Difference between revisions of "Part:BBa K5246043"

Revision as of 19:40, 29 September 2024

Caulobacter crescentus CB2/CB2A HfsE-HfsJ-HfsG-HfsH-HfsK-HfsL Polysaccharide tetrad assembly

Introduction

Vilnius-Lithuania iGEM 2024 project Synhesion aspires to create biodegradable and environmentally friendly adhesives. We were inspired by bacteria, which naturally produce adhesives made from polysaccharides. Two bacteria from aquatic environments - C. crescentus and H. baltica - harness 12 protein synthesis pathways to produce sugars, anchoring them to the surfaces. We aimed to transfer the polysaccharide synthesis pathway to industrially used E. coli bacteria to produce adhesives. Our team concomitantly focused on creating a novel E. coli strain for more efficient production of adhesives.

This is the complete holdfast tetrad assembly system. Parts of this composite can be found:BBa_K5246044 and BBa_K5246045.

This part was used in Vilnius-Lithuania iGEM 2024 project "Synhesion" https://2024.igem.wiki/vilnius-lithuania/.

Biology and Usage

Biology

Caulobacter crescentus is a common freshwater gram-negative oligotrophic bacterium of the clade Caulobacterales. Its distinguishing feature is its dual lifestyle. Initially, C. crescentus daughter cells are in a “swarmer” cell phase, which has a flagellum, enabling them to perform chemotaxis. After the motile phase, they differentiate into “stalked” cells. This phase features a tubular stalk with an adhesive structure called a holdfast, allowing them to adhere to surfaces and perform cell division. [1][2]

Caulobacterales synthesize a polysaccharide-based adhesin known as holdfast at one of their cell poles, enabling tight attachment to external surfaces. It is established that holdfast consists of repeating identical units composed of multiple monomers. Current literature agrees that in Caulobacter crescentus, these units form tetrads composed of glucose, an unidentified monosaccharide (either N-mannosamine uronic acid or xylose), N-acetylglucosamine, and N-glucosamine. These units are polymerized and exported to the outer membrane of the cell, where they function as anchors, securing the bacterium to a surface[3][4]. The C. crescentus holdfast is produced via a polysaccharide synthesis and export pathway similar to the group I capsular polysaccharide synthesis Wzy/Wzx-dependent pathway in Escherichia coli. The holdfast synthesis (hfs) genes include those encoding predicted glycosyltransferases, carbohydrate modification factors, and components of a wzy-type polysaccharide assembly pathway. [4][5][6] The synthesis of holdfast polysaccharides (Fig.1) occurs through a mechanism analogous to the Wzx/Wzy-dependent group I capsular polysaccharide biosynthesis pathway observed in Escherichia coli. The process is initiated in the cytoplasm by the glycosyltransferase (1) HfsE, which transfers an activated glucose-phosphate from UDP to an undecaprenyl-phosphate (Und-P) lipid carrier (1) [7]. Subsequent monosaccharide residues are added to the lipid carrier to form a repeating unit by the action of three glycosyltransferases: (2) HfsJ (adding N-mannosamine uronic acid or D-xylose), (3) hfsG (adds N-acetylglucosamine) and (4) HfsL (most likely adding another N-acetylglucosamine) [8]. Then some of the N-acetyl-D-glucosamine within these repeat units undergoes enzymatic modification through the activity of the deacetylases (5) HfsH and HfsK, which “incorporates” into the tetrad of another saccharide - D-glucosamine [9]. The completed repeat of four monomers is then flipped over the inner membrane to the periplasm by flippase HfsF (6) [8]. In the periplasm, the repeat unit is transferred to copolymerases HfsC and HfsI (7), which assemble holdfast into a mature polysaccharide [10]. Subsequently, following the polymerization, holdfast saccharides are exported through a multi-protein export channel made of HfsB, HfsA, and HfsD (8-10) [11]. After excretion, holdfast polymer is relocated to the anchoring Hfa group of proteins (11), where they function by holding the mature polysaccharide on the cell's surface of, e.g. C. crescentus or H. baltica , and securing it to the surface [8].

Usage

Genes from this composite part are responsible for tetrasaccharide assembly in the holdfast biosynthesis pathway. HfsE (BBa_K5246005 ) transfers an activated glucose-phosphate from UDP to an undecaprenyl-phosphate (Und-P) lipid carrier, HfsJ (BBa_K5246010 ) adds N-mannosamine uronic acid. then HfsG (BBa_K5246007 ) and HfsL (BBa_K5246012 ) joins two N-acetyl-D-glucosamines (HfsG transfers to mannosaminuronic acid, HfsL to another N-acetyl-D-glucosamine) then last N-acetyl-D-glucosamine is deacetylated by HfsH and HfsK (BBa_K5246008 ) (BBa_K5246011 )

In the end, we have a glycolipid - a lipid carrier undecaprenyl phosphate (UndP) with an oligosaccharide (glucose--mannosaminuronic acid--N-acetyl-D-glucosamine--D-glucosamine) attached to it.

Then, the tetrasaccharide can be polymerized by polysaccharide polymerization and export apparatus (BBa_K5246046 ) to get a full polysaccharide if used together

Sequence and Features

Experimental characterization

Part cloning

All of the proteins composing this system are responsible for tetrad assembly. Since the system's proteins are found in the cytoplasm, we concluded that using a high-copy plasmid would ensure sufficient protein concentration for tetrasaccharide repeat synthesis and activation of the whole pathway. This would provide adequate substrate supply for polysaccharide polymerization and export.

To assemble specifically this part into BBa_K5246043 to then further assemble the holdfast synthesis pathway in E. coli , we had to assemble this part first into a backbone of pRSF-Duet-1. We designed a strategy to maximize the success of plasmid assembly by first assembling plasmids with 3 genes and, after verifying the sequences, integrating 3 left genes into that backbone (Fig. 1). In this way, we prevented Golden Gate assembly errors by trying to construct plasmids from 8 or more fragments.

The assembly was done using Golden Gate assembly with IIS AarI restriction enzyme sites introduced during PCR amplification. The backbone of pRSF-Duet-1 (Novagen) and fragments were amplified using Phusion Plus DNA polymerase, as the genome of C. crescentus has a high GC% content making the appearance of non-specific products during PCR amplification more common and primer design more challenging (Fig. 2). Since hfsA gene had an AarI RE site directly in the gene, this site was domesticated during side directed mutagenesis.

Due to the high amount of non-specific products, the fragments were gel-purified. Vectors and fragments composing this operon, were mixed in equimolar amounts with GG reaction components and incubated as described in protocol. The reaction was later transformed into E. coli Mach1 (Thermo Scientific) competent cells. The assembly was then confirmed with colony PCR (Fig.3) and restriction digest analysis (Fig. 3) and positive colonies were sequenced.

To assemble specifically this part into BBa_K5246043 to then further assemble the holdfast synthesis pathway in E. coli , we had to assemble this part first into a backbone of pRSF-Duet-1. We designed a strategy to maximize the success of plasmid assembly by first assembling plasmids with 3 genes and, after verifying the sequences, integrating 3 left genes into that backbone (Fig. 1). In this way, we prevented Golden Gate assembly errors by trying to construct plasmids from 8 or more fragments. In this particular case, we are inserting BBa K5246042 (CB2/CB2A hfsL-hfsH-hfsK) operon into a backbone with BBa_K5246041(CB2/CB2A hfsE-hfsJ-hfsG)

Due to the high amount of non-specific products, the fragments for the second round were also gel-purified. Vectors and fragments composing this operon, were mixed in equimolar amounts with GG reaction components and incubated as described in protocol. The reaction was later transformed into E. coli Mach1 (Thermo Scientific) competent cells. The assembly was then confirmed with colony PCR (Fig.5) and restriction digest analysis (Fig. 4) and positive colonies were sequenced.

-->Full functional project operon assembly with this part and full operon charazterization can be found in composite part BBa_K5246043.

Holdfast polymerization and export apparatus usage for holdfast synthesis

Holdfast synthesis system expression optimization

The intricate holdfast synthesis pathway involves numerous proteins that must be efficiently co-expressed in Escherichia coli. After obtaining plasmids used for full holdfast synthesis pathway assembly, we had to optimize the expression of the whole system in E. coli. Previously, only three studies have tried to recombinantly express C. crescentus proteins in E. coli for unassociated studies with our project's goal [12][13][14]. Since the E. coli strains and protein expression conditions were unrelated to each other, and before our project, no one in iGEM besides the 2009 iGEM ULB-Brussels team ever tried expressing more than two C. crescentus proteins in E. coli at the same time, we had no solid foundation for expression and chose to experiment with different E. coli strains and conditions. Therefore, it was essential to optimize the conditions for simultaneous protein expression by trying different media, temperatures, IPTG concentrations, and expression times on multiple E. coli strains. We used SDS-PAGE analysis of cell lysates and HPLC-MC proteomics to verify the expression results.

Results overview: Optimal C.crescentus protein expression was achieved in the BL21(DE3) strain cultivated in the LB medium. The most favorable conditions included an incubation temperature of 37°C, induction with 0.5 mM IPTG, and an expression duration of 3 hours at 37°C.

KRX(DE3)

To determine the best conditions for the whole system expression, we first used E. coli KRX(DE3) strain. We tried expressing separate plasmids pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL (BBa_K5246043) and pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI (BBa_K5246046) with different IPTG concentrations - 0.1, 0.25, 0.5, 0.75 and 1 mM - and 0.1% rhamnose with protein expression for 3h at 37°C after induction. As we saw, some bands, corresponding to our protein sizes, were appearing in pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI (BBa_K5246046) expression (Fig. 7.1), but we were not sure if they were our system proteins, therefore for pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL (BBa_K5246043) we used negative control with empty pRSF vector and expressed the proteins in similar conditions. We saw that pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL (BBa_K5246043) operon proteins were also expressed (Fig. 7.2) with minimal IPTG concentration impact on protein amount.

Fig. 7.1. SDS-PAGE analysis of pACYC-hfsA-hfsB-hfsD-hfsF-hfsC-hfsI (BBa_K5246046) expression in KRX(DE3) at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26610 (Thermo Scientific).

Fig. 7.2. SDS-PAGE analysis of pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL (BBa_K5246043) expression in KRX(DE3) at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

We decided to use the lower IPTG concentrations - 0.25, 0.5, and 0.75 mM - for gene expression induction of the full system, as it is more cost-effective for upscale in the future. But, unfortunately, full system expression at different temperatures and expression times did not provide clear bands of proteins in SDS-PAGE gel analysis (Fig. 8. 1-5).

Fig. 8.1. SDS-PAGE analysis of CB2 system expression in KRX(DE3) at different IPTG concentrations for 3h at 37°C. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Fig. 8.2. SDS-PAGE analysis of CB2 system expression in KRX(DE3) at different IPTG concentrations for 3h at 30°C. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Fig. 8.3. SDS-PAGE analysis of CB2 system expression in KRX(DE3) at different IPTG concentrations overnight at 30°C. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Fig. 8.4. SDS-PAGE analysis of CB2 system expression in KRX(DE3) at different IPTG concentrations for 3h at 22°C. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Only some proteins, in the size range of 30-50 kDa, appeared, but in general, the results of expression of the whole system were inconclusive, leading to the need to test another E. coli strain.

BL21(DE3)

The next E. coli strain we tested was BL21(DE3). Since the IPTG concentration appeared not to make that big of an impact on the expression, we settled on IPTG concentrations of - 0.25, 0.5, and 0.75 mM - in this way covering a wide range of them and accelerating the optimization effort, if the system would be expressed. We also decided to yet again test different expression temperatures - 37°C, 30°C, 16°C - before and after gene expression induction.

Initially, we tested pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI (BBa_K5246046) operon expression at 37°C for 3h, which did not give promising results (Fig. 9.1), as we could not see distinguishable differences before and after induction. Nevertheless, we proceeded with pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL (BBa_K5246043) operon expression at the same conditions, which appeared to be working (Fig. 9.2), as we could see stark differences between empty E. coli and our operon lysates. As with the expression in the KRX strain, we could not see many differences between the IPTG concentrations.

Fig. 9.1. SDS-PAGE analysis of pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL expression in BL21(DE3) at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Fig. 9.2. SDS-PAGE analysis of pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI expression in BL21(DE3) at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26610 (Thermo Scientific). Note: negative control cultures for 0.25 mM and 0.75 mM were contaminated and subsequently not used for SDS-PAGE analysis.

We advanced with the whole system expression. After analyzing expression at 37°C for 3h conditions, we saw that there were pronounced differences between the uninduced system and the system after 3 hours (Fig. 10.1). As with KRX(DE3) expression, we saw that IPTG concentration used for gene expression induction did not make a big impact for overall expression.

Since the CB2 system was expressing, we tried different temperatures to optimize protein expression further. Remarkably, decreasing the expression temperature to 30°C and expression overnight did not make a significant difference as the proteins were still expressed in similar amounts to that of 37°C (Fig. 10.2). Expression of 16°C overnight produced some of the expected bands but not in the same capacity as expression at higher temperatures (Fig.10.3).

Fig. 10.1. SDS-PAGE analysis of CB2 system expression in BL21(DE3) at different IPTG concentrations for 3h at 37°C. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Fig. 10.2. SDS-PAGE analysis of CB2 system expression in BL21(DE3) at different IPTG concentrations overnight at 30°C. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Proteomic analysis of samples induced by 0.5 mM IPTG from separate parts and the whole system, revealed that proteins responsible for holdfast polymerization - hfsC and hfsI - were not expressed (Fig. 11 (a),(b)). In addition, protein levels during full system expression dropped notably compared to separate part expression. However, proteins were still expressed in slightly higher quantities than in control (Fig. 11. (c)).

Nevertheless, as later experiments showed, these proteins were probably substituted by paralogous proteins found in E. coli as the system without 2 parts was still producing a polysaccharide (see BBa_K5246003 and BBa_K5246009 ). We reason that in the future, we should first test whether the separate proteins - hfsC and hfsI - are expressed and at what conditions before assembling new plasmids with different operon orders or additional promoters. T7/lac could serve as a good starting point, other considerations could involve separately inducible or constitutive promoters available in iGEM Parts Registry.

C41(DE3)

Once the system was successfully expressed in BL21(DE3) strain, we proceeded to optimize the expression further by testing another E. coli strain - C41(DE3). We decided to test separate system parts and the whole CB2 system with different IPTG concentrations - 0.25, 0.5, and 0.75 mM. Since we saw that the proteins were best expressed at 37°C in KRX(DE3) and BL21(DE3) strains, we settled on testing only this temperature. SDS-PAGE analysis of cell lysates before and after gene expression induction revealed that proteins were expressed in separate parts of the system and the whole system (Fig. 12. 1-3). Regrettably, the quantity was visibly less than that seen in BL21(DE3) strain indicating that this strain is not suitable for efficient system expression.

Fig. 12.1. SDS-PAGE analysis of pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI expression in C41(DE3) at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26610 (Thermo Scientific)

Fig. 12.2. SDS-PAGE analysis of pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL expression in C41(DE3) at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Rosetta(DE3) pLysS

Even though to this point we found E. coli strain with, in our perspective, sufficient system expression, we were determined to improve the expression even more. A study was conducted, where C. crescentus hfsJ gene was expressed in E. coli to determine its interaction with hfiA (inhibitor of holdfast development), we investigated the strain they used for expression - Rosetta (DE3) pLysS - for our own purpose [4]. Since we decided to use this strain, we had to carefully reconsidering our CB2 systems’ design, as the pRARE plasmid, native to Rosetta, contains the same antibiotic resistance - chloramphenicol - and the same origin of replication - p15A - as one of our system’s plasmids - pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI (BBa_K5246046). For this we decided to change the backbone’s antibiotic resistance and origin of replication (ori) (Fig.13).

As a donor of new antibiotic resistance and origin of replication we choose pBAD-PhoCl2f plasmid, which was kindly gifted to us by VU LSC Institute of Biotechnology. This particular plasmid contains the ampicillin resistance gene and ColE1 origin of replication compatible with our pRSF operon and pRARE plasmid.

To save time we decided to utilize Golden Gate assembly for resistance/ori switching. We successfully acquired plasmids with new resistance/ori, which, after testing them with colony PCR, restriction analysis (Fig. 14), were sequenced by whole plasmid sequencing by SeqVision. Since after transformation into electrocompetent Rosetta cells they grew without any problems on LB agar plates with 3 different antibiotics, we came to a conclusion that the resistance/ori switch worked.

To test the expression we used different IPTG concentrations - 0.25, 0.5 and 0.75 mM - and our standard expression temperature of 37°C with checking the total expressed protein amount after 3 hours by SDS-PAGE analysis. Unfortunately, we saw very low or almost no protein expression in separate plasmids, leading to the same happening when we tried expressing the whole system (Fig. 15. 1-3). We reason that this might be due to increased metabolic strain on E. coli due to a whole additional plasmid introduced into the system during expression. This E. coli strain might be suitable for specific C. crescentus protein expression but, sadly, is not fitting for our needs.

Fig. 15.1. SDS-PAGE analysis of pRSF-(1)HfsE-hsfJ-hfsG-(2)hfsH-hfsK-hfsL expression in Rosetta(DE3) pLysS at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26614 (Thermo Scientific).

Fig. 15.2. SDS-PAGE analysis of pACYC-(3)hfsA-hfsB-hfsD-(4)hfsF-hfsC-hfsI expression in Rosetta(DE3) pLysS at different IPTG concentrations at 37°C for 3h. M - molecular weight ladder in kDa, Pageruler Unstained Protein Ladder, 26610 (Thermo Scientific).

HMS174(DE3)

12. Liu, Q., Hao, L., Chen, Y., Liu, Z., Xing, W., Zhang, C., Fu, W., & Xu, D. (2022). The screening and expression of polysaccharide deacetylase from Caulobacter crescentus and its function analysis. Biotechnology and Applied Biochemistry, 70(2), 688–696. https://doi.org/10.1002/bab.2390
13. Fiebig, A., Herrou, J., Fumeaux, C., Radhakrishnan, S. K., Viollier, P. H., & Crosson, S. (2014). A Cell Cycle and Nutritional Checkpoint Controlling Bacterial Surface Adhesion. PLoS Genetics, 10(1), e1004101. ,https://doi.org/10.1371/journal.pgen.1004101
14. Patel, K. B., Toh, E., Fernandez, X. B., Hanuszkiewicz, A., Hardy, G. G., Brun, Y. V., Bernards, M. A., & Valvano, M. A. (2012). Functional Characterization of UDP-Glucose:Undecaprenyl-Phosphate Glucose-1-Phosphate Transferases of Escherichia coli and Caulobacter crescentus. Journal of Bacteriology, 194(10), 2646–2657. https://doi.org/10.1128/jb.06052-11

Holdfast biosynthesis

Holdfast synthesis requires glucose

Polysaccharides are produced only in the part of the population

Polysaccharides can not be purified using chemical purification

E. coli with holdfast synthesis pathway produce biofilm-like structures

Different E. coli strains form holdfast polysaccharide

Suitable substrate search for holdfast production

Holdfast composition investigation

Holdfast Protein Expression and Production of Polysaccharides in HMS174(DE3)ΔwecA

Increasing holdfast production efficiency

Investigating the composition of the holdfast polysaccharide

UDP-N-acetyl mannosaminuronic acid (UDP-ManNAc) is a sugar used to make polysaccharides in the CB2 system

FTIR analysis of the holdfast material

References

1. Hendrickson, H., & Lawrence, J. G. (2000). Mutational bias suggests that replication termination occurs near the dif site, not at Ter sites. FEMS Microbiology Reviews, 24(2), 177–183. https://doi.org/10.1111/j.1574-6976.2000.tb00539.x
2. Andrews, S. C., Robinson, A. K., & Rodríguez-Quiñones, F. (2004). Bacterial iron homeostasis. Journal of Bacteriology, 186(5), 1438–1447. https://doi.org/10.1128/jb.186.5.1438-1447.2004
3.Rabah, A., & Hanchi, S. (2023). Experimental and modeling study of the rheological and thermophysical properties of molybdenum disulfide-based nanofluids. Journal of Molecular Liquids, 384, 123335. https://doi.org/10.1016/j.molliq.2023.123335
4. Boutte, C. C., & Crosson, S. (2009). Bacterial lifestyle shapes stringent response activation. Journal of Bacteriology, 191(9), 2904-2912. https://doi.org/10.1128/jb.01003-08
5. Mackie, J., Liu, Y. C., & DiBartolo, G. (2019). The C-terminal region of the Caulobacter crescentus CtrA protein inhibits stalk synthesis during the G1-to-S transition. mBio, 10(2), e02273-18. https://doi.org/10.1128/mbio.02273-18
6.Thanbichler, M., & Shapiro, L. (2003). MipZ, a spatial regulator coordinating chromosome segregation with cell division in Caulobacter. Journal of Bacteriology, 185(4), 1432-1442. https://doi.org/10.1128/jb.185.4.1432-1442.2003
7. Toh, E., Kurtz, H. D., & Brun, Y. V. (2008b). Characterization of the Caulobacter crescentus Holdfast Polysaccharide Biosynthesis Pathway Reveals Significant Redundancy in the Initiating Glycosyltransferase and Polymerase Steps. Journal of Bacteriology, 190(21), 7219–7231. https://doi.org/10.1128/jb.01003-08
8.Chepkwony, N. K., Hardy, G. G., & Brun, Y. V. (2022). HfaE Is a Component of the Holdfast Anchor Complex That Tethers the Holdfast Adhesin to the Cell Envelope. Journal of Bacteriology. https://doi.org/10.1128/jb.00273-22
9. Hershey, D. M., Fiebig, A., & Crosson, S. (2019). A Genome-Wide Analysis of Adhesion inCaulobacter crescentusIdentifies New Regulatory and Biosynthetic Components for Holdfast Assembly. mBio, 10(1). https://doi.org/10.1128/mbio.02273-18
10. Toh, E., Kurtz, H. D., & Brun, Y. V. (2008c). Characterization of the Caulobacter crescentus Holdfast Polysaccharide Biosynthesis Pathway Reveals Significant Redundancy in the Initiating Glycosyltransferase and Polymerase Steps. Journal of Bacteriology, 190(21), 7219–7231. https://doi.org/10.1128/jb.01003-08
11. Javens, J., Wan, Z., Hardy, G. G., & Brun, Y. V. (2013). Bypassing the need for subcellular localization of a polysaccharide export-anchor complex by overexpressing its protein subunits. Molecular Microbiology, 89(2), 350–371. https://doi.org/10.1111/mmi.12281