Bacterial degrons in synthetic circuits

Abstract

Bacterial proteases are a promising post-translational regulation strategy in synthetic circuits because they recognize specific amino acid degradation tags (degrons) that can be fine-tuned to modulate the degradation levels of tagged proteins. For this reason, recent efforts have been made in the search for new degrons. Here we review the up-to-date applications of degradation tags for circuit engineering in bacteria. In particular, we pay special attention to the effects of degradation bottlenecks in synthetic oscillators and introduce mathematical approaches to study queueing that enable the quantitative modelling of proteolytic queues.

Keywords: degradation; oscillatory circuits; proteases; queueing theory.

Figure 1.

Examples of native (a–c) and non-native (d) degrons. (a) Intrinsic degrons can be present anywhere along the protein sequences. Different proteases can recognize these sequences [43]. (b) Ribosomal rescue tags (i.e. SsrA tag) are added at the C-terminal of proteins because of translational errors. The SsrA tag is mainly degraded by the proteolytic complex ClpXP [54]. The degradation of SsrA tagged proteins can be further enhanced by the chaperone SspB, which carries proteins to ClpXP [55]. (c) ClpAPS is the main proteolytic complex that recognizes N-degron tags at the N-terminal of proteins. These specific degrons can be hidden and only exposed by specific enzymes such as chaperones [20]. (d) A common strategy in synthetic biology is to fuse non-native degrons to proteins to reduce their half-life [56]. Left: A protein is unstable because it contains a degron that targets it to be degraded. However, the expression of the non-native protease can lead to protein stabilization because this protease cleaves off the degron sequence. Right: The target protein is stable until the induction of a non-native protease, which then degrades the target protein after recognizing a specific non-native degron.

Figure 2.

(a) Native E. coli SsrA tag and the $\mathrm{DAS}+4\hspace{0.17em}\mathrm{variant}$. The underlying line indicates known binding sites for different E. coli chaperones and proteolytic units in the respective amino acid sequences (SspB: dark blue, ClpX: orange, ClpA: light blue) [81,83]. (b) Schematic representation of the DAS + 4 tagged-based system [83]. In the absence of SspB (e.g. in a ΔSspB mutant), DAS + 4 tagged proteins are stable, and after SspB is expressed the target proteins are degraded by ClpXP. (c) The FENIX system [84] is based on a SsrA/NIa hybrid tag. Active degradation is mediated after recognition of the SsrA tag by ClpXP. When the non-native NIa protease is not expressed, the target protein is stable. When NIa is produced, it cleaves at its recognition site resulting in the removal of the SsrA tag, leading to a stable protein level.

Figure 3.

Proteases can be used to produce localized expression patterns [146]. The TEV cut site is orange and the C-end degron sequence is red. A split TEV protease (blue) is bound to the cell membrane using a PopZ-based polarity system from C. crescentus [147] (magenta). In the presence of the TEV protease, the C-degron is cleaved and the reporter is stabilized. In the absence of the protease, endogenous proteases recognize the C-degron and degrade the reporters. This approach creates patterns of expression in single cells. Figure adapted from [146].

Figure 4.

(a) An oscillatory output can be quantified by its period and amplitude. (b) The design features of a robust oscillator include a delayed negative feedback loop where a repressor (grey) represses all regulatory elements of the circuit. Mathematically, the delayed negative feedback loop is the only essential element in obtaining oscillations. It can also include a positive feedback loop where an activator (yellow) activates all regulatory elements of the circuit. It may also contain a method for rapid protein turnover such as an amino acid degradation tag, which targets proteins to a protease. Proteolytic queues can then form, which can enhance the robustness of the oscillator because the queue can add a consistent time delay to the system. (c) The basic design of two oscillators. Left: the repressilator [5] contains three repressors that repress each other: R1 (LacI), R2 (TetR) and R3 (λCl). Node R1 (LacI) is externally controlled with an IPTG inducer. All repressor proteins are tagged with a SsrA degradation tag (LAA), while the fluorescence reporter (GFP) is tagged with the SsrA variant (AAV). Both LAA and AAV tags are used to target proteins to be degraded by the ClpXP protease. Right: the dual-feedback (DF) oscillator [6] contains a repressor (LacI) that represses all regulatory elements, including itself (negative feedback loop), and an activator (AraC) activates all regulatory elements, including itself (positive feedback loop). Both nodes are controlled with the inducers (IPTG and arabinose). All elements are tagged with the same SsrA-tag (LAA) sequence from E. coli, thus the original repressilator and DF oscillator rely on the formation of proteolytic queues for oscillations.

Figure 5.

(a) Sharing enzymatic resources can lead to coupling events between two independent oscillators. The output of two otherwise independent oscillators can be synchronized via the same degradation tag (ec-SsrA) [11]. Both oscillators show independent oscillations on their own, but when co-expressed in the same host, their period and amplitude synchronize because of the shared enzymatic machinery for degradation (ClpXP). Coupling arises through a proteolytic bottleneck. (b) The Crosstalk Assay [43] allows quantification of the level of crosstalk between two independent degradation tags. The behaviour of two independent fluorescence reporters (CFP and YFP) can be investigated using degradation tags. CFP derivatives are expressed at a constant level. If there is no crosstalk, the induction of YFP can lead to no change in CFP (left and middle). However, if there is crosstalk induction of YFP containing a degradation tag will lead to an increase in CFP (right). This indicates that crosstalk occurs at the protease level because CFP expression is constant. CFP and YFP do not act as transcription factors, and fluorescence crosstalk is only detected when proteins contain degradation tags (right).

Figure 6.

The process of queueing and its corresponding notation. Notations in a queueing process A/S/c/D/K/N are marked in blue in the schematic diagram (top) and are explained in the table (bottom). λ and μ (purple) represent the customer arrival rate and server processing rate, respectively.

Figure 7.

An illustrative model of the Jackson’s network. The servers are denoted by circles. The queue at each server may form due to customers coming from the outside population and customers who finish the service at other servers. That is, once the customers are done with the service at a server, they either choose to leave the system or join the queue of another server.

Figure 8.

A schematic representation of a multi-class proteolytic queueing model, adapted from [181]. Proteins X₁ (yellow) and X₂ (grey) are produced from two independent transcriptional processes, but are being degraded by a common enzyme type (pink). When the enzymes (servers) are fully occupied, a queue with multiple classes of proteins (customers) is formed. Note that proteins join the queue randomly. The queue length depends on the interplay between the enzymatic processing rate and the protein arrival rate to the queue.