A few members of the team, fresh from their exams, began the initial literature reviews necessary to identify the genes and promoters that were required in our biological sensors that would detect and respond to signaling molecules released by Stenotrophomonas maltophilia (DSF), Burkholderia cenocepacia (BDSF) and Pseudomonas aeruginosa (PQS). Finally, rpfC, rpfG, clp and P_manA were to be cloned to detect DSF; BCAM0227, BCAM0228 and P_cbld for BDSF; and pqsR and P_pqsA for PQS. Primers were designed (which included restriction sites required for cloning and also HA-tags to detect for protein expression) for the PQS sensing system and ordered from Sigma Aldrich. All sequences were analyzed with Webcutter 2.0 to detect restriction sites that conflict with the BioBrick assembly standard RFC [10]. It was found clp contained an illegal PstI site; BCAM0228 and rpfg contained an illegal EcoRI site. Primers were then designed for use with QuikChange™ Site-Directed Mutagenesis Kit to remove illegal sites. It was decided at this stage that for the DSF system, homologues of all sequences would be taken from Xanthomonas campestris pv. campestris as they have identical functions but contain fewer illegal restriction sites.
However, BCAM0227 was proving problematic. It was so large (3000+bp), and had so many illegal restriction sites, it would have taken the full ten weeks to QuikChange™ them out! We decided to send it for synthetic sequencing by Dundee Cell Products in two separate parts. We hoped this would allow us to successfully ligate the two pieces together into one without mistakes.

The team are all together! Work begins…
We began organizing our raw materials, setting up overnight cultures of our vectors (pSB1C3 for BioBricks, and our supervisor’s own pUniprom for experiments) and obtaining gDNA from P. aeruginosa, X. campestris and B. cenocepacia from Drs Robert Ryan and Shi-Qi An from the Division of Molecular Microbiology in the College of Life sciences at the University of Dundee By the end of the week we had cloned pqsR and P_pqsA into pSB1C3, phew!
As for the dry-lab, the first week consisted mostly of reading papers and harassing Gillian, our mathematical biologist, for definitions of all the bio-jargon. In fact, most of our early discussions about modelling our systems consisted of furrowed brows from both sides, and questions such as “a differential what?” and “Uhhh, transcriptional whatnow?”

In order to test our biological sensors we decided to clone them into pUniprom, a high copy-number plasmid which has a large and varied multiple cloning site downstream of a strong, constitutive promoter: P_tat. In order to arrange our sequences in such a way that there would be no leakage of our reporter, we had to employ a bit of cloning trickery by inserting our sensing and responding fragments in opposite orientations. This involved cloning them into separate fragments and then flipping them round into pUniprom. And so to the bench we went!
Now that the plan for the plasmid was in place, we began to enter all the information onto Netlogo to model how the system would behave in the presence of different concentrations of exogenous PQS.

With lots of cloning, comes lots of sequencing… which was all kindly carried out by the sequencing service downstairs from our lab. To help us along the way Professor Philip Cohen kindly granted us 100 free sequencing runs (which we quickly exceeded!). After every cloning step for building our PQS and DSF sensor, we sent a sample to be sequenced – the results were a source of merriment and heartbreak in equal measure…it turned out that our cloned PQS sensor wasn’t quite right so we spent the week trying to rectify this.
We also began the lengthy task of removing all the illegal restriction sites from the fragments bound for BioBricking; and so began Dimitrios’ epic journey into the world of BioBricks!
It turned out, (after a bit more sifting through papers), that the Netlogo model for PQS was not quite right. While we figured that out, we began to calculate the ordinary differential equations for the same system.

We were all excited to test our first sensor, an inducible GFP production assay was set up, by spiking overnight cultures of transformed E. coli MC1061 with a range of concentrations of synthetic PQS. Unfortunately, we obtained no mCherry expression....along with a lot of perplexed biologists! We successfully removed the first illegal site from rpfC. Meanwhile clp BioBricking was underway, and work had begun on the BDSF system. We were halfway through the 10 weeks and we were feeling productive! The dry lab were certainly ahead of the game as far as BDSF was concerned; after working through the mammoth DSF system, the closely related two-component BCAM0227/BCAM0228 system was a breeze. In fact, with such a head-start it was decided that there would be a presentation given by the maths boffins of the team entitled: “Modelling for the Perplexed Biologists and Dave”. This was an interactive lesson on how to apply equations to biological systems which helped the lab-rats get to grips with how the figures being generated by the dry team could inform the final design of the biological systems. By the end of the week, rpfG, rpfC and clp were cloned into pUniprom!
Around this time, the “Device Committee” was formed; a small group of wet and dry members who would begin the design of a physical unit to house and interpret the output from our engineered bacteria. This began with background reading on creating a GFP-fluorescence reading device. Order initial parts to build a light sensing device based on LED light source so we could modulate the light levels. We considered that we would have to use MatLab to carry out calculations. Our visits to the clinic took a more direct focus, and during the interviews we were conducting each week, we started asking patients for their thoughts on the design of the physical detector.

Sequencing results for rpfG QuikChange™ showed that it was actually rpfC…oops! Steps were made to rectify this quickly!
All promoters had now been successfully cloned and sequenced and so the job now was to clone them along with their fluorescent reporters into pUniprom. Our two sequences of BCAM0227 were successfully returned from the sequencing company, Dundee Cell Products, for us to work with. The two separate sequences were digested, and work commenced to try and ligate them together into pbluescript...easier said than done!
Initial experiments with a photodiode, Arduino and a green LED for the physical device, commenced. First designs showed us that we would need an excitation source to induce fluorescence, this would necessitate the inclusion of filters and amplifiers…extremely complicated! Discussions were had with the entire team about eventually using firefly luciferase as a reporter in the place of fluorescent proteins.

After lots of modelling of the PQS system, and trying to troubleshoot why our system was not expressing mCherry, the decision was made to design another two biological sensors for P. aeruginosa, which would detect another quorums sensing molecule Pseudomonas autoinducer 1 (PAI-1). Lots of work for the wet team!
An overnight culture of E. coli MC1061 transformed with the DSF plasmid was pelleted and prepared for western blotting for detection of our proteins - alas, RpfC-HA is not being expressed! Back to the cloning board… BioBricking of rpfG was undertaken upon receiving good news from sequencing that our illegal EcoRI restriction site had been removed.

The tricksome BCAM0227 was successfully ligated together into one gargantuan sequence, and swiftly cloned into pUniprom (already containing the P_cbld-gfp fusion).
BCAM0228 QuikChange of the illegal EcoRI site was complete and sent for sequencing to confirm this was actually the case. The Device Committee met up with Dr. Stephen Reynolds from the University’s physics department, who helped us to optimize our light-detecting circuit. We constructed a crude (but effective!) Faraday cage from aluminium foil and a cardboard box to eliminate electrical interference from the buzzing computers in the dry lab…
We managed to clone rpfC into pUniprom, along with the rest of the DSF system; time for some tests!

Sequencing results came back for the QuikChanged BCAM0227 and confirmed that the EcoRI site had been removed and so we handed it over to Dimitrios to clone it into pSB1C3.We obtained a culture of Vibrio fischeri from Dr Sarah Coulthurst’s lab, and began streaking plates to test out our physical device for detection of the bioluminescence from V. fischeri luciferase – it worked! At this point our device was essentially half of a water bottle covered in insulating tape, an Arduino and a voltage meter...However, we had come up with a product name: the Light Amplifying Signal Sensing Object, because every good cowboy needs a L.A.S.S.O., right?! It was time for a more robust design; our thoughts moved to 3D printing, and iGEM alumnus and polymath Nasir Ahmad showed us how to draw up plans using Google Sketch.

At this stage, we had all of our systems (barring PAI-1) cloned into pUniprom and ready for blotting. So with a little, (actually, rather a lot), of help from "The Blotfather" Prof. Frank Sargent, we set to the task with all the elation of a team that can see the end of cloning in sight.
Meanwhile, some of the wet team set to the noble task of cloning our replacement Pseudomonas system. Luckily, all the components we needed were included in the 2014 iGEM distribution pack so we stabbed, rehydrated and got to work. We started by using the P_tet BioBrick as a vector and cloned in lasR downstream.
The Device Committee met with Dr. Mike MacDonald from the Institute for Medical Science and Technology (IMSAT) to discuss plans for building a physical device. They learned from him the importance of following international standards of operation (ISOs). From this point on we endeavoured to follow ISOs 13485 (medical devices) & 9001(general quality management) so that, if the device were to be taken on by a company, our design and parts would be easily replicable.

After western blotting the BDSF system, we obtained results which showed that both BCAM0227 and BCAM0228 were both being expressed. We then set up for an inducible GFP production assay, by spiking overnight cultures of transformed E. coli MC1061 with a range of concentrations of synthetic BDSF. Unfortunately, GFP appeared to be expressed at all concentrations, including 0 µM! This experiment clearly needed to be repeated... The cloning of the two PAI-1 sensors was continued, by ligating two different promoters downstream of LasR; P_lasb and PluxR. The Device Committee were having a great time designing the L.A.S.S.O.; a cavity here, a plunger there, maybe some springs... taking into account the ISO requirements as specified by Dr MacDonald.

Dimitrios has been busy and managed to biobrick P_cblD.
The growing PAI-1 system was complemented with the addition of the nanoluc BioBrick. The dry team set about modelling the potential savings that could be made from the implementation of The Lung Ranger/L.A.S.S.O. They began by investigating the cost for the NHS of keeping patients in hospital for treatment, which could be given domestically if the need was identified earlier. Also under consideration was the amount of days taken off work by the patients when ill or visiting the clinic – a concern raised by some patients we met at the clinic.

The biobrick man strikes yet again.....P_manA was successfully cloned into pSB1C3. More L.A.S.S.O. testing with V. fischeri – this time with plates containing different amounts of bacteria, obtained by serial dilution of an overnight culture. This resulted in our first attempt to map the sensitivity of our device vs. c.f.u. of bacteria. The BDSF GFP blot was repeated with similar results, and the decision was made to begin removing genes from the system until we identified which one was causing the problem.
Western blots of the DSF system showed successful expression of all three proteins; RpfC, RpfG and Clp. We then set up for an inducible GFP production assay, by spiking overnight cultures of transformed E. coli MC1061 with a range of concentrations of synthetic DSF. Similar to our BDSF system, GFP appeared to be expressed at all concentrations, including 0 µM! We need to come up with a plan, quick!! Sequencing of the PAI-1 system showed that the nanoluc sequence was missing, hmm…..

Repeated ligations of PAI-1 system with nanoluc luciferase. The final sequencing results from the BioBricked rpfc was celebrated with party hats...no more BioBricks for Dimitrios, woohooo!!
We received our 3D printed L.A.S.S.O., it looked great but didn’t fit together as planned… out with the Stanley knife! No problem. The GFP assay of the DSF system showed similar results to that of the BDSF system; surely not?! The experiment was repeated and the same results obtained. Oh, dear!
The final L.A.S.S.O. design was almost complete, with only some means of safe disposal for our engineered bacteria to be factored in. We settled on a design that would involve sealed petri-dishes containing freeze-dried E. coli with a one-way valve on the lid, through which a sample of sputum and water would be passed. The plate would then be inserted into the L.A.S.S.O. and all sources of external light eliminated by sliding flaps. The photodiode would pick up luminescence from the luciferase, expressed in response to the bacterial signaling molecule of interest, and send a voltage to the Arduino. This would pass this signal via a µUSB cable to a computer, where an application would interpret the signal into an email containing information about the bacterial species present, which would be sent to a clinician. The plate would then be transferred mechanically into an isolated waste compartment which can be removed and emptied into a biological waste container when full.

Why are our sensors constantly producing GFP?? Instead of painstakingly removing each gene, we decided (on the advice of the University’s top microbiologist Prof. Tracy Palmer) to blot the DSF and BDSF systems for GFP when only the promoter/reporter fusions were present. This was a breakthrough as the P_manA-gfp fusion still showed GFP expression while the P_cblD-gfp fusion did not. This lead us to perform a BLAST search of the Clp sequence from the DSF system against the E. coli genome to look for similar transcription factors in the wild type. CRP from E. coli showed significant similarity so we obtained a CRP-deletion mutant (BW25113 ∆CRP) and transformed it with our promoter/reporter fusion in preparation for blotting.
BCAM0228-P_cblD-GFP was set up in E. coli MC1061 for blotting, as there may be potential crosstalk causing GFP to be produced in the absence of our sensor kinase, BCAM0227.

The two PAI-1 sensors looked good from sequencing and so was transformed into E. coli MC1061 in preparation for blotting for GFP after induction with synthetic PAI-1 obtained from Sigma Aldrich.
Western blot of BCAM0228-P_cblD-gfp showed GFP expression. Contrasted with the negative results from the P_cblD-gfp blot, this suggested that something native to E. coli was activating BCAM0228.
A Western blot of P_manA-gfp in the CRP deletion strain of E. coli still showed expression of GFP.

The two PAI-1 sensors containing different promoters both worked to different extents. The sensor containing P_lasB-gfp showed slow induction of GFP over time, however the sensor containing P_luxR-gfp was initially on but showed increasing GFP production over time. What a relief. After much head-scratching and brow-furrowing over the other two systems it was great to get a positive result. We have learned a lot about the potential cross-talk between signal transduction pathways when they are expressed in different species. We believe that research into untangling these interactions will significantly speed-up the process of applying synthetic biology solutions to real-world problems.

Time to pack up the lab and hand over our equipment to the new PhD students for training, it’s been a rollercoaster of a project that every one of us has benefited from in ways we will likely still be discovering years from now.

See you in Boston!