiquitinates the substrate, the two RNAP halves are brought close enough to combine, forming a complete RNAP capable of transcribing genes. We took advantage of this by placing the gene _gIII_ (coding for pIII, a protein crucial for phage propagation) under the control of a T7 promoter, which only the assembled RNAP can activate. As a result, phages can only propagate if SIAH1/2 effectively ubiquitinates the target, tying the phage’s propagation to the ligase’s activity (Figure 3a). Additionally, we also designed a system that could select against specific off-target effects (Figure 3b). To this end, we would add an additional genetic circuit which ‘punished’ E3 ligases that recognised unwanted targets. Instead of the substrate towards which we want to evolve, we fuse the unwanted protein (dubbed the mock substrate) to the C-terminal  RNAP subunit. This could be the original canonical substrate of the E3 or a protein recognized as an off-target in later stages of the SIAH1/2 evolution. If this mock substrate is recognized by a SIAH1/2 variant it would also trigger the formation of a functioning RNAP through the ubiquitination of the mock substrate. Crucially, the C-terminal RNAP fused to the mock substrate differs from the one present in the positive selection as it does not bind this T7 promoter sequence, but a slightly altered one. This ensures that one can perform the positive and negative selection at the same time. The altered T7 promoter sequence (*PT7; Figure 3) is then recognized by the negative selection RNAP which leads to the transcription of a mock _gIII_. Phages that incorporate the encoded protein are essentially unable to infect new cells, leading to the washout of phages carrying SIAH1/2 variants that recognize unwanted substrates.
 
 <figure markdown>
 ![Figure_positive_negative_selection](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/E3_selection_V1.png)
@@ -73,23 +73,23 @@ We plan to run this system in a bioreactor to create a continuous evolutionary e
 ## Experimental results
 
 ### Is phage propagation dependent on the selection phage? 
-We tested if the E3 ligases SIAH1 and SIAH2 can trigger phage replication when a specific target protein is present. We used the protein EGLN3, which is known to be recognised by SIAH1 and SIAH2. We attached EGLN3 to part of the T7 RNA polymerase enzyme and infected cells with a few phages carrying the SIAH1 or SIAH2 gene. We measured the amount of phage produced after incubating the cells overnight. The phages carrying SIAH1 and SIAH2 replicated more than phages with an unrelated protein. This experiment indicates that our system worked.
+We tested if the E3 ligases SIAH1 and SIAH2 can trigger phage replication when a specific target protein is present. We used the protein EGLN3, which is known to be recognised by SIAH1 and SIAH2. We attached EGLN3 to part of the T7 RNA polymerase enzyme and infected cells with a few phages carrying the SIAH1 or SIAH2 gene. We measured the amount of phage produced after incubating the cells overnight. The phages carrying SIAH1 and SIAH2 replicated more than phages with an unrelated protein (Figure 5). This experiment indicates that our system worked.
 
 
 <figure markdown>
 ![Figure_initial_difference](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/Fig1a_wiki.png)
 <figcaption> Figure 5: Phages carrying SIAH1 or SIAH2 grow much better than phages carrying an unrelated protein. The fold change shows how much the number of phages increased or decreased (If the number of phages doubles, that is a two-fold increase).</figcaption>
 </figure>
 
-We then looked at whether changing the target protein affects phage replication. We replaced EGLN3 (blue) with α-Synuclein (orange), which is also recognised by SIAH1/2. Changing the target protein changes the level of phage propagation. This shows that our system depends on SIAH1 or SIAH2 and the target protein interacting together. 
+We then looked at whether changing the target protein affects phage replication. We replaced EGLN3 (blue) with α-Synuclein (orange), which is also recognised by SIAH1/2. Changing the target protein changes the level of phage propagation (Figure 6). This shows that our system depends on SIAH1 or SIAH2 and the target protein interacting together. 
 
 <figure markdown>
 ![Figure_initial_difference](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/Fig1b_wiki.png)
 <figcaption> Figure 6: Phage propagation is dependent on the substrate. The fold change shows how much the number of phages increased or decreased. </figcaption>
 </figure>
 
 
-Next, we looked at how changes in the degron would affect our system. We found that even when we altered key parts of the degron in EGLN3, phage replication was not significantly affected. This suggests that phage propagation might be triggered by other factors than the degron. 
+Next, we looked at how changes in the degron would affect our system. We found that even when we altered key parts of the degron in EGLN3, phage replication was not significantly affected (Figure 7). This suggests that phage propagation might be triggered by other factors than the degron. 
 
 <figure markdown>
 ![Figure_degron_dependency](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/Fig1c_wiki.png)
@@ -104,7 +104,7 @@ We suspected that the background phage propagation we observed could be caused b
 1. Leaky transcription of _gIII_: the gene controlling phage growth is turned on without the split-RNAP components being present. 
 2. Spontaneous assembly of the split-RNAP subunits: The two halves of the RNAP are coming together on their own, without the need for ubiquitination of the target substrate.
 
-Both hypotheses lead to _gIII_ expression independent of the E3 ligase activity. To test these hypotheses, we quantified phage propagation in cells that had only one half of the RNAP. In these cells, phage propagation was suppressed, confirming that both halves of the RNAP are required to activate _gIII_ expression. This means that accidental activation of _gIII_ wasn’t the issue. Instead, these results show that the two enzyme halves were probably joining together on their own, causing phage propagation without the involvement of the E3 ligase.
+Both hypotheses lead to _gIII_ expression independent of the E3 ligase activity. To test these hypotheses, we quantified phage propagation in cells that had only one half of the RNAP. In these cells, phage propagation was suppressed, confirming that both halves of the RNAP are required to activate _gIII_ expression (Figure 8). This means that accidental activation of _gIII_ wasn’t the issue. Instead, these results show that the two enzyme halves were probably joining together on their own, causing phage propagation without the involvement of the E3 ligase.
 
 <figure markdown>
 ![Figure_split_RNAP_parts](https://idec-teams.github.io/2024_Evolution_Suisse/img/result figures/Fig4_wiki.png)