Engineering | WHU-China

Aiming to increase the efficiency of verifying the feasibility of our project, the students in the wet lab group were divided into three subgroups and carried out three sets of experiments on different goals.

The whole engineering process is conducted under the principles of the DBTL (Design, Build, Test, Learn) cycle. The following are the engineering successes we have completed.

Part 1: Secretion

The overall goal of the secretion module is to design a small peptide surface display system in E. coli and verify the enzymatic cleavage of the system.

Cycle 1: surface display verification

● Design
In order to verify the surface display system, we design two plasmids: Lpp'OmpA_lptE and lptE, using the plasmid pET28a as the backbone.

In lptE, full-length lptE(signal peptide included) is fused with a FLAG tag. The fusion protein is controlled by T7 promoter. Once expressed, the full-length lptE will anchor in the inner leaflet of the outer membrane, which is unable to be displayed on the surface.

In Lpp'OmpA_lptE, lptE whose signal peptide is cleaved with FLAG tag is fused at the C-terminus of Lpp'OmpA, activated by T7 promoter. Once expressed, Lpp'OmpA_lptE will display on the surface of the outer membrane and be detected. T7 promoter is regulated by lac operator and terminated by double terminators(BBa_B0015).

Fig. 1 Genetic circuits of lptE and Lpp'OmpA_lptE.

● Build
For lptE, lptE-FLAG was synthesized by Genscript, and subcloning was performed to get the final product.

Fig. 2 Map of lptE

For Lpp'OmpA_lptE, lpp'ompA-lptE(signal peptide removed)-FLAG was synthesized by Genscript and subcloning was performed to get the final product.

Fig. 3 Map of Lpp'OmpA_lptE

● Test
Plasmids were transformed into E. coli BL21(DE3) separately. Sequencing verified transformation success.

Fig. 4 Sequencing of lptE and Lpp'OmpA_lptE transformant

Western Blot tests were conducted to verify the protein expression in engineered bacteria. Immunofluorescence assay(IFA), FACS analysis, and Cryo-electron microscopy verify the surface display result. To see more details, visit Results page

● Learn
During our testing process, we found that there is a serious leakage of the lac operator. Parameters like temperature, concentration of IPTG, and induction time had minimal effects on the protein amount. A promoter that is more strictly regulated should be applied if we want to put our product into real use.

What's more, the induction condition described in the original literature was not satisfying for our conditions. Since we learned that the lac operator was leaking, we suspected that the high expression rate caused the formation of the inclusion body, preventing the protein from transporting into the outer membrane. After we decrease the induction temperature, we obtain satisfying results.

During our Cryo-electron microscopy experiment, we conducted our investigation with the help of the instrument-assisted manager. After knowing the goal of our experiment, she suggested that Cryo-EM might not be suitable for us. Since Cryo-EM couldn't observe the internal structure of bacteria, we were unable to confirm if the antibody entered the bacteria. She suggested that we try the ultrathin section, which can display the whole structure of Bacteria clearly.

From this cycle, the most important learning was the choice of control group. When similar results were found in the lptE and Lpp'OmpA_lptE. We immediately communicated with our advisors Jincheng Wang and Jiaze Liu. Both Wang and Liu suggested that pretreatment like fixation might affect the polarity of the outer membrane, thereby causing unsatisfying control. To solve this problem, Liu recommended we use lipoprotein(sp removed) as a control, which will be expressed only in the cytosol. Wang advised us to conduct a blocking procedure in the experiment.

Cycle 2: Small peptide secretion and cleavage verification

● Design
In order to verify that the small peptide is displayed and can be cleaved by enterokinase, we designed two sets of plasmids. One set includes AQ with different protein tags(AQ_GST and AQ_6xhis), and the other set QEP(QEP_GST and QEP_6xhis). Take QEP_GST as an example, small peptide QEP is fused with GST tag at the C terminal and enterokinase site at the N terminal. The fusion is linked with the C terminal of lpp’OmpA, activated by the T7 promoter.

Fig. 5 Genetic circuit of AQ and QEP sets.

● Build
For the AQ set, Lpp'OmpA-enterokinase site-AQ-protein tag was synthesized by Genscript, and subcloning was performed to get the final products.

Fig. 6 Map of AQ set

For the QEP set, Lpp'OmpA-enterokinase site-QEP-protein tag was synthesized by Genscript, and subcloning was performed to get the final products.

Fig. 7 Map of QEP set

● Test
Plasmids were transformed into E. coli BL21(DE3) separately. Sequencing verified transformation success.

Fig. 8 Sequencing of AQ&QEP sets transformant.

Similar to cycle 1, Western Blot tests were conducted to verify the protein expression in engineered bacteria. Immunofluorescence assay(IFA), FACS analysis, and Cryo-electron microscopy verified the success of surface display. However, different from cycle 1, the enzymatic cleavage experiment verified the availability of the whole system. To see more details, visit the Results page.

● Learn
Bacteria will lyse either by prolonged induction at low temperature or by extended digestion in the nutrient-deprived buffer.

The results of the enterokinase digestion experiment (37℃-1h is more positive) is not contradict to results of FACS.

In FACS analysis, the instrument only detects intact cells, and the results show a higher positive signal in bacteria induced at 16°C for 12 hours. This suggests that the chimeras on the outer membrane are more accessible to antibodies under these conditions. However, this does not necessarily mean that the chimeras presentation is more pronounced at lower temperatures. Factors such as antibody accessibility and potential damage to the bacteria during the incubation process could influence the results.

In the enterokinase digestion experiment, bacteria induced at 16°C for 12 hours gradually lyse in the medium, releasing Lpp'OmpA-FLAG-protein-GST and some degraded protein-GST. This released protein may be involved in the digestion, leading to a stronger signal compared to bacteria induced at 37°C for 1 hour induction.

All Lpp'OmpA-FLAG signal is missing in the experiments. The reason may be that the FLAG tag is in an inherently undetectable structural domain after digestion or the antibody affinity is not good enough to generate detectable signal. To further investigate the protein we can conduct TEM experiment to detect Lpp'OmpA-FLAG which is anchored on outer membrane after digestion.

Cycle 3. Promoter Changing

● Design
Based on the Learn in cycle 1, we knew that the lac operator was leaking. Therefore, to make our induction more controllable, we decided to replace the T7 promoter and the lac operator with the PBAD promoter, which is induced by L-arabinose. We choose QEP_GST and AQ_GST as our targets and design pBAD_QEP_GST and pBAD_AQ_GST. The genetic circuits are similar to the QEP_GST and AQ_GST, except that the PBAD substitutes the T7 and the lac operator.

Fig. 9 Genetic circuit of pBAD_AQ/QEP_GST

● Build
We conducted PCR on QEP_GST and AQ_GST to get insertion fragment. The subcloning was performed to get the final result.

Fig. 10 Maps of pBAD_QEP_GST and pBAD_AQ_GST

● Test
Plasmids were transformed into E. coli BL21(DE3) separately. Sequencing verifies transformation success.

Fig. 11 Sequencing of pBAD_QEP_GST and pBAD_AQ_GST transformant.

Due to the time limitation, we were unable to test further.

Part 2: Adhesion

In the Adhesion module, we aim to construct a cell-cell adhesion system that enables our engineered bacteria to colonize the intestine with adhesion proteins LAP and HSP60, and the surface display toolbox Neae. To evaluate the adhesion effects of LAP and HSP60 and prove the display box can work, we conducted 3 cycles.

Cycle 1

● Design
To assess the adhesion effects of LAP and HSP60, we individually inserted the corresponding genes into the pET-28a vector backbone. Additionally, we fused the FLAG tag and the 6xHis tag with HSP60 and LAP, respectively. For Neae, the FLAG tag was incorporated into the display region of the outer membrane. All of the coding sequences have gone through codon optimization to improve the expression of the proteins. See more in Parts page.

Fig. 12 The genetic circuits for Adhesion Constrct.

● Build
The plasmids were synthesized by Genscript. We transferred these plasmids into E. coli DH5α and BL21 (DE3) respectively and tested the presence of specific genes through PCR.

            Fig. 13 Overview of plasmids used in the adhesion construct group
               A. Map of pET-28a-[ LAP-His]
               B. Map of pET-28a-[ HSP60-Flag]
               C. Map of pET-28a-[Neae-Flag]

● Test
After strictly following the protocol, we sequentially transformed the plasmids into DH5α and BL21 (DE3) and confirmed this by sequencing afterward.

Fig. 14 The confirmation of transformation respectively.

Fig. 15 The sequencing results of transformation in E. coli BL21(DE3).
            A: The sequencing result of pET-28a-[HSP60-FLAG]
           B: The sequencing result of pET-28a-[LAP-His]
            C: The sequencing result of pET-28a-[Neae-FLAG]

Next, we conducted expression tests on three proteins: Neae, LAP, and HSP60. First, we used IPTG-induced BL21(DE3) cells lysate to directly prepare samples for Western Blot, confirming the expression of three proteins. For the surface display system Neae, we extracted membrane proteins and confirmed the expression of Neae on the membrane using Western Blot. In addition, we used immunofluorescence further to observe its localization on the cell membrane surface. For LAP and HSP60 proteins, we extracted membrane and cytosolic proteins, lysed the bacteria with WB&IP lysate and ultrasonication, and obtained proteins in the bacterial supernatant to perform Co-IP to verify the interaction between the two proteins. For LAP, we also performed nickel column purification and Coomassie Brilliant Blue staining to verify the expression. However, although both proteins are expressed, their concentration in the bacterial lysis supernatant is extremely low, which hinders the verification of protein interactions through Co-IP. See more in Results page.

● Learn

Despite attempting various induction conditions, the protein content in the bacterial supernatant remains low. This might be attributed to the low solubility of the two proteins or the excessively fast expression rate of the T7 promoter that led to the formation of a large number of inclusion bodies. In such a circumstance, it becomes challenging to obtain sufficient soluble proteins for interaction verification. Furthermore, there is a certain degree of leakage expression associated with the T7 promoter.

Cycle 2

● Design
After a conversation with Professor Fei Gan, we obtained advice on changing protein tags to enhance the solubility of LAP and HSP60. SUMO(Small Ubiquitin-like Modifier) is an N-terminus tag that has been proven to be able to enhance protein solubility and improve protein expression in prokaryote^[1]. In addition, the GST tag is a good choice for optimizing the expression of soluble proteins. In light of the situation that the specificity of FLAG was better than the 6xHis tag, we decided to add a SUMO tag at the N-terminus of HSP60 and to replace the 6xHis tag with a GST tag at the C-terminus of LAP. Moreover, we would like to change the promoter T7 to araBAD to settle the issue of inclusion body formation and expression leakage.

Fig. 16 The genetic circuit that after modification

● Build

To add or change protein tags, we designed PCR primers to clone SUMO fragment from pYSW-2(pala-SUMO) (gifted by Gan Lab), and GST fragment from pBAD AQ_GST (See our Parts page). To change promoters, we would clone genes of LAP and HSP60 and insert them into the backbone of pBAD AQ_GST plasmid (pBAD). See all of the primers in the Experiment.

We conducted reverse PCR to linearize the vectors and PCR to amplify the fragments. After linearization, we would ligate the PCR products to construct the plasmids by homologous recombination.

                  Fig. 17 Plasmid construction
            A.Map of LAP-GST (pET backbone)
            B.Map of SUMO-HSP60 (pET backbone)
            C.Map of pBAD-HSP60-Flag
            D.Map of pBAD-LAP (Tag changed, LAP-GST)

                    Fig. 18 The target fragments.
            A.SUMO; Length: 325bp.
            B.GST; Length: 689bp.
            C.HSP60; Length: 1816bp.
            D.LAP; Length: 2654bp.

● Test
The fragments SUMO and HSP60-pET-backbone were successfully amplified through PCR and then joined to form plasmid SUMO-HSP60 through homologous recombination. Ultimately, we successfully transferred the plasmid into E. coli DH5α and BL21 (DE3) (See more in Results page). Moreover, we successfully amplified LAP and HSP60 segments for the backbone of pBAD, yet with the change of GST tag and further optimization (mentioned below) they couldn't be employed anymore.

Fig. 19 The agarose gel electrophoresis of LAP and HSP60 segments PCR products. LAP: 2654bp; HSP60: 1816bp.

● Learn
We found that the map of pBAD AQ_GST that we use to design primers was wrong, for it was the unoptimized version with an original GST sequence instead of the codon-optimized one, which brought about the differences in the specific base pairs. Under this circumstance, we had to redesign the homologous arm of the primers thus leading to the changes of the segments.
Moreover, to obtain more HSP60 proteins, we inserted the SUMO tag into the pBAD-HSP60-Flag plasmid, which also changed the specific segment of HSP60.

● Redesign
We corrected the mistakes in the primers and changed the strategy of inserting HSP60 into the pBAD backbone——we used SUMO-HSP60 instead.

Fig. 20 The new target segments.
A. GST segment. The codon optimization is employed and only the specific bases change. Length: 689bp.
B. LAP segment. Changes as the GST segment changes. Length: 2652bp.
C. HSP60 segment. Changes with the insertion of the SUMO tag to pBAD-HSP60-Flag plasmids; Length: 1760bp.
D. SUMO segment. Changes with the insertion of the SUMO tag to pBAD-HSP60-Flag plasmids; Length: 2654bp.

Fig. 21 The improvement of the pBAD-HSP60-Flag gene circuit

Fig. 22 The new pBAD-SUMO-HSP60-Flag plasmid.

● Build
We conducted reverse PCR to linearize the vectors and PCR to amplify fragments. After which we would conduct homologous recombination to construct plasmids.

● Test
The target sequences HSP60(for pBAD), LAP(for pBAD), GST(for pET), and pBAD backbone(for SUMO-HSP60) were successfully amplified. See more in Results page.

The progress of our experiment is as follows after the primers change:

Table. 1 The amplification and purification processes for the target segments.

● Learn
Next, we will continue to construct the plasmids and try to solve the problems mentioned above, finally verifying the interactions between LAP and HSP60. We may employ agarose gel recovery to acquire purified SUMO-HSP60 pBAD backbone or modify the primers for better specificity to obtain a single band. Moreover, for segments with no PCR success, we may improve the annealing program or change the primers.

Cycle 3

● Design
Finally, our completed adhesion systems Neae-LAP and Neae-HSP60 were constructed to verify the function of the adhesion module.

Fig. 23 The completed adhesion construction, including Neae-LAP and Neae-HSP60.

Fig. 24 Plasmid maps of pET-28a-[Neae-HSP60] and pET-28a-[Neae-LAP]

        Fig. 25 Linearized fragment for pET-28a-[Neae-HSP60] plasmid
        A. Segment of EcoRI-Neae backbone-HindIII. Length: 7087bp.
        B. Segment of HindIII-HSP60-EcoRI. Length: 1955bp

        Fig. 26 Linearized fragment for pET-28a-[Neae-LAP] plasmid
            A. Segment of Neae backbone. Length: 7259bp.
            B. Segment of LAP. Length: 2649bp

● Build
By employing PCR, DNA purification, restriction enzyme digestion, gel extraction and T4 ligase ligation, the fragment and the vector were linearized and then ligated to obtain the target plasmids.

● Test
Agarose gel electrophoresis confirmed the successful construction and transformation of the Neae-HSP60 plasmid. Further WB analysis demonstrated the successful expression of the protein. See more in Results page.
Interestingly, during the plasmid transformation, we found a variation in efficiency when inputting different backbone-to-insert fragment ratios. A ratio of 1 to 3 backbone to insert fragment showed better transformation. Furthermore, the colonies on the resistant plates that from a 1 to 3 ratio also grew larger than the other group.

Fig. 27 The colony PCR from 1:3 and 1:10 transformation ratio. The success segments are marked by the grey arrow.

For the Neae-LAP plasmid, we obtain the target band and successfully transfer the combined plasmid into E. coli BL21(DE3).See more in Results page.

Fig. 28 The output for DNA ligation. The reactant are marked in the red box. Segment LAP is marked in the red box below while the segment Neae backbone is marked in the upper box. Neae backbone: 7259bp; LAP: 2649bp

● Learn
For Neae-HSP60, A ratio of 1:3 which corresponds to the specification of ligase contributes to a better transformation effect and growth. Moreover, although we used restriction enzymes that were overdue for two years, they can still function properly. The restriction enzymes from two different companies also can implement double enzyme digestion if the digestion time is the same.
For Neae-LAP, the primers examined by websites or software may provide better specifications and obtain a single band. A gel extract may provide a more accurate fragment concentration when implemented DNA restriction enzyme digestion and T4 ligase.

Part 3: Biosafety

Cycle 1

● Design
In order to verify how strong the quorum sensing promotors P_lux、P_las are and to explore a proper concentration to induce them, we designed two plasmids: Plux-deGFP and PLas-deGFP, using the plasmid pET28a as the backbone.
In Plux-deGFP plasmid, full-length deGFP(green fluorescence protein) is fused with a P_lux. Moreover, we use T7 promotor to express transcription factor LuxR. Then, promotor P_lux is activated by LuxR which is binding with another external quorum sensing N-acyl homoserine lactones (AHLs) Lux 3OC₆ Once expressed, the protein deGFP can be excited to green fluorescence and thus be detected by Fluorescence microplate reader.
The PLas-deGFP plasmid nearly works the same as PLux-deGFP plasmid, except that the promotor P_las is activated by the LasR coexisting with Las AHL.

Fig. 29 genetic circuit of PLux-deGFP and PLas-deGFP

● Build

For PLux-deGFP, LuxR-P_Lux-deGFP was synthesized by Genscript and subcloning was performed to get the final product.

Fig. 30 Map of Plux-deGFP

For PLas-deGFP, LasR-P_Las-deGFP was synthesized by Genscript and subcloning was performed to get the final product.

Fig. 31 Map of PLas-deGFP

● Test
Plasmids were transformed into E. coli BL21(DE3) separately. Sequencing verifies transformation success.

Fig. 32 Sequencing of PLux-deGFP and PLas-deGFP transformants

Fluorescence microplate reader tests were conducted to verify the protein expression in engineered bacteria. The fluorescence intensity of induced promoter expression was measured in a bivariate between the concentration of the AHLs solution and the induction time. To see more details, visit the Results page.

● Learn
During our testing process, we found that there was a gradient of fluorescence intensity with the gradient difference in induced concentration of the AHLs and time, and the gradient was in line with our expectations.
What's more, some lptE groups showed a high level of leaking expression of deGFP, which is embodied in the fluorescence intensity. We suspect that in E. coli, there exist other AHLs to affect the activation of P_lux and P_las.
From this cycle, we had a better understanding of the Lux system and the Las system, which will guide us in the following cycle.

Cycle 2: CcdA&CcdB verification

● Design
After verifying the functionality of the QS system, we further tested the toxicity of CcdB and the detoxification ability of CcdA. In this cycle, we chose Pet-28a as our backbone and designed one plasmid: CcdB_CcdA. In the plasmid, CcdB is expressed by the T7 promoter, while CcdA is under the control of araBAD promoter, which is triggered by L-arabinose. All transcription processes are terminated by double terminators.

Fig. 33 The genetic circuit of CcdB_CcdA

● Build
In plasmid CcdB_CcdA, the designed fragment was synthesized by Genscript, and subcloning was performed to obtain the final product.

Fig. 34 The map of CcdB_CcdA

● Test
During the test process of this cycle, we encountered various problems. The initial problem was transformation. The fact that CcdB was toxic to E. coli and the lac operator was leaking made it difficult to transform the plasmid into E. coli BL21(DE3). After a month's try, we eventually selected another strain E. coli Tuner(DE3)pLysS, and successfully transformed the plasmid into the strain.

Fig. 35 Sequencing success for CcdB_CcdA transformant.

After the transformation, we conducted 2 sets of experiments: CcdB toxicity verification and CcdA detoxicaiton verification. To see more details, visit the Results page.

● Learn
Firstly, to express a type of protein that is toxic to the host, strain selection must be carefully considered. We chose E. coli Tuner(DE3)pLysS instead of E. coli BL21(DE3) to obtain the transformants. The protein expression level of Tuner series strains can be precisely regulated by IPTG concentration, with strict concentration dependence. The pLysS plasmid can produce T7 lysozyme, which can effectively reduce the basal expression level of the target gene.

Secondly, the IPTG gradient didn't strictly echo with the OD₆₀₀ gradient. But there was a turning point between 10³μM and 10²μM. From 10³μM to 10²μM, bacteria underwent a growth inhibition state to a normal growth state. In contrast, the result of the L-arabinose gradient didn't show this phenomenon. The L-arabinose gradient was in line with the OD₆₀₀ gradient well, except the final OD₆₀₀ in 10g/L L-arabinose was lower than 1g/L. We hypothesized that 10g/L was too high, adding extra metabolic stress to the bacteria and leading to cell death.

Based on this cycle, we better understood the toxicity and detoxication effects of CcdB and CcdA, which will guide us in the following cycles.

Cycle 3

● Design
After verifying the functionality of basic parts (LuxR/LuxI system, LasR/LasI system, CcdA&CcdB), we design an exogenous verification experiment. In this cycle, we combined the QS and CcdA/B toxin-antitoxin systems. We designed two plasmids: Exo_LuxR and Exo_LasR, using pBAD_hisA as the backbone.

Fig. 36 The genetic circuit of Exo_LuxR

In Exo_LuxR, CcdB is activated by araBAD promoter, which is induced by L-arabinose. CcdA is controlled by promoter P_lux. Promoter J23106 constitutively expresses LuxR. All gene transcription processes are terminated by double terminators(BBa_B0015). In this circuit, when we add arabinose into the medium, the transformed bacteria will be killed by the expression of CcdB. When we add 3OC₆, CcdA will be expressed and prevent the toxic effect of CcdB.

In Exo_LasR, LasR is expressed by araBAD promoter, while CcdB is expressed by P_las. All the processes of transcription are terminated by double terminators(BBa_B0015). When we add 3OC12HSL and L-arabinose, the engineered bacteria will be killed by the expression of CcdB.

Fig. 37 The genetic circuit of Exo_LasR

● Build
For both Exo_LuxR and Exo_LasR, the designed fragments were synthesized by Genscript, and subcloning was performed to obtain the final product.

Fig. 38 Map of Exo_LuxR and Exo_LasR

● Test
The test of this cycle was not as expected.

For Exo_LuxR, we successfully transformed the plasmid into E. coli Tuner(DE3)pLysS. The sequencing result was as follows.

Fig. 39 Sequencing success of Exo_LuxR transformants

However, we tried several methods to verify the functionality of the plasmid, all gaining unsatisfying results. Initially, we tried to measure OD₆₀₀ changes over time. We tried the L-arabinose gradient and 3OC₆HSL gradient, but all showed no obvious difference in the growth of bacteria.

Fig. 40 OD₆₀₀changes overtime in L-arabinose gradient and 3OC₆HSL gradient

We also tried the CFU method and the trypan blue staining method. In the CFU method, we found no obvious difference between the situation of induction and without induction. In the trypan blue staining method, we found almost all bacteria that appeared in the field of view were alive. For the time limitation, we didn't conduct testing for the Exo_LasR plasmid.

● Learn
Aiming at figuring out the problem, we discussed it with our advisors Jincheng Wang and Zixuan Ding.

We initially talked about the design of the plasmids. Since all parts were verified in the previous cycle, the sequence wasn't the source of problem. The only difference was that this time, CcdB was expressed under the control of araBAD promoter instead of the T7 promoter. We applied this change due to the finding of leakage of the T7 promoter in the previous cycle and cycle 1 in the Secretion subgroup. We assumed that the strength of araBAD promoter was too weak to express a sufficient amount of CcdB. In the following experiment, we will try to change the promoter.

What's more, since CcdB is toxic to bacteria, it might have random mutations in the sequence during the propagation. We then conducted PCR, using bacteria incubated for 2 weeks as templates. The sequencing result showed that the sequence of CcdB was complete.

Fig. 41 Sequencing result for Exo_LuxR, using bacteria incubated 2 weeks as templates

Cycle 4

● Design
While we designed the exogenous QS verification, we also designed the endogenous verification experiment. In this cycle, We designed two plasmids: LuxR_LasI and LasR_LuxI, using pBAD_hisA as the backbone.

The overall genetic circuits of these two plasmids are similar to exogenous systems. But this time, the quorum sensing substances are synthesized by bacteria. In LuxR_LasI, LasI is expressed, which will synthesize 3OC12HSL. In LasR_LuxI, LuxI is expressed and synthesize 3OC₆HSL. Therefore, bacteria transformed with these plasmids separately will secrete quorum sensing substances and activate the genetic circuit of each other.

Fig. 42 the genetic circuit of LuxR_LasI and LasR_LuxI

● Build
For both LuxR_LasI and LasR_LuxI, the designed fragments were synthesized by Genscript and subcloning was performed to obtain the final product.

Fig. 43 The map of LuxR_LasI and LasR_LuxI

● Test
Due to the time limitation, we were unable to test the endogenous system.

References

[1]

Panavas, T., Sanders, C., Butt, T.R. (2009). SUMO Fusion Technology for Enhanced Protein Production in Prokaryotic and Eukaryotic Expression Systems. In: Ulrich, H.D. (eds) SUMO Protocols. METHODS IN MOLECULAR BIOLOGY™, vol 497. Humana Press, Totowa, NJ. https://doi.org/10.1007/978-1-59745-566-4_20