Cyanobacteria

Our project aims to develop a modular surface display system for Synechococcus elongatus UTEX 2973 to enable expression of proteins, such as carbonic anhydrases, for biocementation applications on Mars. By combining cyanobacterial expertise, rational construct design, and iterative testing, we engineered and validated a recombination-based vector, pRepv3, optimized for stable protein display and counterselection in UTEX. This work establishes a foundation for future cyanobacterial surface engineering and carbon capture strategies in space biotechnology.

Introduction

With CO2 in such abundance in the Martian atmosphere, photosynthetic microbes like cyanobacteria are the perfect chassis to engineer for biocementation on Mars. Because they use CO2 as their carbon source, they can not only be sustained on the red planet, but also be engineered to make other useful bioproducts for the benefit of human settlement.

Having been advised by the Hallam Lab’s in-house cyanobacteria experts, Nannaphat (Patrik) Sukkasam and Kalen Dofher, to choose model organisms over extremophiles for engineering, we landed on Synechococcus elongatus UTEX 2973 (hereafter referred to as UTEX). This strain is already an attractive chassis for carbon-negative bioproduction as one of the fastest-growing cyanobacteria strains [1], and has been successfully engineered by multiple iGEM teams.

Many synthetic biology tools have been developed for cyanobacteria, such as those on the iGEM registry like the Marburg Collection or CyanoGate [2]; however, little published work has been done to effectively display heterologous proteins on the surface of cyanobacteria, particularly in Synechococcus. We find this to be a good balance of having past works to rely on for guidance, and sufficient novelty for our team to make a meaningful contribution to the iGEM community. To fulfill this, we created a surface display vector for S. elongatus, which we used to display a variety of carbonic anhydrases for biocementation applications.

Some works by teams and researchers have tackled cyanobacterial surface display systems in a related species, Synechocystis sp. PCC 6803. In addition to not being directly applicable to UTEX, we also saw room for improvement based on what we know from the Caulobacter S-layer display system design. Cengic et al. tested surface display on membrane proteins[3] , including the S-layer protein (Slp) of Synechocystis, sll1951[4], as well as the pilin proteins that protrude past the S-layer. Among their designs, they fused their heterologous protein, an affibody, to the N terminus of their Slp, and by flow cytometry, saw a 10-fold increase in mean fluorescence intensity (MFI) relative to the wild-type control, suggesting an effective display system. However, wheniGEM Keystone 2023 used this system to display Helicobacter pylori CA (HpCA), they did not see any CA activity using the culture medium.

Construct Design

Through literature review, consultation with our advisors who had experience with surface display, and in silico cloning, we designed our surface display vector for UTEX, pRepv3 (BBa_2570RGH4) Rep is short for replacement, in that it replaces the wild-type Slp at its locus with an Slp fusion protein of our design via homologous recombination. The ‘empty’ state of the vector has an sfGFP reporter fused to the C domain of the Slp, such that it may secrete the reporter if transformed. To make it modular, the vector uses SapI Golden Gate Assembly (albeit with custom fusion sites) to insert a heterologous protein to display. The vector also includes an inducible counterselection system under the control of a theophylline riboswitch and lac operator to select for double crossover recombination where desired. Lastly, the basis of mobility region from pBR322 is included, in case the user wants to transfect via triparental mating using pRL443 and pRL623 plasmids.

Slp-CA fusion protein

Early in the project, we did not know S-layer display on UTEX was possible until we found a recent publication that achieved that very thing[5]. In it, the author identified the S-layer protein to be VCBS (UniProt ID Q31NQ2). Additionally, their characterization work designated the last 150 residues consisting of multiple repeat-in-toxin units to be the secretion signal peptide (hereafter the C domain), and the first 207 residues the section anchoring the Slp to the cell membrane (hereafter the N domain). This aligns with our understanding of Slp’s, using Caulobacter’s RsaA as a model, where the N and C domains are thought to serve a similar purpose, and a ‘secretion’ system could be made by omitting the N domain of RsaA in the fusion protein CDS.

Usai demonstrated that a variety of proteins, including carbonic anhydrases, could be displayed using only the N and C domains of VCBS. This means for the fusion protein, only the necessary signal peptide for secretion and anchor to remain bound to the cell are used[5], unlike our CB2A display system, which simply inserts the heterologous protein in the middle of the otherwise wild-type RsaA.

We sought to recreate this fusion protein expression system ourselves in a way that maximizes modularity and commonality with our CB2A display system. Starting with the N and C domains of VCBS, we added flexible GGGGS linkers on either side to minimize steric hindrance, and a Myc tag upstream of the C domain for surface display validation. Then, inside, we included an sfGFP reporter system from the pJUMP plasmids, flanked by SapI affixes. The 5’ and 3’ SapI fusion sites are TCG and GGC respectively, and are designed for seamless insertion of a CDS without a stop codon to ensure the entire fusion protein is in frame. BsaI was not used because we initially used BsaI to assemble vector variations from modular parts. We also considered BsmBI/Esp3I, but found that it would have conflicted with homology arms for neutral site I of UTEX. As such, any heterologous protein to be inserted should not have a stop codon and have the aforementioned fusion sites flanking the CDS. It was a deliberate choice to omit the terminator on the reporter so that the “empty” state of the fusion protein, i.e. no CDS inserted, the reporter could possibly be secreted. We placed short flexible linkers flanking the insert and Myc tag, which we can use to verify surface expression after S-layer extraction.

We chose regulatory parts such as promoters, terminators, etc. based on what is known to work in Synechococcus and cyanobacteria, such as aforementioned Marburg Collection, CyanoGate[2], and other past iGEM projects.

Theoretically, omitting the N domain in the fusion protein sequence would allow the fusion protein to be secreted into the medium, since the N domain is thought to anchor the protein to the membrane. Either display or secretion would however be limited to single-unit proteins, and while not tested, we speculate that it would have similar limitations to the CB2A display system since they both use the type I secretion system[5], [6]. Per our advisor Beth Davenport, rapidly-maturing or folding proteins also struggle to get secreted in the CB2A system, so we expect a similar limitation with UTEX.

Cycle 1: Display Fusion Protein Design and Validation

Homologous recombination

Patrik and Kalen, the Hallam Lab’s resident Cyano experts, advised that we engineer UTEX via homologous recombination, as is the preferred method for engineering cyanobacteria[7]. It should be noted that cyanobacteria shuttle vectors do exist, such as pPMQAK1 (BBa_J153000) or Marburg 2019’s Golden Gate-compatible pANS shuttle vectors (BBa_K3228089, BBa_K3228069), based on the pANS plasmid native to Synechococcus[8]. However, from a rational design perspective, direct genome integration via homologous recombination is highly advantageous for expression stability, since shuttle vectors have a possibility of being cured without adequate selection pressure. All we would need to do is transfect a “repair template” that is our insert flanked by homology regions or arms targeting the desired locus.

Next came the question of what length of flanking homology arms to use. According to literature, and in Patrik and Kalen’s experience, homology regions as short as 100-250 bp are sufficient for recombination [9]. However, evidence also shows that recombination efficiency increases with homology arm length[9],[10]. We settled on targeting 800 bp as our arm length as a balance between recombination efficiency and construct size, as this is the typical length used by other suicide vector constructs, such as in the commercially available pSyn_6 vector from ThermoFisher, or the homology arms in Marburg’s collection targeting neutral sites in the S. elongatus genome.

We obtained the sequences for the homology arms flanking the WTVCBS CDS from its genome. For selection, we add a Kanamycin (Kan) resistance cassette downstream of the C domain to be integrated along with the fusion protein CDS. Diverging from the CB2A system, we chose not to use Chloramphenicol (Cm) as the marker because a preprint at the time had reported it being less effective for UTEX [11], and because we suspected our Cm stock was contaminated. Since the resistance marker is being added downstream of the Slp, we realized the VCBS N domain would be conserved, meaning a shorter segment of the genome upstream of VCBS was required to reach the 800 bp target size. This also allowed us to avoid a SapI restriction site on the genome upstream.

Notably, we do not have a promoter for the CDS; since we directly target replacing the wild-type gene, we can use the wild-type promoter to control the expression of the fusion protein.

Counterselection

More specifically, what is referred to as homologous recombination is actually a double crossover recombination event[12] . That is, the first single crossover event takes place at one of the homology regions, and there exists an intermediate state where the entirety of the template is part of the genome, before a second crossover event takes place to fully integrate the insert and remove the backbone. The second event does not always take place, so you can be left with the entire plasmid integrated in some colonies. In many cases this may not present an issue; however, we were concerned with the WT Slp still being expressed in such a case, so we added an inducible counterselection system outside of the homology region to specifically select for double crossover events. In theory, with antibiotic and inducer, only transformants having undergone double crossover recombination may survive, because the resistance marker should be integrated along with the fusion protein, but the counterselection marker would have been discarded.

The sacB counterselective marker, while commonly used for various organisms, appears ineffective for UTEX[11]. Instead, we opted to use the native sepT2 toxin gene (BBa_K4880011) found on the pANL plasmid, like the ccdB gene in E. coli. Its use has been demonstrated before as part of a biocontainment strategy[13], as well as an effective counterselective marker when placed under the control of the theophylline riboswitch E* and Ptrc promoter (theoE*, BBa_K3930029) [11]. The theophylline-IPTG inducible expression system has been used extensively for cyanobacteria with demonstrated robustness [14],[15], particularly for counterselection [16]. In theory, this would enable markerless genome editing as well if the resistance marker were placed outside the homology arms; however, for convenience, we decided to integrate the marker as well.

These design elements culminated in our pRepv3 vector. More on why it is “v3” later.

Alternative vectors

As an alternative to targeting the VCBS locus in the genome with pRepv3, we targeted inserting the fusion protein construct at neutral site I of UTEX, since it is a proven engineering strategy. This required minor modifications to the pRep design, namely replacing the VCBS homology arms with those targeting neutral site I from the Marburg Collection, and adding a promoter and RBS to regulate expression. We paired the Anderson promoters BBa_J23107, BBa_J23119, and cyanobacterial promoter Pcpc560, designated low, medium, and high strength respectivey in Usai, 2025[5], with the BBa_B0034 RBS. These vectors were assembled but were ultimately not used.

We also wanted to express the fusion protein using a self-replicating shuttle vector as an alternative to homologous recombination, and to better mirror the CB2A construct. To achieve this, we the pANS_Kanres backbone designed by Marburg 2019.

Cloning in E. coli

As is standard molecular cloning, we would assemble and produce our constructs using E. coli before electroporation into UTEX. The v1 revision of pRep was initially going to be assembled into the pDest vector, which has AmpR. Upon a desire to include the sepT2 cassette and the realization that only one resistance marker was required[8], we decided to make a custom “backbone” fragment with the pUC ori and sepT2 cassette. pRepv1 was only made in silico; pRepv2 was the first that was cloned.

pRepv2 was successfully made using Golden Gate Assembly and transformed in E. coli DH5a, as evident by green fluoresence from the reporter, albeit after 3 days. However, we initially struggled with getting a high miniprep yield, which made the second assembly step to insert CA difficult and gel electrophoresis runs often inconclusive. At one point, we began getting miniprep yields of >1000 ng/uL, having attributed the increase to using a larger culture volume and employing ethanol precipitation during miniprep[17], but upon sequencing, we found a large deletion that inactivated the sepT2 cassette completely. The deletion also shrank the upstream homology region down to roughly 500 bp; while we expected reduced recombination efficiency and will not have the ability to counterselect, these “faulty” pRepv2-CA constructs could still be used to engineer UTEX.

Notably, pRepv2 did not have a lacI gene, which led us to hypothesize that the theoE* riboswitch alone is leaky and insufficient for suppressing expression. In the v3 revision, we added the lacI promoter and gene and placed the clpPX terminator via Gibson Assembly, demonstrated to be effective in both E. coli and UTEX[18],[2]. The transformed colonies not only recovered faster (1.5 days) but also generally looked healthier, having the green fluoresence signal on the entire colony instead of only in the center as was with the v2 colonies. We still had to use a larger culture volume for miniprep (5 mL), but the concentration was plenty high enough for downstream uses. Restriction digest via Esp3I and SapI shows the vector is the correct size and functional for SapI assembly, respectively.

PCR primer pairs were designed to bind to the CA in the forward direction and Myc tag downstream of CA in the reverse, such that amplification only takes place with insertion in the correct orientation. Restriction digest with EcoRI and PCR on the miniprep products shows the constructs are the correct size and contain the correct insert. With this, we proceeded to electroporate into UTEX. Unfortunately, at the time of writing, we were only able to obtain pRepv3-HpCA, and had some setbacks with cloning due to our homemade competent cells losing a lot of their efficiency in the freezer.

Cycle 2: Refining Plasmid Design for Cloning

Cultivating UTEX

While S. elongatus UTEX 2973 is known for its 2-hour doubling time in literature[1], we could not achieve this high-growth phenotype in our growth conditions. While the lab had a dedicated incubator for cyanobacteria, we were limited partially by the 30°C incubation temperature due to it being shared. In addition to higher temperature (>37°C) and light intensity, sources that achieved the fast growth used an air feed supplemented to be 3-5% CO2[1],[19]. Given our incubator does not enrich the air with CO2, we hypothesized that growth was being substrate limited. We attempted to overcome this by supplementing BG-11 with additional bicarbonate or carbonate, but unfortunately did not see improved growth rate, but saw some differences in biomass accumulation and appearance. Generally, we observed with UTEX grown in BG-11 supplemented with carbonate was able to grow to a higher OD750 than regular BG-11. UTEX grown in bicarbonate-supplemented BG-11 appears a brighter, more blue-ish green in the first 2-3 days but often loses viability and turns yellow from day 4 onwards.

Electroporation

It is well-known that unlike its close relative S. elongatus PCC 7942, UTEX lacks the capability for natural transformation, so the preferred method for engineering this strain is triparental mating [1], [15], [16]. Kalen Dofher has experience transforming more difficult cyanobacteria, and reckoned we would be able to transform via this method. Where natural transformation is not possible, conjugation may be favoured over electroporation due to the latter having low efficiency and requiring large amounts of DNA in the µg range. However, this was an acceptable compromise for us, because electroporation is a lot simpler than triparental mating, and we would not need more than one miniprep worth of DNA should one transformation work.

Protocols specific to cyanobacteria and UTEX in particular were initially hard to find, but we eventually found one from Mühlenhoff and Chauvat, 1996[20], which was used by iGEM HK_SSC 2021 and Chen et al. 2013, albeit on a marine Synechococcus strain[21]. After some initial testing, we made minor modifications to the protocol, such as detailing centrifugation speeds and durations, performing the final resuspension in 10% glycerol, and using bicarbonate-supplemented BG-11 for the recovery medium over DTN medium in the original protocol.

In the initial tests, we plated cells on BG-11 plates with 10 µg/mL Kan, which was a concentration commonly used by other teams. However, we were unable to recover any colonies after more than 2 weeks, whereas electroporated cells plated on plain BG-11 plates recovered after 3-4 days. To rule out the electoporation step as a possible cause for lack of recovery, we retried electroporation and this time plated half the cells onto plain plates, and were able to see growth after 3 days.

Cycle 3: Adapting the electroporation protocol

We found another Synechococcus electroporation protocol from Matsunaga et al., 1990[22], [23], which used an antibiotic strength as low as 1 ug/mL Ampicillin. Based on this , we were able to recover lots of colonies after plating electroporated cells on 1 ug/mL BG-11/Kan plates after 3 days. However, the number of colonies recovered looks similar to that of cells electroporated with DNA plated onto plain plates. As such, we were less certain of successful transfection and electroporation, and so verification by colony PCR became critical.

To do so, we designed primers to bind just outside of the homology region on the genome. The wild-type genome will amplify a 3 kb band, but because we also integrate the KanR gene, successful double crossover recombination should amplify to 4 kb. Since single crossover recombination is also a possibility, we can also use two additional primer pairs that amplify from upstream or downstream of the region to the middle of the insert. Suppose only a single recombination event only took place at the upstream region; neither primer pair tageting outside the homology region nor the downstream region should amplify, at least to their correct size.

We eventually obtained the cyanobacterial conjugation plasmids pRL443 and pRL623 for triparental mating[24]. Due to time constraints, we were not able to pursue this in parallel with electroporation, and decided to forego this technique.

At the time of writing, we had electroporated UTEX containing pRepv3-HpCA, pRepv2-BtCAII, pRepv2-SazCA, pDisp-L-BtCAII, and pDisp-L-SazCA and successfully recovered them onto 1 ug/mL BG-11/Kan plates. With colony PCR, we were able to some colonies of pRepv3-HpCA, pDisp-L-BtCAII, and pDisp-L-SazCA, and both streaked them out onto new plates for maintenance and picked into liquid culture, and are still waiting to see green.