Team:William and Mary/Design
Design
In order to engineer biofilms as living materials, we chose to focus on three key aspects: a robust and resilient biofilm, spatiotemporal control, and a programmable final product. For these requirements, we engineered and characterized a diverse library of adhesins, which we incorporated into both chemical and optogenetic circuits. For patterning and thus further control, we designed quorum sensing and Turing pattern systems.
Adhesin Library
Any biomaterial must be able to stand up to the wear and tear of environmental forces. We started out testing biofilm formation with the autotransporter AG43 but soon decided to incorporate a more diverse library of adhesins into our biomaterial toolkit.
AG43 is a well-characterized autotransporter native to E. coli about 3.1 kb in size. It is composed of an alpha region, which is the surface-displaying self-recognizing adhesin, and a beta barrel, which is embedded in the cell’s outer membrane and also transports the alpha domain to the extracellular environment (Klemm et al, 2004).
We used PCR with designer primers to isolate the gene off of the pDAWN plasmid and incorporate 3G Assembly compatible overhangs. With AG43 3G compatible, we were able to design and test its function with different induction methods (IPTG, Tet, Arabinose, and EL222/pBIND). We observed biofilm formation with the pDAWN + AG43 system after induction with blue light [0, 0, 255] for 16 hours and crystal violet staining. We were able to design patterns with extreme precision, both with plate masking and with a simple projected shape. However, there was a low amount of background expression of AG43.
A major drawback of pDAWN + AG43 was the strength of the biofilm formed. After allowing 24 hours of induction, we observed flaking of the biofilm when washed gently with PBS buffer. We realized it would not be feasible to construct biofilms with such poor integrity and began to construct our adhesin library.
Of the adhesins we considered, a particular class stood out to us: amyloid nanofibers. Unlike human amyloid-related diseases, such as Parkinson’s, Alzhimer’s, and prion related diseases, bacteria produce amyloid nanofibers for more productive ends. Various bacteria produce these fibers as a means of strengthening the extracellular matrix and adhering to surfaces, overall increasing the robustness and strength of biofilms. Amyloids are made of multiple subunits which self-assemble and are extremely resistant to denaturation and degradation. In addition to these general characteristics, individual species have adapted these nanofibers for a variety of functions from degrading toxins to transporting electrons (Taglialegna et al., 2016). We have developed circuits for curli fibers, the fap operon, and SaSuhB.
The fap operon is responsible for the expression of functional amyloids in Pseudomonas species. Similarly to curli fibers in E. coli, the fap operon has been shown to increase aggregation and biofilm formation in transgenic E. coli and its native Pseudomonas species (Dueholm et al., 2010, Dueholm et al., 2013). While less characterized, fap does have some advantages over its E. coli analogue: It is orthogonal in most common BSL-1 chassis, can be isolated as a single operon, and is clinically significant to Pseudomonas aeruginosa. The operon itself is composed of six genes (fapA-F) which encompass all levels of functionality from translocation and secretion to subunit nucleation and can be placed under the regulation of a single promoter and RBS (Dueholm et al., 2010).
The fap operon was isolated from the DSM 50071 strain of Pseudomonas aeruginosa in a BSL-2 laboratory via colony PCR [protocol on EHN benchling 190726] (Dueholm et al., 2013). The primers also added W&M PAD 1 and 2 sequences and 3G compatible sticky ends C and D. Half of the PCR product was then PCR purified while the other half was gel extracted. The PCR products had better concentrations and were used to Gibson the fap operon into the plasmid 1C3 with W&M PAD 1 and 2 annealing regions, followed immediately after by transformation into NEB 5-a cells. A diagnostic colony PCR was performed on the resulting transformation, and the colonies displaying the correct bands were inoculated overnight and miniprepped the following morning. At this point, it was necessary to perform site-directed mutagenesis (SDM) to remove a BsaI cutsite from the fap operon as this would have prevented successful 3G Assembly. Gibson SDM and the NEB protocol for KDL enzyme mix SDM [protocols are in the “to review” folder] were performed. Following successful mutagenesis of the BsaI cutsite by the Gibson method, the fap operon was inoculated, miniprepped, and diluted for 3G Assembly. It was not until too late in the competition that we became aware of a SapI cutsite also in the fap operon. We were not able to mutate the part in time and as a result any part containing BBa_K3059636 is neither BioBrick nor Type IIS assembly compatible. However, as our lab works mostly with 3G Assembly, this did not cause any difficulties. The part containing the fap operon, BBa_K3059637, was used to construct three circuits: BBa_K305925 strongly expressed the fap operon with promoter J23100.
Curli fibers constitute the main proteinaceous component of natural E. coli biofilms, making them a crucial addition to our engineered adhesin library. Curli fiber production is regulated by the curli operon, a divergent operon with parts csgBAC and csgDEFG. Each csg component controls a different aspect of curli production—though csgA forms the main subunit of the final fiber, csgG exports the fiber outside of the cell and is thus also necessary to fiber production. csgD on the csgDEFG operon interacts with and thus activates the csgBAC operon, making it unnecessary in a synthetic system that combines both operons (Barnhart & Chapman, 2006).
Our synthetic curli operon is modelled after the operon engineered by Harvard bioengineer Dr. Neel S. Joshi, who has utilized curli fibers for applications ranging from biosynthesis and bioremediation. Joshi combined both native operons to create csgBACEFG, a synthetic curli operon beneath the control of a single promoter. Following Joshi’s example, we isolated csgBAC (including the 22 basepair region upstream of it) and csgEFG (excluding any RBS or upstream region) from the E. coli genome via PCR. Sequences for the csg subunits were found in his paper A Synthetic Circuit for Mercury Bioremediation Using Self-Assembling Functional Amyloids and verified via NCBI Blast (2017). These sequences allowed us to choose annealing sequences for our primers. Our primers also included overhangs with BsaI cutsites and 4 basepair sticky ends to make the two curli operon components 3G-Assembly compatible.
Before we combined the components using 3G Assembly, we used DNA synthesis to create an RBS to precede csgEFG. The sequence for this RBS was again found in A Synthetic Circuit for Mercury Bioremediation Using Self-Assembling Functional Amyloids (2017). It is listed as the sequence between csgC and csgE in csgBACEFG.
The sticky ends flanking csgBAC, the pre-csgE RBS, and csgEFG ensure that they come together in the correct order during Golden Gate Assembly (the first step of 3G Assembly). These sticky ends also ensure that the synthetic curli operon comes together with the chosen promoter, terminator, and universal nucleotide sequences (UNS adapters). The modularity of Golden Gate allowed us to swap out promoters to create various circuits, including constitutive, chemically-inducible, and optogenetic circuits.
For our chemically-inducible system, we placed csgBACEFG beneath the control of a pLac promoter, which is repressed by constitutive expression of the LacI repressor. Addition of IPTG inhibits LacI, alleviating the repression of pLac and thus inducing curli fiber expression.
For our optogenetic circuits, we designed a cI-repressed curli operon (using the cI-repressed promoter R0015) to interface with the pDawn system. If successful, the same cI that represses AG43 in pDawn-AG43 could also repress curli fibers (unless, of course, this repression is relieved by blue light).
In theory, a plasmid with cI-repressed curli fibers could be cotransformed with the pDawn-AG43 plasmid and thus interface with it. Alternatively, we could deconstruct the pDawn system and assemble our own plasmid with pDawn and curli (or pDawn and curli and AG43). To accomplish this, we isolated the pFixK2 and AG43 from pDawn-AG43 via PCR. The primers used in PCR added BsaI cutsites and sticky ends required for 3G DNA assembly. To acquire 3G-compatible YF1-FixJ, we utilized Twist DNA Synthesis and removed an illegal BsaI cutsite that prevented 3G Assembly. Though the final pDawn designs were never successfully constructed from the 3G (and type IIS-compatible) parts, individual components were sequence-confirmed and characterized. For example, AG43 has been implemented into various chemically-induced circuits this year.
While assembling circuits with 3G-compatible pDawn components, we began pursuing an alternative optogenetic circuit utilizing the blue light-activated pBlind promoter. Sequences for pBlind and its repressor EL222 were sourced from Blue light-mediated transcriptional activation and repression of gene expression in bacteria (Jayaraman et al., 2016). We obtained these sequences in 3G-compatible format via DNA synthesis. Constructed pBlind circuits proved incredibly leaky, however; pBlind-AG43 circuits resulted in large aggregates of cells even in the absence of any light. pBlind-curli experiments were inconclusive; for some samples, the circuit acted as we hoped our constitutive curli circuit (J23107-curli) would.
SaSuhB is a protein native to Staphylococcus aureus. While it is characterized as an inositol monophosphatase enzyme, it is also necessary for unimpaired pia-independent biofilm formation in S. aureus (Boles et al., 2010). Moreover, further study of this protein has revealed that it is capable of forming macroscopic amyloid fibers when expressed in E. coli (Dutta et al., 2016). The fibers were characterized as being incredibly sticky, causing the researchers great trouble in trying to wash the fibers off of glass and plasticware. Additionally, SEM imagery revealed that bacteria were readily adhered to these fibers (Dutta et al., 2016). This combination of traits makes SaSuhB appear perfect for our application of forming strong, robust biofilms.
Using the sequence of SaSuhB, as provided by (Dutta et. al, 2016), SaSuhB was ordered as a geneblock through IDT. Furthermore, the primer sequences used by (Dutta et. al. 2016) were modified to include the William and Mary Pad sequences, and the BsaI cut sites and sticky ends needed for gibson, and Type IIs assembly respectively. These primers were also ordered through IDT. Once the primers arrived, we ran a PCR on S. aureus genomic DNA in a certified BSL-2 lab on campus, and obtained the coding sequence for SaSuhB. This coding sequence, along with the sequence obtained through synthesis, were used in Type IIs assembly of several circuits: pLac controlled SaSuhB (033), pL - LacO controlled SaSuhB (040), and pBlind_V1 SaSuhB + pBlind_V1 AG43 (039). Circuits 33 and 40 are repressed by LacI, which is de-repressed upon addition of IPTG. Circuit 39 expresses SaSuhB under the control of the blue light activatable promoter pBlind_v1.
Design Overview
As we engineered our adhesion library, we realized that biofilm potential would always be limited when working with E. coli. We decided that it was necessary to investigate a natively biofilm-forming strain of bacteria. Mycobacterium is a genus of bacteria that spontaneously form robust, resilient biofilm. Their ability to synthesize lipid-rich, waxy coating layers renders them extraordinary resistance to environmental challenges (Chakraborty and Kumar, 2019). Mycobacteria have been isolated from biofilms in numerous environmental sources, including household water pipes (September et al., 2004), where they readily enter drinking water (Caballero, Trugo and Finglas, 2003). This genus also contains human pathogens such as M. tuberculosis and M. leprae, which, strengthened by its biofilm-forming ability, brings severe clinical and therapeutic implications (Esteban and García-Coca, 2018). Thus, it would be ideal if we could engineer Mycobacteria, harness their biofilm-forming machinery so that they outcompete harmful biofilm of its own species, or even other species.
The strain we’ve been working with is Mycobacterium smegmatis (MC2155), which is a non-pathogenic, BSL-1 strain commonly found in soil, water and plants. Due to its similarity to other pathogenic Mycobacterium strains, M. smegmatis has been used as a model for researchers to study Mycobacterium species(Yip, 2007). M. smegmatis has also been used in bioengineering for the synthesis of various organic compounds, such as testosterone (Fernández‐Cabezón et al., 2017) and cofactor F420 (Bashiri et al., 2010). However, it is not widely used in synthetic biology as a chassis, even though it is a great model organism for engineering biofilm. In iGEM registry, M. smegmatis-related parts are not abundant; most of them are coding sequences to be expressed in E. coli. There is a huge need for Mycobacterium-compatible regulatory sequences. Hence, we want to establish a part library containing promoters, RBSs, and terminators that allows protein expression in Mycobacterium smegmatis. We constructed these parts, along with shuttle vectors, in a 3G compatible format. This allows future iGEMers to engineer Mycobacterium smegmatis with this speed, modular cloning method.
Shuttle Vector
In order to assemble circuits in E.coli and express this circuit in Mycobacterium, a shuttle vector with Origin of Replication for both E. coli and M.smeg is required. After searching through literature, we decided to use shuttle vector pSUM36 constructed by Aisan et al(1996). This vector contains pACYC184 origin of replication for E. coli, pAL5000 origin of replication for mycobacteria, and a kanamycin-resistance cassette. We ordered a plasmid from addgene of pSUM36 vector with gfp inserted, which is deposited by Nicolai van Oers Lab. Primers were designed to linearize the pSUM vector as well as flanking in with UNS1 and UNS10 for 3G assembly. The linearized backbone sequence w/ UNS1 and UNS10 can be found here:
Design 3G Compatible Parts
Last year the William and Mary iGEM team introduced 3G assembly and deposited a library of 3G compatible parts. 3G (Golden Gate-Gibson) assembly is a hybrid method of DNA assembly. In 3G assembly, many variants of multi-part circuits can be constructed in a single day with high accuracy and efficiency. We would like to take advantage of this efficient DNA assembly method and apply it to engineering Mycobacterium. To start with, we collected a library of parts that are endogenous to Mycobacterium, which consists of 8 promoters and their corresponding RBS from Li et al. (2017), rpsL promoter from Kenney and Churchward(1996), S16 promoter from Bashyam and Tyagi (1998), and 4 terminators from WebGeSTer database (Mitra et al, 2010). We flanked every part with its corresponding sticky ends, BsaI site for Golden Gate Assembly, and Pad sequences for Gibson Assembly. Eight of the parts have been functionally and sequence confirmed. A full list of parts can be found here:
| BBa_K3059430 | 3G rpsL promoter |
|---|---|
| BBa_K3059432 | 3G pMSMEG 3050 promoter |
| BBa_K3059433 | 3G S16 promoter |
| BBa_K3059434 | 3G MSMEG 2389 5’ UTR |
| BBa_K3059435 | 3G MSMEG 3050 5’ UTR |
| BBa_K3059436 | Terminator #1 |
| BBa_K3059437 | Terminator #2 |
| BBa_K3059423 | weak constitutive mScarlet-I expressing circuit |
| BBa_K3059424 | strong constitutive mScarlet-I expressing circuit |
| BBa_K3059438 | UNS1/10 pSUM36 shuttle vector |
Test Circuits Design
With the library of compatible parts, we assembled two circuits: BBa_K3059423 and BBa_K3059424. mScarlet-I is placed downstream from MSMEG 3050 5’UTR under the control of constitutive promoter pMSMEG_3050 or S16 as a reporter. Circuits are gibsoned onto UNS1/10 pSUM36 shuttle vector and electroporated into electrocompetent Mycobacterium smegmatis.
Patterning
As patterning is inherent to living organisms, we must pattern living biomaterials in order to exert complete control over their gene expression and related function. Patterning can not only control biomaterials, but also expand upon their ability to develop sophisticated structures and accomplish complex tasks. To accomplish patterning in our biofilms, we engineered quorum sensing and investigated Turing patterns.
Quorum sensing is a phenomenon which allows bacteria to communicate with one another. Thus communication allows for bacteria to make decisions as a complete, multicellular unit. Many researchers have taken advantage of these naturally occurring signaling pathways to engineer cell communication systems. Our engineered quorum sensing system uses an IPTG-inducible sender strain and a seperate receiver strain to create a distance-dependent signaling pathway. For the sender plasmid (left diagram), the HSL-producing enzyme is controlled by a pLac promoter, with constitutively expressed LacI. Upon addition of IPTG, LacI no longer is able to bind to DNA, relieving repression. The HSL-producing enzyme, in this example LasI is expressed. When expressed, LasI produces the respective HSL molecule. This HSL molecule is then able to freely diffuse out of the cell. When and if the HSL molecule enters a receiver cell, it is able to bind to constitutively expressed LasR. LasR is then able to bind to DNA and enhance transcription at the pLas promoter. Behind the pLas promoter is a gene of interest that we wish to be expressed. In the example of our test circuit, the gene expressed is mScarlet so that we can measure fluorescence output to verify functionality.
Overview
In 1952, Alan Turing questioned how a group of undifferentiated cells could transform into an organism. This question led him to a fascinating discovery: the formation of Turing patterns seen in nature, such as those found on animal coating and shells, as well as in mineral formation (Yang), are caused by the instability that exists in reaction diffusion systems (Turing). This instability is due to the significant difference in diffusion rates between morphogens (Turing). However, the conditions needed for classical Turing patterns to form are challenging to implement in living systems (Karig). However, in their paper “Stochastic Turing Patterns in a synthetic bacterial population,” researchers Karig et al. developed a system to allow for the creation of stochastic Turing patterns in E. coli.
Turing patterns are caused by the difference in diffusion rates of a slow-diffusing activator and a fast-diffusing inhibitor signals across a lawn of cells. The researchers designed a positive feedback loop for E. coli in which an activator signal, A3OC12HSL, activates production of itself as well as of an inhibitor signal, IC4HSL, which inhibits both the activator and itself. In one of their designs, which used plasmids PFNK-512 and PFNK-804-lacO-lacI, the activator signal starts the first signalling pathway when it binds to LasR, activating the production of both A3OC12HSL and IC4HSL by activating the promoters pLas-OR1, which control lasI (LVA) and rhll. Then, the inhibitor signal binds to RhlR, allowing for the activation of PRhl−lacO and the subsequent expression of the cI repressor. By repressing the p(Las)-OR1 promoters, the cI repressor inhibits production of both the activator and repressor signals. By repressing the λP(R−O1) promoter, the cI repressor also inhibits production of RhlR, and therefore, its own expression. This system is induced by IPTG, which relieves the repression of pRhl-LacO by LacI that occurs before induction, allowing for expression of cI. However, they found that slightly modifying plasmid PFNK-804-lacO-lacI by replacing the λP(R−O1) promoter with the constitutive promoter pLacIq and RBSH with RBSG to prevent the cI repressor from repressing expression of RhlR. This was found to make the differentiation between red and green fluorescence clearer.
Circuit Design
This design was based on the system designed by researchers Karig et al. and described in the Supplementary Information of their paper “Stochastic Turing Patterns in a synthetic bacterial population.” PFNK-512, which activates the activator and inhibitor signals, consists of the genes LasI (LVA), rhlI, ds-Red, and LasR, which are under the control of two identical versions of the hybrid promoter pLas-OR1 (which is made up of the cI binding domain OR1-mut4 and a P. aeruginosa Las promoter so that it can be activated by the activator signal and repressed by the cI repressor). PFNK-816, a modified version of PFNK-804-lacO-lacI, which inhibits the activator and inhibitor signals, consists of the genes lacI, rhlR, cI, and GFP(LVA) and promoters pLacIq and PRhl−lacO. pLacIq, a constitutive promoter, is used for constitutive expression of lacI and the mutant RhlR (RhlR-I124F) that is more responsive to lower concentrations of the inhibitor signal. Promoter PRhl−lacO consists of a LacO operator binding site and a P. aeruginosa RhlR promoter, and controls the expression of genes cI(LVA) and GFP(LVA).
