Characterisation of dCas9-ω bio-transistors
In order to build dCas9-controlled circuits, we aim to use dCas9-inducible synthetic promoters to mimic the behaviour of a transistor (Go take a look at our design (LINK) if you haven't !). To see whether such promoters could be used in wider-scale circuits, we identified and tested the following desirable properties :
- Transistor response : Is the transistor inducible? Is its output modulable allosterically
- Transistor mutability : Is it possible to mutate the dCas9-targeted sites on transistor-like promoters without altering their behaviour and output levels ? Would it be possible to make a family of homogenous transistors that can be regulated independently?
- Transistor orthogonality : is regulation of a transistor specific? Is there cross-talk? Could several transistors work simultaneously?
- Transistor chainability : Can transistors can be linked serially to allow for multiple levels of information processing ?
dCas9-w Transistor response: BBa_K1723001 is inducible. Activation is repressed when both activation and inhibition sites are bound by dCas9-ω
dCas9 binds DNA complementary to its guide RNA (gRNA). The location it binds to may therefore be controlled by producing specific gRNAs.
By targeting dCas9 fused to a RNAP recruiting subunit (dCas9-ω) to an adequate distance from the transcription starting site (TSS), PAM-rich_J23117 (BBa_K1723001) was successfully induced, which represents a PNP transistor in “on” state.
The bio-transistor was designed to produce GFPmut2. Promoter activity for each input combination was assessed by measuring Relative Fluorescence Units (RFU) and normalised by OD600 which should be correlate linearly to the promoter's strength .
We chose to test two sites close enough to the TSS to (likely) sterically hinder RNA Polymerase (RNAP) attachement when bound by dCas9-ω. By itself, targeting dCas9-ω to inhibition sites (R1, R2), sometimes resulted in slight activation (experiment 3) and sometimes in slight inhibition (experiment 1). This might be explained by the already low promoter strength of J23117 (close to autofluorescence in our measurements and consistent with the registry BBa_J23117) and the presence of the activating ω subunit on dCas9-ω. Overall, R1, R2 and R1/R2 produce fluorescence close to basal levels and the transistor is considered to be in “off” state.
Yet, when activation and inhibition sites are targeted simultaneously, dCas9-w bound to inhibition position hinders induction (A/R1, A/R2). We found out that A/R2 may virtually suppress activation; the promoter is in “off” state. Such an effect has been reported for S. cerevisiae (REF) but -to our knowledge- hadn’t been tested in E. coli.
RFU discrepancy between A/R1 and A/R2 could be explained by our modeling simulations. A/R2 produces an output consistent with predictions for a model where activation and inhibition sites may be bound by dCas9-w without steric hindrance between the dCas9-ws: binding is independent. A/R1 on the other hand behaves as if co-binding of the activation and inhibition sites were impossible. Interestingly, we notice that for A/R1 the activation and the inhibition sites are on the same strand, whereas in A/R2 the sites are a bit farther and on different strands.
Transistor Mutability: The promoter’s regulatory sites may be mutated while conserving its transistor behaviour
We selected dCas9-w targets outside of the -10 and -35 regions on BBa_ and mutated them. Thereby, we created an alternative transistor (BBa_K1723005) and set of gRNA inputs.
The basal levels of the original promoter and the alternative one are very similar. Nielsen  has reported the use of mutated synthetic promoters between -35 and -10 regions to build regular dCas9 inhibition-logic circuits. Due to time constraints, we couldn’t produce “double” activation/inhibition inputs. The response of the mutated promoter for single gRNA input, however, suggests that it behaves similarly to the original promoter.
Relative strength between activation of BBa_1723001 and is mutated cousin BBa_1723005 using specific gRNA displayed moderate variability but activation and inhibition patterns were preserved (measurements 4 and 3
Orthogonality : Transistors do not produce activation or inhibition patterns when the input gRNA is not complementary to their regulation sites
All our reporter + wrong gRNA constructs showed similar fluorescence levels and no activation or inhibition pattern. They are also very close to autofluorescence and basal levels; they are in the “off” state. In this particular experiment, basal levels were measured with cells lacking apo-dCas9-ω producing plasmids. This might explain the slight difference in RFU between basal configuration and non-specific gRNA + dCas9-ω constructs. This result is consistent with studies of dCas9’s specificity in the literature .
Unfortunately, we lacked time to test specific regulation of one transistor among several within the same cell. Nevertheless, clues that multiple transistors can work simultaneously can be found in the chaining experiment below, where two non-orthogonal promoters worked in conjunction. The transistor response experiment is also informative on the orthogonality of signals as two gRNA inputs seem to have worked in parallel to produce the A/R1 response (see above). Moreover, the model predicted that despite sharing the intracellular apo-dCas9 pool, the gRNA inputs are still able to form a sufficient amount of gRNA-dCas9 complexes to work as expected
Chainability : One transistor can propagate signal to another transistor
Inducing the upstream transistor successfully resulted in induction of downstream transistor (A_u) to levels comparable to direct activation of BBa_1723005 (A_alt). We notice that the basal expression of the downstream reporter BBa_1723005 in chained configuration (B_d) is higher than the basal expression of BBa_1723005 alone (B_alt). This is probably due to leakage from the upstream promoter, as repressing the upstream promoter (R_u) resulted in lower RFU/OD600. We therefore believe that our bio-transistors could be chained to form more complex circuits. As far as we know, chaining dCas9-ω-regulated promoters hadn’t been attempted. This might be the first exclusively dCas9-based bio-circuit in E. coli !
Kinetic response of a dCas9-ω transistor
Bacterial cultures were monitored in a plate reader for 17 hours. We restarted the cultures twice, which is revealed by two sharp drops in OD600. We observe most of the log phase correlates with decreasing RFU/OD600. We conjecture that E. coli replicate faster than the time needed for the transistor + reporter system to follow: if OD600 grows faster than the RFUs, the normalised fluorescence result will decrease. This pattern was observed in all our kinetic experiments when bacteria were observed in log growth phase. In this particular experiment the time needed to tend toward equilibrium is slower than usual. In most measurements, the transistor stabilises faster (~ 300 min).
In this experiment, we used arabinose(ara)-inducible promoters to produce gRNA inputs. This measurement provides clues as to how the transistor responds with increasing signal strength. We notice that activation (A) is impacted by the input levels. On the other hand, in the inhibition condition the levels are already close to basal and [ara] has little effect on the transistor's output. Testing simultaneous activation and repression constructs should, however, provide more information of transitor modulability. The measurement of simultaneous activation and repression has been planned and should take place soon...
dCas9-ω production in our constructs is anhydrotetracycline (aTc) inducible. Contrary to the experiments presented above - which were performed at a fixed concentration of aTc - we investigated the impact of [aTc] on transistor response.
Surprisingly we observe that activation decreases when [aTc] is too high. Conversely, what should have been an inhibiting condition shows increasing activation.
We believe that the presence of the ω subunit plays a key role in explaining this result. An interesting further experiment would be to target the same activation and inhibition sites with regular dCas9.
 Shiue, S.-C. et al., 2015. Expression profile and down-regulation of argininosuccinate synthetase in hepatocellular carcinoma in a transgenic mouse model. Journal of Biomedical Science, 22(1), p.10. Available at: http://dx.doi.org/10.1186/s12929-015-0114-6.
 Nielsen, A.A. & Voigt, C.A., 2014. Multi-input CRISPR/Cas genetic circuits that interface host regulatory networks. Molecular Systems Biology, 10(11), pp.763–763. Available at: http://dx.doi.org/10.15252/msb.20145735.
 O’Geen, H. et al., 2015. A genome-wide analysis of Cas9 binding specificity using ChIP-seq and targeted sequence capture. Nucleic Acids Research, 43(6), pp.3389–3404. Available at: http://dx.doi.org/10.1093/nar/gkv137.
Characterisation of dCas9-VP64 bio-transistors
As the transistor is the simplest unit in complex circuits, we want to characterize it for further applications. We will test different features, such as the activation, the repression, both activation and repression simultaneously and orthogonality.
As a little reminder, we use 2 repressing gRNAs, c6 & c7, whenever we repress the promoter.
As dCas9-VP64 is under control of a Galactose + aTc promoter, we tested different aTc concentrations to obtain a better response. We discovered that the response of the system seems completely independent of the concentration of aTc!
These two figures show the response of the transistor with different gRNAs:
- An activating one (in clear green, BBa_K1723009)
- An inhibitory one (in brown, BBa_K1723013 and BBa_K1723017)
- Both activating and inhibitory gRNAs (in orange)
- gRNA that target an other promoter (to check the orthogonality, in dark green)
- No gRNAs (basal level, in blue)
- Only dCas9 (no gRNAs and no promoter, in dark red)
- Only the activating gRNA (no promoter, in red)
The activation is clearly above the basal level (at the end of the measure, when the equilibrium tend to be reached, the ratio between activation and basal level is approximately 1:4). Furthermore, the activating gRNA induce a higher GFP expression as the inhibitory one (ratio 1:2 approximately). This difference is even bigger with both activation and inhibition. This could easily be explained: when inducing activation and inhibition, three gRNAs are expressed (c3_0, c6_0 and c7_0). It means that three dCas9 proteins are present in a region less than 100 base pair long. This will obviously cause steric inhibition and prevent the RNA Polymerase to bind the promoter (just go check our modeling page for the dCas9 shape and minimal binding distance).
The orthogonality showed good results as well! When targeting the CYC0 promoter with a different gRNA (the inhibitory gRNA for promoter CYC1 in this case), the fluorescence is not influenced. It means that the building of orthogonal circuits that don’t interfere with each other is possible and feasible!
Unfortunately, due to lack of time, some tests as the chainability, the mutability or the measure of a logic gate response were not feasible.