Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS The Official PLOS Blog

Controlling translational resource allocation

 

Guest post by Alexander Darlington

We often design and build genetic devices in isolation and then bring the plasmids encoding them together in one cell; most of the time this results in an unexpected behavior. The devices function differently when sharing the same cell than they did when alone. Recent work suggests that this is, in part, due to competition for the cell’s gene expression machinery. We recently studied how feedback controllers can be used to manage the cell’s resources in an effort to reduce the impacts of this competition.

 Bringing modules together: Failure of modularity

When constructing genetic devices, synthetic biologists and biotechnologists use gene regulation to link different modules; by regulating the transcription of a module it is assumed that the levels of the protein will follow. Consider a biosensor where the metabolite of interest (i.e. the input) changes the binding of a transcription factor to the promoter of another gene. This downstream gene may be an enzyme to produce a useful product or another transcription factor which goes on to activate/inhibit other downstream processes. These different modules are said to independent; each can function individually, be characterised separately and have the same predictable function when assembled together.

Read more: Promoters, initiators of transcription and drivers of synthetic biology

However, these different modules do not function independently as each module draws on the same shared pool of RNA polymerase (RNAP) and ribosomes. At a given growth rate the concentrations of RNA polymerases and ribosomes are constant. Therefore, as one gene is expressed and utilises RNAP for its transcription and ribosomes for its translation, fewer of these resources are available for the expression of other module. As seen in Figure 2, GFP levels fall as RFP is induced. Effectively, the activation of one gene indirectly represses other genes as it sequesters resources away from them. These non-regulatory interactions result in module coupling which are not immediately apparent from circuit topologies. This competition for the same resources can result in the breakdown in the expected relationship between transcriptional regulation (input) and protein levels (output). Current experimental evidence suggests that competition for translational resources (such as ribosomes) is crucial in mediating these interactions.

How do we control allocation of ribosomes between different modules?

Due to the universality and complexity of the cell’s translational machinery, there is currently exist no sufficiently distinct ribosome which to create a truly orthogonal ribosome pool. However, ribosomes can be divided into host and circuit specific functions by the use of a synthetic ribosomal RNA (rRNA) component to create a quasi-orthogonal ribosome system. The ribosome is a large protein and RNA complex and the binding interactions between an mRNA and the core 16S ribosomal RNA are known to be a key regulator of translation initiation.  By expressing a synthetic 16S ribosomal RNA and expressing an mRNA with a complementary 5’ sequence (ribosome binding site, RBS) an orthogonal pair of ribosomes and mRNAs can be created. The orthogonal ribosomes (‘o-ribosomes’) translate only mRNAs with the corresponding ‘orthogonal RBS’. This allows the creation of a circuit specific pool of ribosomes.

We showed that dividing circuit genes between circuit-specific ‘orthogonal’ ribosomes and host ribosome pools can relieve the effects of resource competition by reducing the competition for that piece of translational machinery. Applying the concept of feedback control, we designed and implemented a prototype controller for orthogonal ribosome production. This controller acts to automatically match the supply of circuit specific translational capacity to the demand for protein synthesis by the circuit. In our recent follow-up work, we further investigated the function of this negative feedback mechanism, specifying design rules and offering potential biological implementations.

Building a translation resource allocation controller

Our controller acts to dynamically allocate the cell’s translational capacity between host and circuit genes in response to circuit demand using the circuit-specific translational activity conferred by the use of o-ribosomes (Figure 1A).

a Figure 1 (A) The topology of the feedback controller. As circuit genes are induced they compete with the controller mRNA for ribosomes. (B) The simulated dynamics of the feedback controller. A circuit gene is induced at 12 h (not shown). This results in a fall in the controller protein and hence increase in the o-rRNA promoter activity (due to the reduction of its repression by the controller). As o-rRNAs are produced they co-opt host ribosomes and so the total number of circuit specific orthogonal ribosomes increases. Work presented in Darlington et al (CC-BY 4.0).

 

Negative feedback utilises the output of a system to modify the system’s input such that disturbances or fluctuations can be rejected. For example, a room thermostat controls the temperature by sensing the room temperature and adding additional heat if it is too low, while if the temperature is too high the thermostat turns off the heating, to maintain a stable temperature. Here we consider a simple implementation of such a negative feedback controller which converts a disturbance in circuit translation due to induction of another gene into a change in o-ribosome production, such that demand for ribosomes is approximately matched by the supply of ribosomes.

Figure 2: A prototype controller was constructed based on the previous characterised o-ribosomes and the repressor LacI. In the absence of control, the orthogonal ribosome pool is constant and so the circuit genes (RFP and GFP) complete for ribosomes (‘Open loop’). As RFP is induced, GFP falls as described in the main text. This can be quantified by the gradient of the resulting RFP-GFP relationship. The controller successfully reduces this gene coupling by 50% as measured by the slope of the isocost ‘trade-off’ line (‘closed loop’). Work presented in Darlington et al (CC-BY 4.0)

To create our negative feedback controller, we take advantage of resource competition. The orthogonal ribosome pool is expressed under the control of a repressor. This repressor protein is constitutively active (i.e. its transcription is said to be constant) and utilises the o-ribosome pool for its own translation. This means that the repressor’s protein levels are a direct function of the orthogonal ribosome pool (Figure 1B). When a circuit gene is induced, its mRNA competes with the repressor mRNA for ribosomes. Therefore, the concentration of the repressor falls as o-ribosomes are shared (Figure 1B). The reduction in repressor concentration results in less inhibition of the o-rRNA promoter; so the transcription of orthogonal rRNAs increases (Figure 1B). This means that more ribosomes are co-opted from the host ribosome pool to the circuit specific orthogonal ribosome pool (Figure 1B). Our experimental prototype successfully reduces resource competition by 50% (Figure 2).

 

Next generation translational controllers: Elucidating further design rules

In our recent paper in ACS Synthetic Biology, we develop a detailed model of the controller which we simplified to a tractable level of complexity which allows for rational and optimised design. Analysis of the model reveals a fundamental design trade-off; that reducing coupling comes at the expense of gene expression. Whilst this trade-off is apparent in the absence of the controller, the controller allows higher levels of decoupling at a given level of gene expression at all levels of gene expression.

 

By perturbing each parameter in turn, we were able to determine how each parameter contributes to the controller behaviour. We find that both controller multimerization and ribosome binding strength are crucial to determining the trade-off. Increasing multimerization decreases coupling and increases gene expression, while increasing RBS strength increases coupling and decreases gene expression. High levels of decoupling are observed with high levels of controller mRNA production (e.g. via a high copy number plasmid or strong promoter) or a tightly binding repressor. We carried out an extensive literature search and simulated potential biological implementations and found candidate designs across the optimal design front.

Outlook: Implications for more complex circuits

As synthetic circuits increase in number of modules and complexity, resource competition will become an increasingly important factor in the synthetic circuit design process. Increasing the number of modules results in an increase in competition as more genes complete for the same pool of resources. We investigated the ability of our controller to decouple genes in multi-module circuits. We successfully showed that the controller can mitigate oscillations in a constitutively expressed gene which are brought about by sharing resources with an oscillator. For example, it has previously been shown that an activation cascade is sensitive to resource competition. If the upper node sequesters too many resources then the lower node cannot be expressed; if competition is sufficiently strong then the motif can even invert its function with the output of the activation cascade falling as the input increases. Our model predicts that our resource allocation system can abolish this effect.

At present, bringing different modules together into a single circuit can result in the emergence of unexpected interactions due to the emergence of non-regulatory interactions as resources are shared across circuit modules. Our recent results suggest that the use of circuit-specific ribosomes coupled with feedback controllers can relieve these interactions at the translational level.

 

Alexander Darlington is a Post-Doctoral Research Associate at the University of Warwick. His research interests lie in synthetic circuit design and control of biological processes; with a focus on host circuit interactions and developing designs in light of these constraints. He recently completed his PhD at the UK’s Synthetic Biology Doctoral Training Centre where he developed models of genetic architectures to implement feedback control mechanisms to manage translational resource limitations.

Back to top