Analogous to the transfer of data as electronic packets in a computer network, the concept of engineering bacteria to transport packets of chemical signals between bacterial populations arose. In this way, the bacteria themselves can form network connections.

In order to aspire towards this ultimate objective, we decided to focus on reprogramming bacterial motility. Each bacterium will travel in a stimulus-directed fashion to specific points of the network (the nodes), where they can communicate signals. Such a stimulus should be unique to the bacterium and not be part of the natural environment. With these specifications in mind, we needed a way of manipulating the chemotactic behaviour of bacteria.

During chemotactic movement, the expression of CheZ induces counter-clockwise (CCW) flagellar activity - resulting in a “run and tumble” bacterial motion (Figure 1). For more on bacterial chemotaxis, see here.

The WITS iGEM 2011 team focused heavily on the key idea of guiding E.coli to move up a concentration gradient of a specific exogenous analyte – through controlled expression of this motility factor, CheZ. Our ambitious goal was to show that we can reprogram bacterial chemotaxis and thereby transfer products of any coding sequence to a desired location.

“Biotweet” is a theoretical concept of a framework for bacterial networks, using reengineered bacterial motility. Any network requires the establishment of connections between nodes of the network, allowing for the transport of node-to-node data packets. Reengineering the mechanisms of chemotaxis is the first step in the assembly of such a bacterial communication system. With “Biotweet”, we have attempted to construct such a network within a microbiological setting: a challenging task, considering that bacterial cells cannot be linked by wires.

Riboswitches are ligand-inducible RNA protein expression regulators that are comprised of an aptamer domain and an expression platform (Gallivan, 2007). The aptamer is a sequence of nucleotides that is designed to specifically bind to ligands, while the expression platform consists of a ribosome binding site (RBS) and a downstream coding region (CDS). When the specific ligand or analyte binds to the aptamer domain, the riboswitch undergoes a structural change. This results in the exposure of the RBS and the expression of the downstream CDS (Figure 2, top left panel).

By linking CheZ expression to post-transcriptional activation by a unique analyte (using an analyte-specific riboswitch), bacterial motility can be guided towards a specific stimulus. We chose to use artificial riboswitches to control the expression of the flagella rotation regulator protein CheZ. Using this approach, the expression of CheZ can be regulated in a ligand concentration-dependent manner, outside the control of the natural chemotaxis pathway (Figure 2, right panel).

We have decided to include two fluorescent reporters as fusion proteins (both N- and C-terminal CheZ fusions are well-tolerated, according to Gallivan, 2007). This allowed us to observe activation of the riboswitches in the context of our experiments.

We borrowed heavily from studies which have described riboswitch-controlled bacterial chemotaxis (Topps and Gallivan. 2007). Topps and Gallivan (2007) demonstrated that these riboswitches can be used to reprogram the chemotactic response, allowing bacteria to move up a concentration gradient of the analyte that activates the riboswitch (a process known as pseudotaxis). Bacteria can theoretically be reprogrammed to respond to any stimulus, if the appropriate riboswitch is made. This provides the versatility needed for biological network connections to be established in any application. Furthermore, this eliminates the need to engineer novel chemoreceptors.

Figure 2. Using ligand-responsive riboswitch activation of CheZ expression. CheZ can be fused to a reporter and direct bacterial motility in a ligand-dependent manner up a concentration gradient. Note: Fluorescent reporter proteins were fused to CheZ for observation of riboswitch activation during experimentation. (Click to enlarge)

Using analyte-induced CheZ expression, the main objective of project “Biotweet” is to illustrate the ability to reengineer bacterial chemotactic behaviour through the activation of riboswitches. Construction of such a mechanism is done in a modular fashion, producing an opportunity for scientists to not only direct bacterial movement to a specific location, but additionally, the prospect of delivering products of any protein-encoding sequences to that same site. This can be done by fusing any translational unit to the CheZ sequence in the riboswitch.

Figure 3.WITS iGEM 2011 riboswitch design. The riboswitch is linked to the motility activator CheZ protein, which can in turn be fused to any coding sequence. (Click to enlarge)

Figure 3.“send” and “receive” riboswitch-controlled motility module. Featured is also an Analyte B-responsive toggle switch which activates Crerecombinase expression. Cre, in turn, causes “receive” motility activationby flipping constitutively active promoter, and allowing motility to be controlled by Analyte A. (Click to enlarge)

Unfortunately, other than Tetracycline (Dox) riboswitches for use in yeast, at present, there isn’t an engineeredriboswitch which is sensitive to a known transcriptional repressor. Therefore, we designed system which allows the “receive” function (Crerecombinase) to be activated by IPTG (lacI). Also, we focused this design on the use of two known synthetic riboswitches – against theophylline and atrazine (see riboswitch design considerations below).

Figure 4. “send” and “receive” riboswitch-controlled motility module. In this design, Cre-recombinase is activated by IPTG which is present only at the theophylline position (not present in the theophylline gradient - green). As with Fig.3, Cre, in turn, causes “receive” motility activation by flipping constitutively active promoter, and allowing motility to be controlled by an Atrazine gradient (red). (Click to enlarge)

Our initial aim was to develop a theophylline-sensitive riboswitch which is capable of regulating CheZ (motility factor) translation. Several theophylline-sensitive riboswitches have been developed, most notably by the Gallivan lab (Emory, USA).

To summarize, using a high-throughput screen(Lynch et al., 2007) and a high-throughput selection (Topp and Gallivan, 2008) in E.coli, the Gallivan lab has been able to develop a powerful theophylline-dependent riboswitch. It is worth noting that the motility-based screen(Topp and Gallivan, 2008) made use of a theophylline riboswitch library activating the expression of CheZ, the motility factor in E.coli which allows for running (as opposed to tumbling) movement. Therefore, activation of CheZ allows bacteria to move along the theophylline gradient. It is worth mentioning that the best riboswitch identified in these earlier screens was the theophylline riboswitch clone 8.1 (see figure bellow), which displays 36-fold expression activation ratio (Fig.5).

Figure 5. Theophylline responsive riboswitch (clone 8.1). We have referred to this riboswitch as theophylline riboswitch type 1 (thRS1). (Click to enlarge)

In order to improve on their earlier riboswitches, Topp and Gallivan developed an in vivo-based selection strategy aimed at selecting bacteria based on fluorescence activated cell sorting (FACS) (Wieland et al., 2008; Lynch et al., 2009) (Fig.6). By randomizing the sequence encoding the RBS (+ Shine Delgarno/SD), and by sorting a N12 library, clone 12.1 was identified, which displays a 96-fold expression activation in the presence of theophylline (1 mM) (see figure below). The authors of this study suggest this is the largest activation ratio to date, for synthetic or natural riboswitches. It is worth noting that this riboswitch contains a strong (and longer) UAAGG SD which is found in a tight (inaccessible) duplex in the “OFF” (ie no theophylline) state. Moreover, the UAAGG sequence is spaced optimally, with 6 nt separating the 5’ A of the anti-SD sequence and the start codon of the translation (AUG) (Fig.6).

Figure 6.Theophylline responsive riboswitch (clone 12.1). We have referred to this riboswitch as theophylline riboswitch type 2 (thRS2). (Click to enlarge)

The atrazine riboswitch was developed recently by Sinha et al (2010). In order to make this riboswitch compatible with biobrick assembly, Pst1 site was removed by creating a G:U wobble base pair in lower stem region (Fig 7).

Figure 7. Atrazine responsive riboswitch. We have referred to this riboswitch as atrazine riboswitch (atRS) (Sinha et al., 2010). (Click to enlarge)

We have identified that several theophylline riboswitches have been previously described by other iGEM teams (For reference, see Lethbridge 2007, Lethbridge 2009 and NYMU Taipei 2010). One of our initial concerns was whether it was possible to isolate the riboswitch from the associated translational unit (coding region). Being able to do so would allow the riboswitch to be defined independently as a BioBrick, with the useful ability to act in a modular fashion for fusion to an adjacent translational unit. All previous teams have made this assumption. We will show that the riboswitch and its adjacent coding region should be considered together and should be cloned as one biobrick, or an alternative "acanonical" assembly standard should be used.

Team NYMU Taipei assumed that a created theophylline riboswitch (Gallivan lab clone 8.1) biobrick could be fused to a GFP coding region by standard biobrick assembly techniques. While it is possible to do this, we feel that the resulting 6 nt scar sequence insertion increases the distance between the RBS and the ATG start codon, a result which is likely to cause a decrease in riboswitch efficiency (see team Taipei switch below, riboswitch highlighted in yellow, inserted biobrick “scar” region highlighted in violet, ATG start underlined). In fact, Taipei 2010 did show data which showed only a 2.5 fold increase in gene activation at approx. 4 mM theophylline. Compared to the 36-fold activation at 1mM theophylline by Gallivan lab, this seems too low! We therefore feel that this riboswitch is not behaving efficiently.

Figure 8. BBa_K411001,BBa_K411003 - Team NYMU-Taipei iGEM2010

Team Lethbridge 2009, used the same theophylline riboswitch (clone 8.1) to generate a riboswitch-CheZ construct. However, this team created a 11 nt insert after the ATG start codon. This result created a TAG stop codon (in the 3rd amino acid position) as well as a frameshift mutation. Therefore it is unlikely that a functional CheZ could have been produced from this system. (see team Lethbridge 2009 below, riboswitch highlighted in yellow, inserted biobrick “scar” region highlighted in violet, ATG start underlined)

Fig 9.BBa_K249028, TaemLethbridge iGEM2009. Theophylline riboswitch - scar - CheZ -dT

Gallivan JP. Toward reprogramming bacteria with small molecules and RNA.CurrOpinChemBiol 2007;11:612-9

Topp S, Gallivan JP. Guiding bacteria with small molecules and RNA. J Am ChemSoc 2007;129:6807-11.

Lynch S.A. Desai S.K.,Sajja,H.K., and Gallivan J.P.A high-trhoughput screen for synthetic riboswitches reveals mechanistic insight into their function. 2007, ChemBiol 14:173-184

Topp S. and Gallivan J.P. Random walks to synthetic riboswitches – a high throughput selection based on cell motility. 2008, ChemBiochem 9:210-213

Wieland M., and Hartig, J.S. Improved aptazyme design and in vivo screening enable riboswitches in bacteria. 2008, Agnew.Chem.Iny.Ed.Engl 47:2604-2607

Lynch S.A. and Gallivan J.P., A flow cytometry based screen for synthetic riboswitches. 2009, Nucleic Acids Res. 37(1)184-192

Sinha J, Reyes SJ, Gallivan JP.Reprogramming bacteria to seek and destroy an herbicide.Nat Chem Biol. 2010 Jun;6(6):464-70. Epub 2010 May 9.