As with C. crescentus we tested different IPTG concentrations (0.25, 0.5 and 0.75 mM). Since the 37°C temperature appeared to be optimal we proceeded with this temperature as our starting point. For expressed protein analysis we used SDS-PAGE on cell lysates before and after induction.

Due to time constrains we decided to start with expressing the system first and then proceed with separate system parts expression (see parts BBa_K5246049 and BBa_K5246052) . After numerous attempts to grow cultures for protein system expression we were only able to grow one culture to appropriate optical density for gene expression induction. Since it took much more time than we expected to reach the required optical density, we decided to leave the culture overnight at 16°C and analyze the lysates from overnight expression.

Unfortunately, all of the cultures with the full H. baltica system either did not grow overnight or were of almost non existent optical density indicating that the culture died. Nonetheless, we prepared samples for SDS-PAGE gel analysis of before and after induction. Surprisingly, there were some bands corresponding to recombinant protein sizes appearing in the after induction lanes suggesting that parts of the system were expressing (Fig. 18).

Protein name	Size, kDa
hfsE	58
hfsA	55
hfsF	52
hfsC	50
hfsI	49
hfsK	41
hfsG	37
hfsL	36
hfsJ	29
hfsH	28
hfsD	27
hfsB	25

However, due to the unpredictable nature of the culture growth and weak expression of some proteins, we concluded that this was definitely not the right condition for this system’s expression in E. coli. Because of the natural high salinity environment of the system’s proteins, we theorize, it might be impossible to express the system in E. coli as it seems that the system is toxic to the cell. Although, it is plausible to try expressing this system in other host organisms used for expressing halophilic proteins [6][7].

Multiple experiments revealed that having the proteins alone is insufficient for holdfast synthesis. We attempted to purify the holdfast from the media (see Experiments), but the results were not as expected (Fig. 19). No holdfast was produced under the given conditions, despite the presence of all the proteins. No differences were observed between the control and target samples in the dot blot assay of the purification samples.

So we considered adding precursors for the biosynthesis pathway to facilitate its’ activation. We hypothesized that we should be able to adapt methods, suitable for polymer production in E. coli [9][10]. Holdfast synthesis begins from UDP-D-glucose transfer to an UndP lipid carrier, so having an excess of available glucose could solve the problem. We designed a protocol based on previous studies[11][12][13]. Adding 1% (w/v) glucose after target protein expression, as the addition of glucose lowers expression levels by targeting lac operator, and incubating overnight at 30°C with a shaking speed lowered to 170 rpm resulted in distinct rings forming around the edges of the shaking media (Fig.20) [25][27].

In addition, for further experiments, e.g. holdfast purification, we expressed and synthesized ring-like formations, later deemed to be the holdfast, in larger flasks. This increase in size - from 500 mL to 1 L flask did not diminish the ability to form the rings (Fig. 21). Not only that but after multiple day incubation and consistently lowering the shaking speed, we were able to get multiple rings to form around the 1 L flask walls. This information was later used to produce rings for other experimental applications where larger amounts of starting material were required.

Wheat Germ Agglutinin (WGA) specifically binds to N-acetyl-D-glucosamine - a characteristic component of the holdfast, found in abundance in the polysaccharide [14]. This allowed us to determine the localization of the holdfast. We tested various samples, including cells from the media, the supernatant, material from the ring, sonicated cells, and the supernatant of the sonicated cells. After performing the dot blot assay (see Experiments), we saw that the holdfast is produced (indicated by considerable presence of N-acetyl-D-glucosamine) but is localized exclusively within the ring material (Fig. 22).

We made the conclusion that holdfast was located in the ring, but we still did not know the nature of the material, so we investigated if there were any living bacteria in it. Following the expression, after the rings formed, we resuspended the samples in LB media and streaked them on LB agar plates under four different conditions: (1) with appropriate antibiotics, (2) antibiotics + 0.5mM IPTG, (3) antibiotics + 1% glucose, and (4) antibiotics + 0.5mM IPTG + 1% glucose. The plates were then incubated overnight at 30°C. Variety of plates was supposed to show if there are any visible morphological differences between living bacterial cells from the ring, bacterial cells that express proteins of interest, and bacterial cells that not only express proteins but also, in theory, produce holdfast. 

It seems like ring formation consists of living bacteria that can grow on agar plates. Although they do not have any morphological differences (Fig. 23), they do appear to be more of a liquid-like appearance.

Following the discovery, we hypothesized that only part of the bacterial population is able to effectively express all 12 proteins, so we used plated bacteria from rings for standard protein expression and ran a SDS-PAGE analysis (Fig. 24). It is clearly noticeable that bacteria from the ring expresses proteins much better than the fresh ones.

Surprisingly, after incubating overnight, we did not notice any significant difference in the quantity of holdfast produced, based on ring appearance.

After understanding that the ring contains our target polysaccharides with bacteria, we further investigated the exact location of the holdfast. We expanded the dot blot assay method, including more samples from the ring: lysed ring cells, lysed ring cells’ supernatant, and lysed ring cells’ sediment. This lets us pinpoint the location of the polysaccharides and confirm where the holdfast is produced. Dot blot assays showed that holdfast is produced only in the ring cells mostly located in the soluble fraction of the lysate. These results were replicable for multiple biological repetitions (Fig.25.1-3).

The addition of glucose is required to activate the holdfast synthesis pathway. Even after the activation, only part of the bacterial population is capable of synthesizing polysaccharides, forming ring-like structures on surfaces. After the synthesis, most of the holdfast is attached to the cells.

After successful holdfast biosynthesis, we aimed to purify the polysaccharides for further research and analysis. We adapted the only available C. crescentus holdfast chemical purification method and verified the results using dot blot assays with WGA [8].

During the purification process, it was noticed that holdfast-producing cells stick together and can not be homogenized via pipetting or vortexing. Unfortunately, despite multiple attempts, we could not obtain pure holdfast material (Fig. 26.1-3.). None of the 3 attempts yielded significant differences between the CB2 system sample and the empty (control) sample. As the other experiments have showed, the E. coli with the CB2 system produces polysaccharides with distinct chemical features of holdfast attached to the cells, purification protocols specific for exopolysaccharide purification from the cell envelope could be tried upon the ring-formed cells [15][16][17].

Understanding that holdfast stays attached to the producing cells led us to the hypothesis that they ought to have some noticeable morphological differences. SEM analysis revealed that the polysaccharide-producing bacteria form tight conglomerates and differ morphologically compared to the control group. SEM pictures clearly show that holdfast-producing bacteria have altered cell wall morphology appearing bigger, and more “puffy” with uneven envelope topology .Moreover, the CB2 system containing bacteria seems to form large aggregates, sticking together into difficult-to-disrupt lumps. This characteristic was noticed beforehand in holdfast purification experiments, but only in SEM, it became clear that the underlying cause of the phenomenon is biofilm-like structures (Fig. 27. 1-3., Fig. 28. 1-3.) [18][19].

Similar results were seen in flow cell bright-field microscopy, where it is noticeable that the CB2 sample has a considerable amount of big bacterial aggregates (Fig.29). Biofilm-like clumps noticeable in the video can not be homogenized by pipetting or vortexing, suggesting strong associations between cells. This could explain why CB2 sample bacteria were not evenly coating the glass slide, as the aggregates prevented enough bacteria from covering the bottom of the flow cell.

SEM and bright field microscopy analysis revealed that the ring material of E. coli cells producing the holdfast are morphologically different and produce biofilm-like structures.

We decided to test E. coli strain testing, we used C41(DE3), Rosetta(DE3) pLysS, and HMS174(DE3), and repeated the expression conditions with different IPTG concentrations (0.25, 0.5, and 0.75 mM) with protein expression for 3h at 37°C. After the expression, we added 1% glucose and left the flasks to incubate overnight at 30°C, 170 rpm. All of the flasks, after incubation, had prominent rings in CB2 system flasks compared to the control (Fig. 30.1-3.). However, the formed rings were not as prominent as the ones formed in BL21(DE3).

This indicated that even though these strains were not producing visible amounts of CB2 system proteins, the low amount of the expressed system was still sufficient enough to promote ring formation. As we ran out of time, we did not perform dot blot assays of the formed rings, but it is likely that they would also contain holdfast with N-acetyl-D-glucosamine as they appear during the same conditions used for BL21(DE3) rings production.

We chose sugars that could be relevant to the polysaccharide composition: glucose, mannose, fructose, saccharose, N-acetyl-D-mannosamine, D-glucuronic acid, D-glucosamine, and xylose. After performing holdfast synthesis with standard conditions (in BL21(DE3)) optimized in earlier iterations, we observed the formation of rings with some substrates (Fig. 31.1-8). Notably, D-glucosamine substrate - a direct component of holdfast polysaccharide does not produce prominent rings, which indicates that it is incorporated into the holdfast as a N-acetyl-D-glucosamine that, after deacetylation with hfsH, only then becomes D-glucosamine (Fig. 31.7.) [1].

Then, we examined the amount of holdfast produced with each monosaccharide substrate by dot blot assay (Fig. 32). As with visual representation of glucose formed rings (Fig. 31.1.), it appears that glucose is still the best choice as a substrate for holdfast production, with mannose coming second and D-glucosamine - third.

Additionally, crystal violet staining (see Experiments) experiments yielded similar results (repeated in 3 biological replicates), with glucose and mannose samples producing the largest amount of “biofilm”. This data was used in the model construction to find and verify if glucose is indeed the most cost-effective and efficient holdfast production substrate (see Model).

After multiple attempts to eliminate the ECA pathway from E. coli expression strains KRX and BL21(DE3) by using CRISPR/Cas9 and λ Red system, we could not achieve it. We could transform cells with the plasmids and dsDNA templates required for efficient editing, but the few colonies we obtained showed false positive results.

The ECA pathway was eliminated by deleting the wecA gene in these E. coli expression strains: BL21(DE3), Rosetta(DE3), and HMS174(DE3). We achieved it by changing the genome editing strategy and choosing a homologous Red/ET recombination system. Our edited colonies had kanamycin-resistance cassettes as markers in the edited region, and we tested them by colony PCR. We were able to distinguish edited and non-edited colonies by data: PCR’ed non-edited region 1544 bp and with edited region 1840 bp. From the results it could be seen that gene deletion happened in most of the cells in three different E. coli expression strains (Fig.33.a-c). Some colonies obtained 2 bands because there was a mix of edited and dead cells on the plate.

CB2 holdfast system proteins are expressed, and polysaccharides were produced in HMS174(DE3)ΔwecA. We discovered that wecA gene deletion did not interfere with CB2 system protein production. Additionally, we compared it to the not edited HMS174(DE3) strain (Fig 34.).

We analyzed polysaccharide and ring production after overnight incubation with 1% glucose in an ECA-deficient HMS174(DE3)ΔwecA strain (Fig. 35). Although cells with the CB2 system grew slower than an empty control, they still produced rings with the polysaccharides, with thicker bands of the ring visible in figure 35.

We wanted to inactivate WecB - the enzyme that catalyzes the initial reaction in the production of the substrate UDP-N-acetyl mannosaminuronic acid (UDP-ManNAc), which is thought to be in the composition of our polysaccharide.

Goal: Knock out and test inactivation of WecB - enzyme producing one of the substrates for our polysaccharide biosynthesis pathway.

Strategy: Homology recombineering and usingtheRed/ET recombination system.

Results: The wecB gene was knocked out from E. coli expression strainsBL21(DE3), Rosetta(DE3), and HMS174(DE3).

The UDP-N-acetyl mannosaminuronic acid (UDP-ManNAc) substrate was withdrawn by deleting the wecB gene in these E. coli expression strains: BL21(DE3), Rosetta(DE3), and HMS174(DE3). We achieved it by using a homologous Red/ET recombination system. Our edited colonies had kanamycin-resistance cassettes as markers in the edited region, and we tested them by colony PCR. We were able to distinguish edited and non-edited colonies by data: PCR’ed non-edited region 1577 bp and with edited region 1846 bp. From the results it could be seen that gene deletion happened in most of the cells in three different E. coli expression strains(Fig.36 a-c). Some colonies obtained 2 bands because there was a mix of edited and dead cells on the plate.

CB2 holdfast system proteins are expressed but polysaccharides are not produced in HMS174(DE3)ΔwecB. We discovered that wecB gene deletion did not interfere with CB2 system protein production (Fig. 37).

We analyzed polysaccharide and ring production after overnight incubation with 1% glucose in the HMS174(DE3)ΔwecB strain. Cells with the CB2 system grew faster than unedited HMS174(DE3) and did not produce rings with polysaccharides. The flask with the E.coli containing the CB2 system looked the same as the E. coli without (Fig 38).

Additionally, we comparedHMS174(DE3)ΔwecA with HMS174(DE3)ΔwecB containing CB2 systemto see the difference of polysaccharide and ring production (Fig. 39).

These experiments were done by our colleagues at the Vilnius University Center of Physical Sciences And Technology. Dr. Martynas Talaikis and Dr. Ilja Ignatjev. FTIR analysis reveals chemical properties of our holdfast. 

A shows the ATR-FTIR spectra of intact cell samples between the control and CB2 system samples (Fig.40.1.A). Since they are very similar, a difference spectrum was constructed by subtracting the control spectrum from the sample spectrum (Fig. 40.1.B) to highlight the differences. In the difference spectrum, the vibrational modes directed upwards belong to the CB2 sample. Vibrational modes of the amide functional group appear at 1618 cm-1 (Amide-I, C=O stretching), 1517 cm-1 (Amide-II, N-H in-plane bending), and 1235 cm-1 (Amide-III, N-H in-plane bending coupled with C-N stretching, and C-C stretching) were related to the amide fragment of N-Acetyl glucosamine. Another important vibration mode observed at 1169 cm-1 belongs to the C-O stretching vibration mode in glycosidic linkage, COH stretching, and C-H in-plane bending. Also, glycosidic linkage was observed at 978 cm-1 (symmetric C-O stretching), 931 cm-1 (C‑O stretching), and 833 cm-1 (C-C stretching of α-glycosidic linkage). All assignments were done according to literature [20][21]. All vibrational modes individually and collectively indicate the presence of polysaccharides.

Figure 40.2 represents the control and sample spectra of cell debris with their difference spectra. The spectral assignment is similar to that presented in Figure 40.1. In both cases, the positive intensities in the difference spectra are related to sample-contained N-Acetyl glucosamine fragment and polysaccharides.

We chose BL21(DE3) strain for adjustable and efficient expression of target proteins since systems’ proteins were the best expressed in this strain. Given the lack of time, we went with conditions optimized beforehand in earlier experiments for the whole pathway expression: temperature of 37°C, induction with 0.5 mM IPTG concentration, and expression for 3 hours. 

After SDS-PAGE gel analysis, we concluded that we successfully expressed all 5 proteins from C. crescentus CB2 and CB2A strains (Fig. 41.1-2). We noticed, that HfsJ and HfsL glycosyltransferases are visible in lower quantities in comparison with the other proteins. Both of these protein expression conditions need to be further investigated and optimized.

Protein name	Size, kDa
hfsG	34
hfsH	27.9
hfsJ	34.7
hfsK	43.3
hfsL	33.3

Protein name	Size, kDa
hfsG	34
hfsH	27.9
hfsJ	34.7
hfsK	43.3
hfsL	33.3

Surprisingly, proteins of H. baltica wereexpressed in even higher quantities than C. crescentus in the same conditions (Fig. 51). Looking back at the holdfast synthesis system expressions we were unable to succeed in the expression optimization of the H. baltica pathway, however, proteins are clearly synthesized in the same conditions when expressed on their own. HfsJ protein is, similar to C. crescentus, not visible (Fig. 41-42).

Protein name	Size, kDa
hfsG	37
hfsJ	29
hfsK	41
hfsH	28
hfsL	36

After successful expression, we proceeded to work on the purification of his-tagged proteins. We went with immobilized metal ion affinity chromatography (IMAC). Adapting protocols from the little research that was available, we used HisPur Ni-NTA Spin Columns (Thermo Scientific) [25][5][4]. Equilibration, wash, and elution buffers contained 10mM Tris pH 7.4, 150mM NaCl, and 10mM, 75mM and 500mM imidazole, respectively. All proteins were successfully purified, unfortunately, in different quantities. HfsG, HfsH, and HfsK are purifiable in substantial quantities in the conditions used that the purification conditions are optimal (Fig. 43.1, 2, 4). HfsL is also clearly visible, although in lesser amounts (Fig. 43.5). HfsJ eluted earlier than expected at 75mM imidazole, suggesting that either the position of the 6xHis-Tag is on the wrong terminus or the tag is flexible, making weak interactions between Ni-NTA resin (Fig. 43.3).

Similar results were expected from the purification of C. crescentus CB2A proteins. HfsG, HfsH, and HfsK were purified in significant quantities (Fig. 44.1, 2, 4). Surprisingly, in this case, HfsJ was better purified and eluted in the same fraction as the other proteins — at 500 mM imidazole (Fig. 44.3). Another unexpected result was observed during HfsL protein purification (Fig. 44.5), where it eluted earlier than in the case of CB2, at 75 mM imidazole. This may be because proteins are dynamic structures that can fold in multiple ways. Our His-tagged end might have folded into the globular region of the protein, making it inaccessible for purification and causing weak interactions and premature elution.

H. baltica proteins were much trickier to purify. Only two of them were present after analyzing SDS-PAGE gel after purification: HfsH and HfsK, despite most of them being well-expressed (Fig. 45.1, 2). HfsK protein eluded at lower imidazole concentration - 75 mM. These results are likely to be explained by the natural environment of the host organism - Hirschia baltica. Its’ natural marine environment is high in salt concentration, thus proteins must be adapted to that specific ionic strength. Our purification buffers do not contain large amounts of salt and it might cause misfolding and aggregation of said proteins into insoluble inclusion bodies making them difficult to purify. More optimisation is required in order to purify these proteins, focusing more on salt composition of the buffers.

In hopes for further research, applications, and ideas for future iGEM teams, we showed that individual proteins of the holdfast synthesis system can be successfully expressed and purified. More testing on their enzymatic activity is needed, but we give a strong foundation for their future application for liposome labeling or any other implementation of choice.