Table 2 The advantages and challenges of different regulatory systems

Transcription factors

• The TF systems are extensively studied, and design automation tools^9,23,35 and well-established libraries are available.

• The modular design allows concatenating different modules into complex topologies.

• The programmable TFs can target endogenous pathways.

• The transcriptional circuits show slow kinetics, especially in layered design.

• The transcriptional circuits suffer from retroactivity, burden, and large genetic footprints.

• The layered design is sometimes overcomplicated for simple logic gates.

• Sensor

• Genetic “wire” to connect different modules and tune signal levels

• Genetic controller

• Layered logic

• Dynamic control

Recombinase

• The low basal level produces tight OFF states and digital response.

• Recombinase-based circuits are genetically compact.

• Irreversible recombination is suitable for memory and cascade.

• Design automation tools^15,42,137 and well-established libraries are available.

• The use of repetitive recognition sites may decrease recombination specificity and genetic stability¹⁵.

• Some recombinase recognition sites show cryptic promoter or terminator activities¹⁵.

• The recombinase-based circuits show slow kinetics.

• Digitizer

• Recording and memory

• State machine (synthetic differentiation)

• Single-layer DNA-level logic

Plasmid copy number (PCN) control

• Changing PCN can simultaneously alter the expression of all genes encoded in the plasmid.

• Reduced PCN is coupled with cell viability under antibiotic selection.

• The kinetics is slow, constrained by cell division.

• The number of plasmids used in the same cells is restricted by plasmid incompatibility.

• Global gene expression control

• Cell viability control (kill switch)

Riboregulator

• The RNA base-pairing is programmable and predictable, affording in silico design, prediction, and large orthogonal libraries.

• The de novo-designed assembly of multiple RNA strands is suitable for multi-input signal processing.

• The RNA circuits have small genetic footprints, fast kinetics, and low metabolic load.

• Riboregulators can respond to endogenous RNAs.

• The riboregulators usually require high expression levels of trans-acting RNAs.

• Promiscuous interactions with host transcriptomes potentially affect bacterial growth¹⁰⁴.

• Promiscuous interactions with surrounding genetic contexts (e.g., insulator¹⁴¹) affect modularity.

• Some riboregulators change the protein sequences of genes of interest.

• Endogenous RNA sensor

• Post-transcriptional multi-input logic

• Dynamic control

Riboswitch & Ribozyme

• Riboswitches and ribozymes are functional across different host organisms.

• Ribozymes are amenable to aptamer insertion, allowing external regulation.

• Computational tools to design riboswitches are available.

• Limited parts are available for RNA aptamers, and identifying new aptamers tends to be hard.

• The dynamic ranges of riboswitches are usually low.

• Sensor

• Dynamic control

RNA inference (miRNA & asRNA) & RNA binding protein

• The circuits encoding these modalities could be delivered by RNA and regulated by external inducers and endogenous miRNA biomarkers

• The RNA-binding proteins could amplify post-transcriptional signals and interface with protein-level regulation

• The function of miRNAs and asRNAs involves endogenous machinery, potentially causing queuing effect and burden.

• The gene activation mechanisms are few; in most cases the gene expression can only be repressed.

• The lifetime of RNA-delivered circuits is relatively short.

• Endogenous miRNA sensor (cell classifier)

• Genetic controller

• Post-transcriptional logic based on NOT gates

• Transient gene expression for a short period

Programmable RNA-targeting system

• The sequence-specificity is high and the ADAR-based systems can distinguish dinucleotide variants⁷⁶.

• The RCas system can deliver diverse effector proteins to execute functions more than activation and repression.

• The systems are not fully programmable, as the ADAR system recognizes specific codons in target RNA.

• The ADAR-based translational control can only produce protein signals and produce peptides that may vary in immunogenicity⁷⁷.

• The large size and bacterial origin of the dCas protein hinder its application in eukaryotic systems⁸⁰.

• Endogenous RNA sensor (cell classifier) and editor

Protein regulators

• The protein circuits operate at fast kinetics¹⁰⁰

• The protein interactions are functional across different cellular compartments and various host contexts

• The protein circuits can easily interface with endogenous protein processes¹⁰¹

• The protein regulators can function in RNA-delivered circuits

• Limited parts are available, restricting the scale-up of protein circuits.

• Overloading protein degradation machinery causes queuing effect⁸⁷.

• Protein proteolysis costs relatively high levels of energy like ATP⁸⁷.

• Inserting protein interaction domains to target effectors may disrupt the protein functionality and needs carefully selecting insertion locations.

• Signal transmitter between endogenous signaling pathway and synthetic circuit

• Genetic controller

• Post-translational logic

• Dynamic control