TIBTEC-1223; No. of Pages 10

Review

Synthetic protein switches: design principles and applications Viktor Stein and Kirill Alexandrov The University of Queensland, Institute for Molecular Biosciences, Brisbane, QLD 4072, Australia

Protein switches are ubiquitous in biological signal transduction systems, enabling cells to sense and respond to a variety of molecular queues in a rapid, specific, and integrated fashion. Analogously, tailor-engineered protein switches with custom input and output functions have become invaluable research tools for reporting on distinct physiological states and actuating molecular functions in real time and in situ. Here, we analyze recent progress in constructing protein-based switches while assessing their potential in the assembly of defined signaling motifs. We anticipate such systems will ultimately pave the way towards a new generation of molecular diagnostics and facilitate the construction of artificial signaling systems that operate in parallel to the signaling machinery of a host cell for applications in synthetic biology. Synthetic biology of signal transduction The rational construction of artificial signaling systems is a key goal of synthetic biology. This encompasses all levels of complexity, ranging from proteins to pathways, networks, and, ultimately, organisms, and has application for molecular diagnostics, cell-based biosensors, therapeutics, and industrial biotechnology [1,2]. In addition, a capacity to engineer biological signaling systems with predictable behavior provides ultimate proof to scientific models describing biological processes [1]. Constructing artificial signaling systems has been realized predominantly with synthetic gene circuits, in which rational engineering strategies are supported by the modular organization and function of transcription factors and their DNA response elements [3,4]. Similarly, aptamers and ribozymes have been recombined to create functional nucleic acids that can sense and amplify distinct molecular cues [5] or exert post-transcriptional control on gene expression [6]. However, the limited chemical diversity of nucleic acids compared with amino acids ultimately limits their functionality. Furthermore, transcription-based signaling circuits are inherently slow, with typical response times on the scale of hours [7]. By contrast, protein-based signaling circuits operate orders of magnitude faster and feature diverse enzymatic outputs [7]. However, engineering such Corresponding authors: Stein, V. ([email protected]); Alexandrov, K. ([email protected]). Keywords: synthetic biology; protein switches; protein engineering; molecular diagnostics; signaling; biosensors. 0167-7799/ ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tibtech.2014.11.010

systems has proven challenging, especially at the molecular level, where signaling is controlled by distinct protein switches. These can either be based on allosterically regulated proteins that couple input to output solely through conformational changes or be composed of modular receptors, transducers, and actuators that process molecular cues in a concerted fashion through the induced colocalization of distinct signaling components. The individual designs can range from highly integrated based on structurally intertwined receptors and actuators, where conformational changes are transmitted through networks of residues that are adjacent in tertiary, but not necessarily in primary structure (Figure 1A), to modular, where conformational changes are limited to the linker regions that separate functionally and structurally distinct receptor and actuator domains (Figure 1B), to highly modular, where signaling cues are transmitted through the induced colocalization of molecularly distinct signaling components (Figure 1C). Applications for protein switches are numerous. In diagnostics, protein switches can detect analytes as components of inexpensive homogeneous assays that do not require specialized equipment or time-consuming incubation and washing steps characteristic of immunoassays [8]. Protein switches have also proven invaluable in the quantitative imaging of molecular processes in cells [9]. In addition, protein switches can control the activity of key signaling proteins by non-invasive means, such as with light or biochemically inert ligands, which act orders of magnitude faster than inducible gene expression-based systems. Here, we review recent progress in the construction of protein-based switches for monitoring and actuating molecular and cellular functions while identifying aspects critical for their successful design. Overall, this should greatly facilitate the challenging task of constructing protein-based switches, which has so far proven intractable to computational design methods [10,11] (Box 1). Allosteric fluorescent protein switches Genetically encoded Ca2+ sensors constitute the first generation of protein switches that exploited fluorescent proteins (FPs, see Glossary) for generating a measureable read-out [12,13]. Allosteric binding receptors for Ca2+ were generated by fusing calmodulin (CaM) to a CaM-binding peptide (CaM-BP) derived from the myosin light chain kinase. As CaM binds Ca2+, CaM-BP associates with CaM, causing the receptor to transition from an extended to a compact state. In the original design, this conformational change was detected through distance-dependent Trends in Biotechnology xx (2014) 1–10

1

TIBTEC-1223; No. of Pages 10

Review Glossary Actuator: protein modules that undergo changes in easily detectable biophysical properties, including binding, catalytic activity, or other type of signaling chemistry, such as fluorescence or bioluminescence. Affinity clamps: a class of artificially engineered allosteric binding receptors where a primary protein-binding domain is fused to a secondary enhancer domain that specifically recognizes the ligand-bound conformation of the primary binding domain resulting in the formation of a sandwich complex. Allosteric binding receptor: refers to protein-based receptors that undergo conformational changes upon binding of their target ligand. Alternative frame folding (AFF): refers to a generally applicable procedure for creating allosteric binding receptors and actuators based on competing, partially folded protein fragments that can be selectively stabilized in response to specific molecular queues. Autoinhibitory domains (AI domains): directly bind the active or allosteric site of an enzyme to inhibit catalytic or binding functions. Avena sativa phototropin 1 Lov2 domain (AsLov2): refers to a flavin mononucleotide-containing protein domain that undergoes a reversible conformational change upon illumination, which is exploited by several light-responsive synthetic protein switches. b-Lactamase (BLA): a model reporter enzyme that confers resistance to blactam antibiotics, the activity of which can be readily detected using both colorimetric and fluorescent read-outs. b-Lactamase inhibitory protein (BLIP): a competitive, active site-directed inhibitor of BLA that is used as an AI domain to create allosterically regulated BLA-based protein switches. Bioluminescent resonance energy transfer (BRET): refers to the distancedependent energy transfer between a bioluminescent donor and a fluorescent acceptor. BRET-dependent read-outs are frequently used to resolve bimolecular interaction events or conformational changes in allosterically regulated protein switches with luciferase and FP-tagged binding receptors. Bottom-up design: the assembly of protein switches, signaling motifs, pathways, and networks from well-characterized component parts using rational engineering principles. Calmodulin (CaMs): a family of calcium-dependent allosteric binding receptors that transition from an extended to a compact conformation upon binding CaM-BPs. CaM-binding peptides (CaM-BP): a family of peptides that specifically bind to CaM. Circular protein permutation: a protein engineering procedure where the native N and C termini are fused and new N and C termini are created at a different site, leaving the overall structure intact. Dihydrofolate reductase (DHFR): a model reporter enzyme that catalyzes the conversion of dihydrofolic acid to tetrahydrofolic acid using NADPH as an electron donor. Fibronectin type III (FN3) domain: a class of immunoglobulin-like protein domains that, similar to antibody fragments, can be tailor-engineered to recognize a variety of protein-based targets. Firefly luciferase (FFL): a model reporter enzyme that catalyzes the oxidation of D-luciferin (LH2) to oxyluciferin with the concomitant emission of light. FK506-binding protein (FKBP12): forms part of a model protein–protein interaction module that binds FRB in a rapamycin-dependent manner. FKBP rapamycin-binding protein (FRB): forms part of a model protein–protein interaction module that binds FKBP12 in a rapamycin-dependent manner. Fluorescent proteins (FP): proteins that fluoresce upon illumination with light and are frequently used to visualize bimolecular interaction events or conformational changes in allosteric protein receptors by means of distancedependent FRET between two FPs. Fluorescence resonance energy transfer (FRET): refers to the distancedependent energy transfer between a donor and acceptor fluorophore. FRETdependent read-outs are frequently used to resolve bimolecular interaction events or conformational changes in allosterically regulated protein switches with FP-tagged binding receptors. Guanine nucleotide exchange factors (GEFs): key regulatory proteins that mediate the activation of small GTPases as they catalyze the exchange of GDP for GTP. Luciferase-based indicators of drugs (LUCIDs): a class of semisynthetic small molecule sensors where binding of a target analyte triggers conformational changes in the sensor that is subsequently detected by BRET. Maltose-binding protein (MBP): an allosteric protein binder belonging to the periplasmic binding protein family that undergoes a rigid domain movement upon binding maltose. In association with BLA, MBP has been used to construct various allosteric protein switches by means of domain insertion. Mitogen-activated protein kinases (MAPK): a highly conserved, eukaryotic signaling system that transmits signals from receptors at the plasma membrane through a phosphorylation cascade to the nucleus. Protein fragment complementation assays (PCAs): refer to bimolecular protein–protein interaction assays that transduce a binding event through the assembly of a fully functional reporter enzyme.

2

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

PSD95-Dlg1-Zo1 (PDZ): a family of protein-binding domains that specifically bind distinct C-terminal peptide motifs. Rapamycin: refers to a small molecule drug that mediates association between FKBP12 and FRB. SNAP-Tag: a site-specific protein conjugation system based on engineered mutants of human alkyl guanine transferase (hAGT) that mediate the irreversible conjugation of SNAP-tag fusion proteins with benzyl-guaninelabeled small molecule tags that include fluorophores, affinity handles, and small molecule drugs. SNAP-tag based indicator with a fluorescent intramolecular tether (SNIFIT): a class of semisynthetic small molecule sensors where binding of a target analyte triggers conformational changes in the sensor that is subsequently detected by FRET. Src homology 2 (SH2) domains: a family of protein domains that specifically bind phosphopeptide motifs. Src homology 3 (SH3) domains: a family of protein domains that specifically bind proline-rich peptide motifs. Top-down design: the construction of protein-based switches, signaling motifs, pathways, and networks through the step-wise modification of existing systems. Transducer: generically refers to protein-based binders, enzymes, modular linker elements, and networks of interacting residues that translate a conformational change in the binding of a receptor to an actuator.

changes in the efficiency of fluorescence resonance energy transfer (FRET) between a yellow- (YFP) and cyan-fluorescent protein (CFP) [12]. In a more integrated design, the Ca2+ receptor was inserted internally into GFP, resulting in Ca2+-triggered modulation of GFP fluorescence intensity [13,14]. Iterative improvements over the course of 15 years have yielded highly sensitive calcium sensors that can detect the comparatively small calcium bursts that occur during axon potential firing in neurons [15,16]. Based on these initial blueprints, more than 100 intramolecular FRET sensors have been developed to monitor a variety of molecular queues, ranging from protein–peptide, antibody–epitope, and protein–small molecule interactions to the detection of proteolysis, post-translational modifications, and mechanical properties, such as tension at cell–cell junctions [17–19]. Biophysical studies demonstrated that intramolecular FRET sensors with the largest signal change exploit the natural propensity of FPs to dimerize, which enables them to toggle between two defined molecular states in the presence and absence of a molecular queue. This sets them apart from FPs that do not form such interactions and exist in poorly defined conformational and spectral ensembles [17]. In practice, this requires the construction of linkers that are sufficiently rigid to counteract the dimerization of FPs in the absence of a molecular queue, but sufficiently loose to ensure efficient FRET and folding of the fluorophores. Allosteric fluorescent protein switches based on mutually exclusive binding interactions Given that only a few naturally occurring protein families exhibit conformational changes sufficiently large for sensor construction, allosteric binding receptors frequently have to be engineered. Here, modular design strategies that utilize generic receptor modules and minimize the need to engineer protein-based binders de novo are preferred. Affinity clamps In ‘affinity clamping’, two binders are connected through a flexible linker, which enables them to form a sandwich complex around a target ligand. Proof-of-concept was achieved by using a flexible glycine-serine linker to connect a circularly permuted Erbin PDZ domain to an engineered

TIBTEC-1223; No. of Pages 10

Review

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

Box 1. Challenges of engineering protein switches

Allosteric protein switch (Domain inserted)

(A)

+

L

Integrated

Allosteric coupling

R MB P

R L

A

Ligand Receptor

Actuator

A

Change in funcon: e.g., binding or catalyc, or biophysical property e.g., fluorescence.

Allosteric protein switch (Modular)

(B)

Mutually exclusive binding interacons

R1 L

AI

+

AI

R1 L

A

R2

Autoinhibitory domain

R2

A Linkers

R1

+

T Transducer

L

+

R2

A AI Autoinhibited actuator

Induced colocalizaon

R₂

Proximity protein switch

(C)

R₁ L R₂

T

A AI

Binding, post-translaonal modificaons

Numerous technological advancements have facilitated the construction of protein switches. Notably, affordable DNA synthesis in association with the abundant availability of sequenced genomes from diverse phyla has yielded an unlimited repertoire of biological parts from which binders, linkers, and enzymes can be sourced to construct recombinant sensors, transducers, and actuators. This is complemented by powerful computational design algorithms that, in some cases, could predict protein structures to angstrom resolution, engineer enzyme active site de novo, and introduce protein- and small molecule-binding sites [10,11]. Yet, the rational design of protein switches that respond to specific molecular queues with distinct input and outputs remains challenging. This is primarily due to our limited understanding of the dynamic processes that underlie signaling at the molecular level, such as protein folding and flexibility, reversible binding interactions, and the role of post-translational modifications. Therefore, engineering protein switches typically requires extensive empirical optimization using technically challenging multiplexed screening and selection systems. As a result, most switches have been based on a few model proteins, including GFP, BLA, DHFR, and FFL. However, their utility is limited to reporter functions and, thus, is neither amenable to actuate cellular functions nor compatible with point-of-care readouts. Therefore, generally applicable strategies that allow facile construction of protein-based switches using any combination of receptor, transducing, and actuating domains are desirable.

Modular TRENDS in Biotechnology

Figure 1. Principle designs of protein-based switches increasing in modularity. (A) Integrated designs are generated, for instance, by means of domain insertion where ligand binding-induced conformational changes between a receptor ‘R’ and an actuator ‘A’ are allosterically coupled and transduced through a network of interacting residues that are adjacent in tertiary structure, but not necessarily in the sequence. (B) In modular allosteric receptor designs, ligand binding-dependent conformational changes are confined to the linker regions that connect structurally and functionally autonomous binding receptors ‘R1’ and ‘R2’ with an actuator ‘A’ and its AI domain. (C) In proximity-regulated protein switches, ligand-binding events induce colocalization of molecularly distinct receptors ‘R1’ and ‘R2’, transducers ‘T’, and autoinhibited actuators ‘A’. Activation of the actuator can either occur through binding interactions or post-translational modifications that are mediated by the transducer ‘T’ and displace the AI domain from the active site of the actuator ‘A’.

fibronectin type III (FN3) domain that was evolved to recognize the PDZ domain in its ligand bound form [20,21]. Biophysical studies demonstrated that binding was associated with a large conformational change as the receptor transitioned from a loosely structured, ligandunbound conformation to a sandwich-like, ligand-bound complex with a rigidified linker [21]. This conformational change was sufficiently large to create an intramolecular FRET sensor [22] (Figure 2A). Furthermore, highlighting the general utility of the approach, the repertoire of ‘affinity clamps’ was recently expanded to phosphotyrosine-specific motifs [23]. SNIFITs and LUCIDs Detecting small molecules is considered more challenging, given the limited molecular recognition features that small ligands can contribute towards a conformational change. To this end, a new class of semisynthetic protein switches, termed SNAP-tag based indicator with a fluorescent intramolecular tether (SNIFITs), promises a generic route to fluorescent small molecule sensors [24–27]. SNIFITs feature a protein-based binder tethered to an intramolecular ligand that can be displaced by a target

analyte, which causes the SNIFIT to transition from a compact to an extended conformation. This conformational change is, in turn, resolved by distance-dependent FRET between two synthetic fluorophores that were introduced through highly specific protein conjugation tags based on SNAP-tag technology. In this way, SNIFITs for sulfonamides [24], glutamate [25], g-aminobutyric acid (GABA) [26], and acetylcholine [27] have been developed. Recently, the repertoire of SNIFITs was expanded for the quantitative detection of several clinically important drugs in whole blood using conventional photographic imaging equipment compatible with point-of-care diagnostic devices [28] (Figure 2B). In this new biosensor class, termed luciferase-based indicators of drugs (LUCIDs), one of the site-specific protein conjugation tags is replaced with NanoLuc [29], an engineered luciferase derived from the deep-sea shrimp Oplophorus gracilirostris. Its high physical stability and brightness enable bioluminescent resonance energy transfer (BRET) read-outs with easily recordable and electronically amplifiable signals. A notable design feature of LUCIDs is a rigid 30 amino acid-long poly-proline linker that introduces steric strain to modulate the strength of interaction between the protein-based binder and its intramolecular tether. This potentially provides a generic element for fine-tuning intramolecular binding interactions, especially when conformational changes in the receptor are not effectively transduced to the actuator, but further shortening of the linkers restricts the functional folding of adjacent domains. Allosteric fluorescent protein switches based on partially unfolded protein fragments Beyond the movement of structurally well-defined protein domains, many naturally occurring allosterically regulated protein functions rely on partially unfolded and intrinsically disordered protein fragments [30]. Equivalent synthetic 3

TIBTEC-1223; No. of Pages 10

Review

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

Fluorescence-based biosensors PDZ pepde sensor (Affinity clamps)

(D)

cAbl SH2 Domain Sensor (Intermolecular AFF, FN3)

et

(A)

Fn3 YPet

+

P

P

High FRET

Low FRET

Low FRET

P

CyPet

YPet

PDZ et

+ FN3

FN3 Binding incompetent FN3

Drug sensors (LUCIDs)

(B)

Tether

+

SNAP 30 Pro

Drug

High BRET

Low BRET

T B

Ternary complex assembly

D B NLuc

SNAP

NLuc

SH2 SH2 cAbl kinase SH2

T

D

CyPet

YPet

CyP

PDZ pepde

DZ

TVMV CyPet

High FRET

YP

Fn 3

30 Pro

Drug binder

(E)

Thrombin sensor (Intramolecular AFF, GFP/YFP)

NanoLuc Thrombin

(C)

Calcium sensors (Intramolecular AFF, calbindin)

Tenth strand YFP

YFP

GFP Ca Ca High FRET

+

+ Light photodissociaon

+ High FRET

Low FRET

Tenth strand GFP

TRENDS in Biotechnology

Figure 2. Summary of fluorescence-based allosteric protein switches used in molecular imaging and diagnostics. (A) Affinity clamps comprise a circularly permutated PDZ domain that is fused to a fibronectin type III (FN3)-based enhancer domain that recognizes the PDZ domain in its ligand-bound state. PDZ ligand binding is associated with the formation of a sandwich complex and a large conformational change that can be detected through the distance-dependent fluorescent resonance energy transfer (FRET) between the two fluorescent proteins YPet and CyPet. (B) Luciferase-based indicators of drugs (LUCIDs) are based on intramolecular tethers that compete with a target analyte for the same binding protein while the resulting conformational changes are detected by the distance-dependent bioluminescent resonance energy transfer (BRET) between NanoLuc and the Cy3 fluorophore conjugated to SNAP-tag. (C) Calcium sensors generated by alternative frame folding (AFF) of calbindin where the circular permutated form was stabilized by Ca2+ ions while the associated conformational change was detected by distance-dependent FRET of two chemically conjugated fluorophores. (D) Intermolecular allosteric receptors for the cAbl kinase Src homology 2 (SH2) domain were generated by mutating an FN3-based binder such that binding to the SH2 domain was disrupted, but readily restored through the functional complementation of the duplicated binding competent fragment in trans. Destabilizing packing mutations in the FN3 core domain ensured efficient exchange between the binding incompetent and competent fragments. (E) Protease-responsive fluorescent ratiometric sensors were generated by duplicating the tenth strand of GFP while a point mutation in the native tenth C-terminal strand changed the fluorescent output from green to yellow. Depending which of the two strands associated with the fluorescent protein (FP) core, the resulting FP can either emit yellow or green fluorescent light; photo-assisted, thrombin-mediated cleavage of the green fluorescence-emitting strand results in a permanent yellow fluorescent protein.

protein switches can be artificially engineered by means of alternative frame folding (AFF). Here, a portion of a protein that makes contact with a desired target ligand is duplicated and fused to the opposite end of the molecule, yielding a continuous polypeptide that equilibrates between native and circularly permuted conformations (Box 2) [31] while leaving parts of the protein unfolded. Proteins are turned into allosteric binding receptors by point mutations that selectively stabilize the circularly permuted conformation while destabilizing the native conformation upon ligand binding (Figure 2C). Both calbindin and a ribose-binding protein (RBP) were converted by AFF into Ca2+- and ribosespecific allosteric receptors, while ligand-induced conformational changes in the two receptors were detected through distance-dependent intramolecular FRET between two chemically conjugated fluorophores [32,33]. Recently, AFF was used to convert an FN3-based binder specific to the Src homology 2 (SH2) domain of cAbl kinase into an allosterically regulated, intermolecular fluorescent switch (Figure 2D). Binding to cAbl kinase co-operatively stabilized the assembly of the FN3-based binder from two 4

partial fragments, and was resolved by intermolecular FRET between YFP and CFP [34]. This promises to be a generic method for converting globular binding scaffolds into allosteric fluorescent sensors. Genetically encoded, intramolecular FRET sensors incorporating FPs have yet to be developed, but their construction could be complicated by the fact that AFF-based allosteric receptors comprise partially unfolded proteins, which have loosely defined molecular states [32,33]. AFF was also used to engineer allosterically regulated actuators including a zymogen of barnase [35] and a protease-sensitive ratiometric GFP sensor [36]. The latter was constructed by duplicating the tenth C-terminal strand of GFP at the N terminus while separating it with a thrombin cleavage site; in addition, a point mutation was introduced in the native tenth C-terminal strand shifting the emission spectrum from green to yellow (Figure 2E). As a result, two duplicated strands that conferred yellow and green fluorescence competed for the same binding interactions with the GFP core; photo-assisted proteolytic cleavage of the native N-terminal strand subsequently enabled the

TIBTEC-1223; No. of Pages 10

Review Box 2. Circular protein permutation In circular protein permutation, the topology of a polypeptide chain is altered as the native N and C termini of a protein are covalently linked while new termini are generated at different sites [31]. This does not modify the protein fold, but changes the order of residues in which they appear and, thus, influences the folding trajectory of a protein. Structural studies show that overall structures of circularly permutated and nonpermutated proteins are similar, but can display pronounced conformational changes in the vicinity of the native and new termini. This can favorably influence several biophysical properties, such as thermostability, catalytic efficiency, and substrate specificity. Circular protein permutation is frequently used in the construction of protein-based switches: for instance, protein switches generated by AFF feature an equilibrium between a circularly permutated and a nonpermutated conformation, one of which is stabilized upon binding a target analyte [32,33]. Circular protein permutation can also be used to relocate termini into allosteric hotspots, which forms the basis of constructing allosteric protein switches by means of domain insertion [38,48,49]. Circular protein permutation can also be used to move the termini of a protein closer to the active site or the binding interface, shortening intramolecular distances and increasing effective concentrations between an AI domain and the active site of an enzyme [55] or between a primary protein–peptide and a secondary enhancing interaction, as required in ‘affinity clamping’ [20,23]. However, due to the drastic change in folding trajectory, the outcome of circular protein permutation is generally unpredictable and, thus, usually requires extensive empirical testing. Here, naturally occurring sets of circularly permutated protein families may provide leads as to which families are amenable to circular protein permutation and which are not. Proteins that are amenable to circular permutation can usually be split and, thus, adopted in PCA. Arguably, this is because subfragments define sufficiently selfcontained structural domains that can fold independently. A prominent example includes DHFR, which has been used in PCAbased split-protein sensors [51], is amenable to circular permutation [47], and has been utilized to create allosterically regulated enzymes by domain insertion [48,49].

C-terminal strand to associate permanently, resulting in a >2000-fold ratiometric dynamic fluorescence change. It is easily conceivable that the thrombin cleavage site could be replaced with allosteric binding receptors that stabilize one strand more than another in the presence of a target analyte. Such a design could lead to a new category of ratiometric fluorescent sensors with properties superior to those of intramolecular FRET sensors. Allosterically regulated enzymes for imaging, diagnostic, and therapeutic applications While fluorescent sensors provide excellent spatiotemporal resolution in cell culture, their sensitivity and compatibility with point-of-care diagnostics and molecular imaging of multicellular organisms is limited. In this regard, enzymebased protein switches are considered superior due to their ability to amplify a biomolecular signal. Integrated designs by domain insertion Domain insertion is a popular strategy for engineering allosterically regulated enzymes. Proof-of-concept was achieved by inserting a circularly permuted library of b-lactamase (BLA) into maltose-binding protein (MBP) (Figure 3A) [37,38]. Potent maltose-inducible BLA switches with >600-fold induction ratios were identified in a highthroughput screen for maltose-dependent resistance to

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

ampicillin in Escherichia coli [38]. To expand the utility of potential target ligands from sugars to peptides, an allosterically regulated mutant of BLA was engineered by inserting CaM at a split site that was previously used in BLA-based protein fragment complementation assays (PCA) [39]. In this way, the conformational change in CaM as it transitioned from a dumbbell to a compact state upon binding its cognate peptide, modulated the association of the split-BLA halves and, thus, enabled different CaMBPs to induce BLA activity >100-fold. In a similar fashion, a bioluminescent sensor for Akt kinase activity was constructed by fusing an Akt consensus substrate to a promiscuous phosphopeptide-binding domain, FHA2, and inserting it into the hinge region that separates the N- and C-terminal domains of firefly luciferase (FFL) [40] (Figure 3B). Upon phosphorylation, the binding receptor separating the N- and C-terminal domains transitioned from a flexible to a constrained conformation, thus reducing the luciferase activity. There was a modest three times decrease in the bioluminescent reaction, but it was sufficient to monitor Akt-dependent phosphorylation activity in mice. While this design is commonly referred to as a split-luciferase, it is noteworthy that the N- and C-terminal domains are structurally autonomous, although each contributes key catalytic residues to separate half-reactions. The N-terminal domain charges the substrate firefly D-luciferin (LH2) with AMP, and the C-terminal domain catalyzes oxidation of LH2 to oxyluciferin, resulting in bioluminescence [41]. Therefore, the way that the split-luciferase protein switches generate a signal is more akin to colocalized substrate channeling than assembling a fully functional enzyme. This feature has recently been exploited in a bimolecular in vitro assay based on the rapamycin induced colocalization and functional complementation of two partially functional FFLs tagged with FKBP12 and FRB [42,43] with superior shelf lives over purified split-FFL halves [44]. Domain insertion was also used to convert cytosine deaminase into an allosterically regulated enzyme therapeutic that, in the presence of hypoxia-inducible factor 1a (HIF-1), rendered cancer cells ten times more sensitive to the cytotoxic pro-drug 5-fluorocytosine [45]. Such designs could eventually be used as combined gene- and chemotherapeutic modalities for treating cancer with reduced toxicity. A recurring issue in the field is the extent to which domain insertion could constitute a generally applicable strategy for engineering allosteric protein switches. The target protein needs to be sufficiently stable to tolerate insertion of an unrelated protein domain. Tolerance to insertion is largely unpredictable and needs to be tested empirically. Additionally, insertion sites in the receptor and the actuator need to be allosterically coupled. A binding event in the receptor needs to be transduced through the inserted domain, by a series of conformational changes, to the active site of an actuator. To this end, statistical coupling analysis (SCA) can identify coevolving residue networks that mediate allosteric coupling [46], while systematic circular permutation studies can point out a suitable insertion site [47]. In one example, SCA successfully predicted allosteric hotspots in dihydrofolate reductase 5

TIBTEC-1223; No. of Pages 10

Review

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

Enzyme-based biosensors (A)

Maltose sensors (BLA-MBP)

Phosphorylaon sensor (Split-firefly luciferase)

(B)

Maltose binding protein

+

M

MB P MB P

Phospho-pepde binding domain

LH2

MBP M BLA

BLA

N LH2-AMP

N N-terminal domain FFL N

FHA2

C

C-terminal domain FFL

Phosphorylaon FHA2 PO

Maltose

β-lactamase

(C)

C

Oxyluciferin

Monovalent anbody sensors (BLA-BLIP)

Bivalent anbody sensors (BLA-BLIP)

(D)

β-lactamase inhibitory protein BLIP

BLIP BLA

+

BLA

Single stranded DNA sensor (BLA-BLIP) 3′

+

BLIP

5′ ssDNA

Fn 3 3′ 5′

BLA

P

5′3′

Fn3 AI

+

PDZ pepde

BLA 5′ 3′

BLA

Receptor + amplificaon cascade (Autoinhibited protease)

(F) BLIP

(E)

BLIP

BLA

BLIP

+

AI

Autoinhibitory domain

TVMV

TVMV protease

P

TVMV

PD

PDZ

Z SH3

3′ 5′

SH3

PPI proximity sensors (Autoinhibited protease)

(G)

FRB

+

HCV

AI

P

R

FKBP

HCV

TVMV

AI

FRB R

P

AI

FKBP

Proteolyc acvaon

AI

FRB R HCV

TVMV

P

+

P

HCV protease

FKBP

HCV

Proteolyc acvaon

AI

SH3 pepde

AI

TVMV

P

P HCV

P

AI TRENDS in Biotechnology

Figure 3. Summary of enzyme-based allosteric protein switches used in molecular imaging and diagnostics. (A) Maltose sensors were generated by inserting maltosebinding protein (MBP) into a circularly permutated version of b-lactamase (BLA) such that the rigid domain movement upon binding maltose in MBP modulates the activity of BLA. (B) Phosphorylation sensors for Akt kinase activity were generated by fusing a consensus substrate of Akt kinase to FHA2, a promiscuous phospho-binding domain. Upon phosphorylation, the receptor transitioned from a flexible to a constrained conformation separating the N- and C-terminal domains of firefly luciferase (FFL) and bringing them into a catalytically inactive conformation. (C–E) Allosterically regulated BLA switches based on BLA inhibitory protein (BLIP) with monovalent (C) and bivalent (D) versions for detecting antibodies or versions for detecting single stranded DNA (E) through mutually exclusive binding interactions. (F) Modular, allosteric protease receptors generated by recombining Tobacco Vein Mottling Virus (TVMV)-based autoinhibited protease modules with an ‘affinity clamp’ receptor. Binding of a ligand stabilizes the receptors in an open conformation based on mutually exclusive binding interactions, while the signal generated by the TVMV-based allosteric receptor could be further amplified through a Hepatitis C Virus (HCV)-based amplifier. (G) Protease-based proximity sensors of specific protein–protein interactions (PPIs) were generated by fusing the TVMV- and HCV-based signal transducer to FK506-binding protein (FKBP12) and FKBP rapamycin-binding protein (FRB), which colocalize in the presence of rapamycin to cause cleavage and activation.

(DHFR), AsLov2, and PDZ to construct light- and PDZ peptide-regulated mutants of DHFR [48,49]. Furthermore, the insertion site in DHFR of E. coli identified by SCA was distinct from the insertion site of estrogen-a into FKBP12, which was used to create DHFR-dependent small molecule sensors in yeast [50], and the split-sites used in DHFRbased PCAs [51]. Modular designs by autoinhibition Considering the unpredictable effect of domain insertion on protein structure and fold, modular strategies to engineer allosterically regulated enzymes are generally preferred. In one recent example, BLA was fused to blactamase inhibitory protein (BLIP), while allosteric receptors in the connecting linker regulated BLIPdependent inhibition of BLA following binding of the target analyte. In this way, allosterically regulated protein 6

switches were generated to recognize clinically important proteases and antibodies as well as single-stranded DNA (Figure 3C–E) [52–55]. Antibodies were recognized either by monovalent or bivalent peptide-based epitopes. Interestingly, a bivalent antibody-specific receptor previously developed for an intramolecular FRET sensor [56] was transferred to the BLA-BLIP system with only minor optimization of affinities [52]. This highlights the advantages of modular design strategies that exploit structurally and functionally distinct autoinhibitory (AI) domains. However, BLA-based reporter assays are limited to colorimetric and fluorescent read-outs and, thus, are neither easily integrated with point-of-care diagnostics nor able to actuate cellular functions. Thus, in an effort to expand the repertoire of signaling chemistries, a signaling toolbox was developed featuring artificially autoinhibited proteases as elementary signaling

TIBTEC-1223; No. of Pages 10

Review modules [57]. Proteases were sourced from highly specific viral proteases from Tobacco Vein Mottling Virus (TVMV) and Hepatitis C Virus (HCV) and AI domains were artificially engineered using product-based competitive inhibitors as leads. Structure-guided design and directed protein evolution were used to generate different types of proteasebased signal transducers and allosteric receptors that can be activated following site-specific proteolysis or binding of a PDZ ligand peptide to an ‘affinity clamp’ (Figure 3F). Notably, the activity of the allosteric protease receptors could be induced from fivefold switch-OFF to 37-fold switch-ON, solely depending on the length of the connecting linker. Furthermore, protease-based allosteric receptors and transducers could be assembled in an integrated signal sensing and amplification circuit improving absolute signals between one and two orders of magnitude. In addition, protease-based proximity sensors could be generated by fusing the TVMV- and HCV-based signal transducers to FKBP12 and FRB modules, such that the HCV-based signal transducer was activated by TVMV upon rapamycininduced colocalization, and the dynamic range of activation was fine-tuned through partially autoinhibitory interactions (Figure 3G). Considering the ease with which artificial zymogens can be constructed, protease-based signaling cascades can in principle be connected to any read-out and, thus, become a useful tool in analytical and diagnostic applications. Allosteric protein switches as actuators of cellular functions Beyond analytical and diagnostic applications, allosteric protein switches are being increasingly used as research tools to actuate and perturb cellular signaling systems. Considering their response rates are orders of magnitude faster than inducible gene expression systems or RNAimediated knockdowns, this makes them suitable for the analysis of rapid cellular processes. Light-dependent protein switches Light-responsive proteins have revolutionized cell biological research, because they enable control of protein function with unprecedented spatiotemporal resolution [58]. Most light-responsive allosteric protein switches are based on the Lov2 domain of Avena Sativa phototropin 1, termed AsLov2. This contains a flavin chromophore that forms a covalent cysteine bond with the protein core upon illumination at 458 nm, which in turn induces unwinding of the C-terminal helix (Ja) of AsLov2. AsLov2-dependent switches are typically constructed by fusing a desired target protein to the C terminus of Ja, which then controls its association with secondary effector proteins. Proof-ofconcept was first realized with the small GTPase Rac1, for which light-dependent conformational changes controlled access of Rac1 to its effector protein kinase PAK and, thus, cell motility [59] (Figure 4A). Similar types of light-inducible protein switch have since been constructed to activate kinase inhibitors [60], control membrane and promoter localization through PDZ affinity-clamp interactions [61,62], control ion channel permeability [63], regulate gene expression and proteosomal degradation in E. coli through ipaA–vinculin and SsrA–SspB interactions

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

[64] and control nuclear localization [65]. In the latter two cases, Ja of AsLov2 was engineered such that the molecular specificity feature-binding of the AsLov2 core overlapped with effector binding. allosterically regulated protein Light-inducible switches were also generated based on the intramolecular reversible association of FP Dronpa dimers that could block the effector-binding interface of a Cdc42-specific guanine nucleotide exchange factor (GEF) and substrate binding to HCV protease by steric bulk [66] (Figure 4B). In addition, light-dependent protein interaction switches were generated using the Cry2-CIBN and PhyB-PIF photo-inducible dimerization modules to control the activity of small GTPases [67,68] and mitogen-activated protein kinase (MAPK) signaling [69] through the induced colocalization of key signaling proteins. Rapamycin-inducible protein kinase switches Although light confers excellent spatiotemporal control in cell culture, its use is difficult in nontransparent animals, for which pharmacological ligands must be used instead. In addition, pharmacological ligands enable synchronized control of protein functions in cell populations on the timescale of seconds or longer. In one recent example, synthetic control was semirationally engineered into focal adhesion kinase (FAK) by inserting a truncated mutant of FKBP12, termed iFKBP12, with juxtaposed N and C termini into a conserved loop near the kinase active site [70]. Upon formation of the rapamycin iFKBP12–FRB complex, the loop rigidifies and concomitantly increases kinase activity, which controls cellular morphogenesis. Similar constructs were engineered for Src and p38 [70] as well as Fyn, Src, Yes, and Lyn protein kinases [71]. Furthermore, FRB can be directly fused to a kinase substrate, in which case rapamycin controls the formation of the enzyme–substrate complex, in addition to kinase function [72]. In a simplified system, termed UniRap, a single-chain rapamycin-binding module based on intertwined FKBP12– FRB domains was developed to control Src kinase activity (Figure 4C) and induce the formation of epidermal cell protrusions in zebrafish [73]. In addition, a rapamycin-inducible mutant of protein kinase A (PKA) was generated based on split-inteinmediated excision [74] of an active site-directed competitive inhibitor of PKA (Figure 4D) [75]. However, proof-ofprinciple is limited to in vitro applications. Towards the bottom-up design of autonomously operating signaling systems A key goal of synthetic biology is to create tailor-engineered programmable cells that autonomously sense, process, and respond to distinct molecular cues. This has been realized predominantly through top-down designs; for instance, by artificially rewiring key signaling nodes of modularly organized signaling pathways. In the simplest case, ectopic overexpression of G protein-coupled receptors (GPCR) was sufficient to reprogram input control of GPCR signaling and yield various designer cell-based therapeutics [76– 79]. Alternatively, key regulatory and catalytic domains in the yeast mating pathway were interchanged to improve mating efficiency [80], and artificial protein–protein 7

TIBTEC-1223; No. of Pages 10

Review

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

Cellular protein switches Light-inducible PPIs (AsLov2)

(A)

Light

+

Effector

AsLov2

Light-inducible GEFs (FP Dronpa)

(B) Effector recruitment

Effector

Rac 1 GTP

+

Dronpa FP FP

Dronpa

Synthec intersecn GEF

Jα

Rapamycin-inducible protein kinase (UniRapR)

(C)

Dronpa FP

DH PH

AsLov2

Rac1 GTP

Cdc42 GTP

Light

Cdc42 GDP

Dronpa

DH PH

Small GTPase recruitment + acvaon

Rapamycin-inducible protein kinase (Autoinhibited)

(D)

Rapamycin

FR B

+

R

Src kinase

Src

FKBP

IntN IntC

R FRB

PKA

Kinase inhibitor excision

FKBP

R FRB

IntN IntC

(F)

+ PKA

Protein kinase A

Phosphorylaon-inducible GEFs

AI

MBP

MBP

PKA

IntC

AI

AI

MBP Acve site structuring

R FRB

FKBP

+

IntN

Src

Rapamycin

(E)

R

FKBP

FRB R FKBP

Phosphorylaon inducible split-TEV protease

Small GTPase recruitment + acvaon

P Cdc42 GDP

+

DH PH

PDZ

P

DH PH

Yki

Yki

Cdc42 GTP

Phosphorylaon

14.3.3

PDZ

N-TEV

N

+

Phosphorylaon

N C

C C

C-TEV

14.3.3

TEV protease assembly

Synthec intersecn GEF

TRENDS in Biotechnology

Figure 4. Summary of allosteric protein switches that can sense or actuate cellular functions. (A) Avena sativa phototropin 1 Lov2 domain (AsLov2)-dependent protein switches are based on the light-dependent unwinding of the C-terminal Ja-helix, which can modulate binding interactions of effector proteins attached at the C terminus. (B) Light-inducible allosterically regulated guanine nucleotide exchange factors (GEFs) were generated through recombination with mutant Dronpa fluorescent proteins (FPs) that, upon illumination, dimerized and sterically blocked recruitment and, thus, activation of its small GTPase effector. (C) Rapamycin-inducible protein kinase (UniRapR)dependent protein switches rely on an artificially engineered rapamycin-binding module inserted into the active site loop of a protein kinase. Upon rapamycin binding, the active site loop structures with a concomitant increase in kinase activity. (D) Rapamycin-inducible split-inteins can permanently excise an autoinhibitory domain from protein kinase A (PKA). (E) Allosterically regulated GEFs generated through recombination with phosphorylation-responsive peptide-PDZ autoinhibitory interactions. Upon phosphorylation of the PDZ peptide, inhibition is relieved, causing the complex to transition from a constrained to an extended conformation, enabling synthetic GEFs to interact with their small GTPase substrate proteins. (F) Split-TEV protease mutants can be selectively assembled following phosphorylation-dependent protein–protein interactions between Yki and 14.3.3.

interactions (PPIs) were sufficient to reprogram input and/ or output control of the MAPK pathway [81] and alter its signaling dynamics [82]. Synthetic PPIs were also used to reprogram two-component signaling pathways in E. coli, while artificial AI interactions reduced cross-activation of related two-component signaling pathways in the crowded environment of the bacterial plasma membrane [83]. Furthermore, the activity of the small GTPase Cdc42 was controlled through allosterically regulated GEFs, which were designed based on phosphorylation-sensitive peptide–PDZ autoinhibitory interactions (Figure 4E) [84]. While rewiring studies highlight the potential of PPIs in reprogramming cellular behavior and shed light on the functional plasticity of naturally occurring signaling networks, a good mechanistic understanding of the underlying signaling networks is usually required. In addition, the majority of these circuits rely on transcriptional outputs because the construction of protein-based actuators that can be interfaced with the most prevalent intracellular signaling chemistries, notably phosphorylation, remains challenging. Ultimately, bottom-up assembly strategies using wellcharacterized, genetically encoded protein-based switches that can be easily interfaced with defined cellular functions 8

are desired. In particular, there is a need to develop enzyme-based signal transducing and actuating systems and expand the repertoire of synthetic protein-based parts beyond orthogonal PPI modules [85,86]. Highly specific viral proteases that have naturally evolved to operate in the complex environment of the cytosol may be well suited for this purpose. Protease-based switches have been developed that mediate biomolecular signals through the induced colocalization of constitutively active proteases [87,88] or the assembly of split-protease mutants [89–95] in response to ligand binding, receptor dimerization, and post-translational modifications (Figure 4F). In addition, protease-based signals can be actuated through diverse mechanisms ranging from apoptosis, transcription, cleavage of a reporter protein, and protein degradation [96,97]. The latter can effectively reverse the proteolytic signal, which can limit noise by introducing negative feedback, and potentially enable the creation of more complex network motifs beyond linear signal amplification cascades [98]. Concluding remarks and future perspectives While the rational engineering of protein-based switches has yet to be fully developed, emerging empirical rules

TIBTEC-1223; No. of Pages 10

Review Box 3. Outstanding questions  How can we build repositories of protein-based parts that can be readily assembled into protein-based switches?  Which classes of protein-based binders, naturally occurring or artificially engineered, are best suited as allosteric binding receptors and protein interaction modules?  How can we expand the repertoire of autoinhibited enzyme modules with signaling chemistries and catalytic functions that are compatible with point-of-care diagnostic read-outs or readily interfaced with cellular functions?  How do structural features in linkers transduce conformational changes between an allosteric binding receptor and an autoinhibited enzyme module?  How do we identify insertion sites in protein-based receptors and enzyme-based actuators that tolerate sufficiently large domain insertions and are amenable to allosteric coupling?

facilitate the construction of tailor-engineered proteins with custom input and output parameters. Both in the context of molecular diagnostics and cellular signaling, the future trend will be to assemble separately engineered protein-based sensors, transducer and actuators into autonomous signaling motifs that operate both independently and in parallel to their environment (Box 3). Acknowledgments This work was supported in part by ARC DP grant DP1094080, ARC FF FT0991611, and NHMRC project grant 569652 and NHMRC program grant APP1037320 to K.A. This study was funded in part by Movember and Prostate Cancer Foundation of Australia’s Research Program.

Disclaimer statement V.S. and K.A. are co-inventors on a Patent Cooperation Treaty (PCT) patent application that covers aspects of the protease-based biosensor technology discussed in this publication.

References 1 Lim, W.A. (2010) Designing customized cell signalling circuits. Nat. Rev. Mol. Cell Biol. 11, 393–403 2 Lienert, F. et al. (2014) Synthetic biology in mammalian cells: next generation research tools and therapeutics. Nat. Rev. Mol. Cell Biol. 15, 95–107 3 Slusarczyk, A.L. et al. (2012) Foundations for the design and implementation of synthetic genetic circuits. Nat. Rev. Genet. 13, 406–420 4 Nielsen, A.A. et al. (2013) Advances in genetic circuit design: novel biochemistries, deep part mining, and precision gene expression. Curr. Opin. Chem. Biol. 17, 878–892 5 Jung, C. and Ellington, A.D. (2014) Diagnostic applications of nucleic acid circuits. Acc. Chem. Res. 47, 1825–1835 6 Benenson, Y. (2012) Synthetic biology with RNA: progress report. Curr. Opin. Chem. Biol. 16, 278–284 7 Olson, E.J. and Tabor, J.J. (2012) Post-translational tools expand the scope of synthetic biology. Curr. Opin. Chem. Biol. 16, 300–306 8 Vallee-Belisle, A. and Plaxco, K.W. (2010) Structure-switching biosensors: inspired by Nature. Curr. Opin. Struct. Biol. 20, 518–526 9 Mehta, S. and Zhang, J. (2011) Reporting from the field: genetically encoded fluorescent reporters uncover signaling dynamics in living biological systems. Annu. Rev. Biochem. 80, 375–401 10 Khoury, G.A. et al. (2014) Protein folding and de novo protein design for biotechnological applications. Trends Biotechnol. 32, 99–109 11 Feldmeier, K. and Hocker, B. (2013) Computational protein design of ligand binding and catalysis. Curr. Opin. Chem. Biol. 17, 929–933 12 Miyawaki, A. et al. (1997) Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388, 882–887 13 Nagai, T. et al. (2001) Circularly permuted green fluorescent proteins engineered to sense Ca2+. Proc. Natl. Acad. Sci. U.S.A. 98, 3197–3202 14 Tian, L. et al. (2009) Imaging neural activity in worms, flies and mice with improved GCaMP calcium indicators. Nat. Methods 6, 875–881

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

15 Chen, T.W. et al. (2013) Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499, 295–300 16 Akerboom, J. et al. (2012) Optimization of a GCaMP calcium indicator for neural activity imaging. J. Neurosci. 32, 13819–13840 17 Merkx, M. et al. (2013) Rational design of FRET sensor proteins based on mutually exclusive domain interactions. Biochem. Soc. Trans. 41, 1201–1205 18 Palmer, A.E. et al. (2011) Design and application of genetically encoded biosensors. Trends Biotechnol. 29, 144–152 19 Constantinou, A. and Polizzi, K.M. (2013) Opportunities for bioprocess monitoring using FRET biosensors. Biochem. Soc. Trans. 41, 1146– 1151 20 Huang, J. et al. (2008) Design of protein function leaps by directed domain interface evolution. Proc. Natl. Acad. Sci. U.S.A. 105, 6578– 6583 21 Huang, J. et al. (2009) Structural basis for exquisite specificity of affinity clamps, synthetic binding proteins generated through directed domain-interface evolution. J. Mol. Biol. 392, 1221–1231 22 Huang, J. and Koide, S. (2010) Rational conversion of affinity reagents into label-free sensors for Peptide motifs by designed allostery. ACS Chem. Biol. 5, 273–277 23 Yasui, N. et al. (2014) Directed network wiring identifies a key protein interaction in embryonic stem cell differentiation. Mol. Cell 54, 1034–1041 24 Brun, M.A. et al. (2011) Semisynthesis of fluorescent metabolite sensors on cell surfaces. J. Am. Chem. Soc. 133, 16235–16242 25 Brun, M.A. et al. (2012) A semisynthetic fluorescent sensor protein for glutamate. J. Am. Chem. Soc. 134, 7676–7678 26 Masharina, A. et al. (2012) A fluorescent sensor for GABA and synthetic GABA(B) receptor ligands. J. Am. Chem. Soc. 134, 19026–19034 27 Schena, A. and Johnsson, K. (2014) Sensing acetylcholine and anticholinesterase compounds. Angew. Chem. 53, 1302–1305 28 Griss, R. et al. (2014) Bioluminescent sensor proteins for point-of-care therapeutic drug monitoring. Nat. Chem. Biol. 10, 598–603 29 Hall, M.P. et al. (2012) Engineered luciferase reporter from a deep sea shrimp utilizing a novel imidazopyrazinone substrate. ACS Chem. Biol. 7, 1848–1857 30 Motlagh, H.N. et al. (2014) The ensemble nature of allostery. Nature 508, 331–339 31 Yu, Y. and Lutz, S. (2011) Circular permutation: a different way to engineer enzyme structure and function. Trends Biotechnol. 29, 18–25 32 Stratton, M.M. et al. (2008) A Ca2+-sensing molecular switch based on alternate frame protein folding. ACS Chem. Biol. 3, 723–732 33 Ha, J.H. et al. (2013) Stepwise conversion of a binding protein to a fluorescent switch: application to Thermoanaerobacter tengcongensis ribose binding protein. Biochemistry 52, 600–612 34 Zheng, H. et al. (2014) Converting a binding protein into a biosensing conformational switch using protein fragment exchange. Biochemistry 53, 5505–5514 35 Mitrea, D.M. et al. (2010) Engineering an artificial zymogen by alternate frame protein folding. Proc. Natl. Acad. Sci. U.S.A. 107, 2824–2829 36 Do, K. and Boxer, S.G. (2013) GFP variants with alternative betastrands and their application as light-driven protease sensors: a tale of two tails. J. Am. Chem. Soc. 135, 10226–10229 37 Guntas, G. et al. (2004) A molecular switch created by in vitro recombination of nonhomologous genes. Chem. Biol. 11, 1483–1487 38 Guntas, G. et al. (2005) Directed evolution of protein switches and their application to the creation of ligand-binding proteins. Proc. Natl. Acad. Sci. U.S.A. 102, 11224–11229 39 Meister, G.E. and Joshi, N.S. (2013) An engineered calmodulin-based allosteric switch for peptide biosensing. Chembiochem 14, 1460–1467 40 Zhang, L. et al. (2007) Molecular imaging of Akt kinase activity. Nat. Med. 13, 1114–1119 41 Sundlov, J.A. et al. (2012) Crystal structure of firefly luciferase in a second catalytic conformation supports a domain alternation mechanism. Biochemistry 51, 6493–6495 42 Ohmuro-Matsuyama, Y. et al. (2014) Improved protein-protein interaction assay FlimPIA by the entrapment of luciferase conformation. Anal. Chem. 86, 2013–2018 43 Ohmuro-Matsuyama, Y. et al. (2013) A protein-protein interaction assay based on the functional complementation of mutant firefly luciferases. Anal. Chem. 85, 7935–7940 9

TIBTEC-1223; No. of Pages 10

Review 44 Ohmuro-Matsuyama, Y. et al. (2013) Demonstration of proteinfragment complementation assay using purified firefly luciferase fragments. BMC Biotechnol. 13, 31 45 Wright, C.M. et al. (2011) A protein therapeutic modality founded on molecular regulation. Proc. Natl. Acad. Sci. U.S.A. 108, 16206–16211 46 Suel, G.M. et al. (2003) Evolutionarily conserved networks of residues mediate allosteric communication in proteins. Nat. Struct. Biol. 10, 59–69 47 Iwakura, M. et al. (2000) Systematic circular permutation of an entire protein reveals essential folding elements. Nat. Struct. Biol. 7, 580–585 48 Lee, J. et al. (2008) Surface sites for engineering allosteric control in proteins. Science 322, 438–442 49 Reynolds, K.A. et al. (2011) Hot spots for allosteric regulation on protein surfaces. Cell 147, 1564–1575 50 Tucker, C.L. and Fields, S. (2001) A yeast sensor of ligand binding. Nat. Biotechnol. 19, 1042–1046 51 Remy, I. and Michnick, S.W. (1999) Clonal selection and in vivo quantitation of protein interactions with protein-fragment complementation assays. Proc. Natl. Acad. Sci. U.S.A. 96, 5394–5399 52 Banala, S. et al. (2013) Switchable reporter enzymes based on mutually exclusive domain interactions allow antibody detection directly in solution. ACS Chem. Biol. 8, 2127–2132 53 Nirantar, S.R. et al. (2013) A generic scaffold for conversion of peptide ligands into homogenous biosensors. Biosens. Bioelectron. 47, 421–428 54 Janssen, B.M. et al. (2014) DNA-directed control of enzyme–inhibitor complex formation: a modular approach to reversibly switch enzyme activity. ACS Synth. Biol. Published online September 12, 2014. http:// dx.doi.org/10.1021/sb500278z 55 Kim, H. et al. (2014) Engineering beta-lactamase zymogens for use in protease activity assays. Chem. Commun. (Camb.) 50, 10155–10157 56 Golynskiy, M.V. et al. (2010) Antibody detection by using a FRET-based protein conformational switch. Chembiochem 11, 2264–2267 57 Stein, V. and Alexandrov, A. (2014) Protease-based synthetic sensing and signal amplification. Proc. Natl. Acad. Sci. U.S.A. 111, 15934– 15939 58 Tischer, D. and Weiner, O.D. (2014) Illuminating cell signalling with optogenetic tools. Nat. Rev. Mol. Cell Biol. 15, 551–558 59 Wu, Y.I. et al. (2009) A genetically encoded photoactivatable Rac controls the motility of living cells. Nature 461, 104–108 60 Yi, J.J. et al. (2014) Manipulation of endogenous kinase activity in living cells using photoswitchable inhibitory peptides. ACS Synth. Biol. 3, 788–795 61 Strickland, D. et al. (2012) TULIPs: tunable, light-controlled interacting protein tags for cell biology. Nat. Methods 9, 379–384 62 Muller, K. et al. (2014) Orthogonal optogenetic triple-gene control in mammalian cells. ACS Synth. Biol. 3, 796–801 63 Schmidt, D. et al. (2014) A fully genetically encoded protein architecture for optical control of peptide ligand concentration. Nat. Commun. 5, 3019 64 Lungu, O.I. et al. (2012) Designing photoswitchable peptides using the AsLOV2 domain. Chem. Biol. 19, 507–517 65 Niopek, D. et al. (2014) Engineering light-inducible nuclear localization signals for precise spatiotemporal control of protein dynamics in living cells. Nat. Commun. 5, 4404 66 Zhou, X.X. et al. (2012) Optical control of protein activity by fluorescent protein domains. Science 338, 810–814 67 Levskaya, A. et al. (2009) Spatiotemporal control of cell signalling using a light-switchable protein interaction. Nature 461, 997–1001 68 Toettcher, J.E. et al. (2013) Using optogenetics to interrogate the dynamic control of signal transmission by the Ras/Erk module. Cell 155, 1422–1434 69 Wend, S. et al. (2014) Optogenetic control of protein kinase activity in mammalian cells. ACS Synth. Biol. 3, 280–285 70 Karginov, A.V. et al. (2010) Engineered allosteric activation of kinases in living cells. Nat. Biotechnol. 28, 743–747 71 Chu, P.H. et al. (2014) Engineered kinase activation reveals unique morphodynamic phenotypes and associated trafficking for Src family isoforms. Proc. Natl. Acad. Sci. U.S.A. 111, 12420–12425

10

Trends in Biotechnology xxx xxxx, Vol. xxx, No. x

72 Karginov, A.V. et al. (2014) Dissecting motility signaling through activation of specific Src-effector complexes. Nat. Chem. Biol. 10, 286–290 73 Dagliyan, O. et al. (2013) Rational design of a ligand-controlled protein conformational switch. Proc. Natl. Acad. Sci. U.S.A. 110, 6800–6804 74 Mootz, H.D. et al. (2003) Conditional protein splicing: a new tool to control protein structure and function in vitro and in vivo. J. Am. Chem. Soc. 125, 10561–10569 75 Mootz, H.D. et al. (2004) Activation of an autoregulated protein kinase by conditional protein splicing. Angew. Chem. 43, 5189–5192 76 Auslander, D. et al. (2014) A designer cell-based histamine-specific human allergy profiler. Nat. Commun. 5, 4408 77 Auslander, D. et al. (2014) A synthetic multifunctional mammalian pH sensor and CO2 transgene-control device. Mol. Cell 55, 397–408 78 Ye, H. et al. (2013) Pharmaceutically controlled designer circuit for the treatment of the metabolic syndrome. Proc. Natl. Acad. Sci. U.S.A. 110, 141–146 79 Rossger, K. et al. (2013) Reward-based hypertension control by a synthetic brain-dopamine interface. Proc. Natl. Acad. Sci. U.S.A. 110, 18150–18155 80 Peisajovich, S.G. et al. (2010) Rapid diversification of cell signaling phenotypes by modular domain recombination. Science 328, 368–372 81 Bashor, C.J. et al. (2008) Using engineered scaffold interactions to reshape MAP kinase pathway signaling dynamics. Science 319, 1539–1543 82 Park, S.H. et al. (2003) Rewiring MAP kinase pathways using alternative scaffold assembly mechanisms. Science 299, 1061–1064 83 Whitaker, W.R. et al. (2012) Engineering robust control of twocomponent system phosphotransfer using modular scaffolds. Proc. Natl. Acad. Sci. U.S.A. 109, 18090–18095 84 Yeh, B.J. et al. (2007) Rewiring cellular morphology pathways with synthetic guanine nucleotide exchange factors. Nature 447, 596–600 85 Negron, C. and Keating, A.E. (2014) A set of computationally designed orthogonal antiparallel homodimers that expands the synthetic coiledcoil toolkit. J. Am. Chem. Soc. 136, 16544–16556 86 Thompson, K.E. et al. (2012) SYNZIP protein interaction toolbox: in vitro and in vivo specifications of heterospecific coiled-coil interaction domains. ACS Synth. Biol. 1, 118–129 87 Howard, P.L. et al. (2003) Redirecting tyrosine kinase signaling to an apoptotic caspase pathway through chimeric adaptor proteins. Proc. Natl. Acad. Sci. U.S.A. 100, 11267–11272 88 Barnea, G. et al. (2008) The genetic design of signaling cascades to record receptor activation. Proc. Natl. Acad. Sci. U.S.A. 105, 64–69 89 Djannatian, M.S. et al. (2011) Studying G protein-coupled receptor activation using split-tobacco etch virus assays. Anal. Biochem. 412, 141–152 90 Botvinnik, A. et al. (2010) Integrated analysis of receptor activation and downstream signaling with EXTassays. Nat. Methods 7, 74–80 91 Wehr, M.C. et al. (2008) Analysis of transient phosphorylationdependent protein–protein interactions in living mammalian cells using split-TEV. BMC Biotechnol. 8, 55 92 Wehr, M.C. et al. (2006) Monitoring regulated protein-protein interactions using split TEV. Nat. Methods 3, 985–993 93 Wehr, M.C. et al. (2013) Salt-inducible kinases regulate growth through the Hippo signalling pathway in Drosophila. Nat. Cell Biol. 15, 61–71 94 Daringer, N.M. et al. (2014) Modular extracellular sensor architecture for engineering mammalian cell-based devices. ACS Synth. Biol. Published online February 25, 2014. http://dx.doi.org/10.1021/ sb400128g 95 Gray, D.C. et al. (2010) Activation of specific apoptotic caspases with an engineered small-molecule-activated protease. Cell 142, 637–646 96 Taxis, C. and Knop, M. (2012) TIPI: TEV protease-mediated induction of protein instability. Methods Mol. Biol. 832, 611–626 97 Tasaki, T. et al. (2012) The N-end rule pathway. Annu. Rev. Biochem. 81, 261–289 98 Chen, S. et al. (2013) Building robust functionality in synthetic circuits using engineered feedback regulation. Curr. Opin. Biotechnol. 24, 790–796