Photodimerization systems for regulating protein-protein interactions with light

Abstract

Optogenetic dimerizers are modular domains that can be utilized in a variety of versatile ways to modulate cellular biochemistry. Due to their modularity, many applications using these tools can be easily transferred to new targets without extensive engineering. While a number of photodimerizer systems are currently available, the field remains nascent, with new optimizations for existing systems and new approaches to regulating biological function continuing to be introduced at a steady pace.

Introduction

A fast-evolving area in protein engineering involves the development of systems that use photosensory proteins to control biochemical processes with light. Such efforts have taken two main directions. First, a photosensory domain or protein may be tethered or inserted into a target protein domain, resulting in light-mediated changes in the photosensory domain inducing allosteric or steric changes in target protein activity or interactions. While this approach has been highly effective and can provide tight regulation in a single molecule, often significant protein engineering and optimization is required for each target to achieve robust light control of protein function. A second, more modular approach to regulating protein activity with light uses natural or artificially engineered protein pairs that undergo alterations in binding affinity with light. These paired molecules, termed photodimerizers or light-regulated dimerizer systems, can be affixed to target proteins or domains in diverse ways leading to changes in target protein activity.

Types of Photodimerizer Systems

Photodimerization systems developed to date are all derived from natural photoreceptor domains originating from plants, animals, algae, and fungi, although some have additional engineered components. Such systems consist of two proteins that undergo changes in binding affinity upon light illumination. At least one of the partner pairs is comprised of a light-sensitive photoreceptor domain that undergoes a conformational change with light illumination. The conformational change can lead to altered binding affinity to a partner protein, or altered homomeric interactions [1,2]. Central to the light-induced conformational change is the initial absorption of a photon by the photoreceptor. Each photoreceptor is conjugated to a light-responsive chemical, a chromophore that undergoes a local conformation or oxidation state change upon photon absorption, leading to a structural change within the protein [3,4]. Table 1 describes the different photodimerizer systems that have been developed to date for use in optogenetic regulation, highlighting the chromophore used for each system. In some photoreceptors, including cryptochromes, phytochromes, LOV domains, and BLUF domains, the chromophore consists of a small molecule cofactor conjugated to the protein [3]. In other cases, as with UVR8 or Dronpa, the chromophore is intrinsically contained within the protein amino acid sequence [5–7].

Table 1.

Optogenetic modules for controling protein-protein interactions

System	Photoreceptor	Co- factor	Typical/peak excitation wavelength		Size (amino acids)	Approx. lifetime (half-life)	References
			Association	Dissociation
phyB/PIF3	Arabidopsis thaliana phytochrome B	PΦB, PCB	Pr: 660 nm	Pfr: 740 nm	621/524	light inducible	[24]
phyB/PIF6	Arabidopsis thaliana phytochrome B	PΦB, PCB	Pr: 660 nm	Pfr: 740 nm	908/100	light inducible	[9]
BphP1/PpsR2	Rhodopseudomonas palustris bacteriophytochrome	Bilive rdin	Pfr: 740 nm	Pr: 650 nm	732/465	light inducible, dark reversion 15min	[35]
CRY2/CIB1	Arabidopsis thaliana cryptochrome 2	FAD	450 nm	dark	612/335 498/170 (PHR/CIBN)	5 min	[25]
CRY2PHR/ CRY2PHR (clustering)	Arabidopsis thaliana cryptochrome 2	FAD	450 nm	dark	498	5 min	[49,50,59]
CRY2olig, CRY2clust	Arabidopsis thaliana cryptochrome 2	FAD	450nm	dark	498	5 min	[52,54]
TULIPs	Avena sativa Phot1 LOV2 domain	FMN	450 nm	dark	153/194	tunable	[10]
iLID/SspB	Avena sativa Phot1 LOV2 domain	FMN	450 nm	dark	144/110	tunable	[63,64]
FKF1/GI	Arabidopsis thaliana FKF1	FMN	450 nm	dark	619/1173(wt) 178/1173(N-LOV)	62.5 h	[27]
VVD/VVD	Neurospora crassa Vivid (VVD)	FAD	450 nm	dark	150	tunable	[29]
nMag/pMag (Magnets)	Neurospora crassa Vivid (VVD)	FAD	450nm	dark	150/150	tunable	[65]
LOVTRAP	Avena sativa Phot1 LOV2 domain	FMN	dark	450 nm	143/59	tunable	[58]
EL222	Erythrobacter litoralis HTCC2594 EL222	FMN	450 nm	dark	208	< 50 s	[28]
Aureochrome	Vaucheria frigida Aureochrome 1 LOV	FMN	450 nm	dark	136	7 min (tunable)	[47]
PixD/PixE	Synechocystis sp. PCC6803 PixD, PixE	FAD	dark	450 nm	150/380	s-min	[66,67]
Dronpa	Dronpa	None	405 nm	488 nm	257	light inducible	[60]
UVR8/COP1	Arabidopsis thaliana UVR8	None	280-315nm	N/A	440/340		[68]
UVR8/UVR8	Arabidopsis thaliana UVR8	None	dark	280-315nm	440	>8 hrs	[55]

In addition to cofactor differences, the photodimerizer systems vary with respect to size, spectral sensitivity, and reversibility kinetics. The rate of dissociation of protein interactions can range from seconds to days, depending on the system. While all photodimerizer systems described in Table 1 revert back to their dark state spontaneously, some such as phytochromes and Dronpa are optically reversible with light illumination, allowing user-defined control of reversibility. Fast reversal kinetics are particularly important in achieving local, spatially-resolved subcellular control of processes: fast turnoff allows maintenance of a gradient of activated protein that counteracts cellular diffusion. Conversely, systems with slow reversal kinetics are useful for other applications that seek to induce more permanent phenotypes. As in vivo use of these tools can be quite varied, the availability of a variety of photodimerizer systems with different parameters and properties allows tunability and optimization for a specific application.

While all photodimerizer systems result in changes in binding protein interactions upon light absorption, light can alter affinity of interactions in different ways (Figure 1), resulting in increased affinity of partners (light-induced dimerization), decreased affinity of partners (light-induced dissociation), increased or decreased affinity depending on the light wavelength (eg., phytochrome), or in some cases, increased or decreased self-association (light-modulated oligomerization). Each of these approaches has been utilized in different ways for downstream applications, and in some cases have been used in conjunction for complex regulation of biological processes.

Figure 1. Controlling protein interaction with light.

A. Light induced dimerization. Light illumination promotes interaction between two proteins. B. Light induced dissociation. Two proteins interact in dark but are dissociated upon light illumination. C. Light induced dimerization and dissociation. Different wavelengths of light illumination promote either association or dissociation. The ground state is adopted in dark. D. Oligomerization. Light illumination triggers self-association.

Applications of photodimerizer systems

Basic approaches for regulation of protein activity using photodimerizer systems include manipulation of protein localization, reconstitution of split proteins, activation through induced dimerization or oligomerization, and sequestration/release (Fig. 2). Many of these approaches are directly derived from methods originally implemented with chemical dimerization systems (in which protein domains are induced to dimerize by addition of a small molecule that promotes association) [8], but provide advantages of reversibility and spatial control. Below are described some of the central strategies using protein dimerizers for protein photocontrol, highlighting both pioneering and recent studies.

Figure 2. Photodimerizer approaches to regulate protein function.

Schematic showing approaches used to regulate protein activity using photodimerizers. The target protein regulated is colored in green, with photosensory proteins in blue, partner proteins in purple, and anchoring proteins in orange. A) Recruitment to anchored subcellular location. One of the photodimerizer components is fused to a protein or peptide allowing anchored subcellular localization (plasma membrane is shown as an example), while the other is fused to a target protein of interest. Light illumination allows recruitment of the target to the anchored location. B) Reconstitution of a split protein. Activity is achieved by fusing the N-terminus and C-terminus of a split protein to the interacting photodimerizers, allowing functional reconstitution with light. C) Reconstitution of a split transcription factor. Variation of (B), where one of the photodimerizer partners is fused to a DNA binding domain, where it can bind to DNA at a promoter site. The partner photodimerizer is fused to a transcriptional activation domain. Light allows recruitment of the activation domain, resulting in activation of transcription. D) Dimerization of two different proteins. Photodimerizers can be used to bring two different (or the same) target proteins together with light. E) Oligomerization. A target is fused to a photoreceptor (such as CRY2) that undergoes light dependent oligomerization, which can be used to induce or disrupt activity. F) Sequestration/release. A target protein is anchored (sequestered) at an inactive subcellular location, then released with light illumination to allow function. G) Dissociation of protein clusters. Used with light-dissociated dimerizers, protein assembles into clusters that impede trafficking or activity, which are dissociated/dissolved with light. H) Single chain caging. Interacting partners are presented on the same protein chain, along with a target protein or domain. Function of the target protein is sterically or allosterically impeded when photodimerizers are bound, but enabled when partners dissociate.

Manipulating protein localization

Because protein function is intrinsically tied to protein localization, manipulation of protein localization using dimerizers (Fig. 2A) has been a useful approach to regulate function. In initial demonstrations, the PhyB/PIF6 optogenetic system was used to recruit the DHPH catalytic domains of GDP-GTP exchange factors (GEFs) Tiam or intersectin to the plasma membrane (PM), resulting in induction of Rac1 or Cdc42 activity, respectively, with light application [9]. Remarkably, recruitment could be locally controlled, thus light could be steered to specific subcellular locations to regulate activity at these locations. With Tiam DHPH recruitment, the authors demonstrated activation of Rac1 and formation of cellular protrusions extending in the direction of the light [9]. Optogenetic PM recruitment has also been successful for regulating kinase signaling pathways. For example, in yeast, TULIPs and CRY2/CIB1 were used to recruit the MAP kinase scaffold Ste5 to the PM, resulting in activation of the MAP kinase mating pathway [10,11]. Dimerizer tools have also been successfully used to modify lipid signaling networks, through recruiting kinases or phosphatases targeting phosphoinositide species. Idevall-Hagren et al. used CRY2PHR/CIBN to recruit an inositol 5-phosphatase domain or a PI3-kinase domain (inter-SH2 (iSH2) region of the p85α regulatory subunit of PI3-kinase) to the PM, resulting in rapid, local control of phosphoinositide signaling [12]. Recruitment of the inositol 5-phosphatase domain of OCRL resulted in PI(4,5)P₂ dephosphorylation and loss of membrane ruffling, while recruitment of PI3-kinase resulted in induction of PI(3,4,5)P₃ synthesis and membrane ruffling [12].

As light can be applied at specific times for defined durations, the use of optogenetic tools can allow examination of the responses of signaling networks to different light-entrained inputs. In one study, PhyB/PIF6 dimerizers were used to inducibly recruit the SOS catalytic domain to the PM for different durations and frequencies, leading to activation of the Ras-ERK signaling pathway [13]. In a separate study also examining ERK signaling, CRaf was inducibly recruited to the PM for specific durations using CRY2PHR/CIBN dimerizers [14]. In both studies, the authors demonstrated the powerful ability of such tools to stimulate signaling nodes for precise, user-specified amounts of time, allowing fine-tuned probing of the circuitry and feedback mechanisms inherent to these pathways.

While light focused locally can be used to recruit signaling enzymes to specific regions of the PM (for example as in the PhyB/Pif6 studies noted above [9]), another strategy for local PM recruitment involves tethering one dimerizer to a protein that is differentially localized on the PM, combined with global illumination. Studies by De Renzis et al. used CRY2/CIB1 dimerizers to recruit the catalytic domain of RhoGEF2 to the plasma membrane of Drosophila epithelial cells, resulting in activation of myosin-II in Drosophila embryonic tissue [15,16]. By fusing one of the dimerizers to a protein localized to only the basal PM, they were able to specifically activate myosin-II at the basal side using global light, resulting in disruption of normal cell shape changes that occur during ventral furrow formation in gastrulation [16]. Another approach to increase spatial resolution of PM recruitment involved immobilizing one dimerizer component to both the PM and a secondary anchor, the microtubule cytoskeleton, resulting in slower diffusion of the anchored molecule [17]. Finally, a newly characterized photoreceptor, BcLOV4, undergoes structural changes in light allowing direct recruitment to the PM, showing promise for use in single-component systems for inducible PM recruitment [18].

In addition to recruitment to the PM, studies have used dimerizers to recruit proteins to many other intracellular structures. TULIPs and CRY2PHR/CIBN were used to recruit cytoskeletal motor proteins to cargos, including peroxisomes and mitochondria, resulting in light control of intracellular organelle positioning [19,20]. Dimerizer tools have also been used to localize proteins to subcellular compartments, such as the nucleus or cytoplasm. One study fused phyB(1-908) and a NES to a target protein, resulting in cytosolic localization of the target in dark but induced nuclear targeting by light-triggered interaction with PIF3 (which contains an endogenous NLS) [21]. CRY2/CIBN have also been used to guide cargo into exosomes, by fusing the cargo to CRY2PHR and anchoring CIBN to an exosome-associated tetraspanin protein, CD9 [22]. A recent study compared three different light-induced dimerizer tools, CRY2/CIB1, iLIDs, and Magnets for efficient local recruitment to a precise subcellular location (for example, a specific site on mitochondria) [23]. While each system was effective in recruitment, systems possessing very fast dark reversion rates showed enhanced local recruitment [23]. Using Magnet dimerizers, the authors also demonstrated recruitment to single organelles, including lysosomes or endosomes.

Reconstituting split proteins

Another approach effective with photodimerizers has been to reconstitute split proteins (Figure 2B-C). Proteins are split into two fragments, with each fragment attached to one half of a dimerizing pair. While the fragments do not have sufficient affinity on their own, light induced-dimerization increases the proximity of the of the split protein fragments, leading to reconstitution of activity. The earliest demonstration of reconstitution of split protein fragments using photodimerizers used phyB and PIF3 to regulate a split Gal4 binding domain and activation domain [24]. In this case, red light-induced dimerization of phyB and PIF3 led to reconstitution of activity, while far-red light dissociated fragments resulting in inactivation. Subsequently, similar approaches have been used with other photodimerizers, allowing for precise control of transcription in space and time [25–34]. In a recent publication, control of a split transcription factor in mammalian cells was induced with a bacterial phytochrome BphP1-PpsR2 interaction, allowing transcriptional control by red/far-red light without the requirement of exogenous cofactor (needed with plant phyB-based systems) [35].

While the above examples involved reconstitution of split protein fragments, they did not attempt to reconstitute a split enzyme active site, which can be more difficult. In one of the first examples of split enzyme reconstitution, the phyB/PIF3 system was used in yeast to reconstitute a split protein splicing enzyme, the yeast vacuolar ATPase intein [36]. Cre DNA recombinase was also reconstituted with light allowing recombination at loxP sites in the genome using CRY2/CIB1 photodimerizers [25]. Further optimization yielded an optimized (paCre2.0) system with higher dynamic range, that can be activated by a single pulse of light [37]. A different split Cre system was also developed using nMag/pMag photodimerizers, which was shown to be functional in vivo [38]. A split version of Cas9 (paCas9), allowing induction of targeted genome modifications in mammalian cells, was also developed using nMag/pMag photodimerizers [39], which was optimized for enhanced transcriptional control through additional coactivator recruitment [40]. Split versions of T7 RNA polymerase were also effectively reconstituted with light, using dimerization of Magnets and/or Vivid [41,42]. In many of these studies generating split protein systems, multiple split sites were systematically tested for efficacy, prior to chosing an optimal location. A recent study reports a new computational method to determine effective sites for splitting proteins for inducible assembly, which has the potential to be highly useful for such studies [43].

Protein homodimerization, heterodimerization, and oligomerization

A common mechanism used in Nature to regulate protein activity is through induced homodimerization, heterodimerization, and oligomerization. Using photodimerizers or photooligomerizers, scientists can utilize the same mechanisms to inducibly regulate cell signaling or activity with light (Fig. 2D-E). In one of the first studies to stimulate protein-protein interactions using light, PhyB and PIF3 were used to inducibly recruit a weak-affinity, GDP-bound form of Cdc42 to its effector, WASP (Wiskott-Aldrich Syndrome Protein), stimulating actin assembly [44]. This study demonstrated that two proteins with low affinity could be inducibly forced to associate upon dimerization of the tethered light-interacting molecules.

Photodimerizers have since been used in many ways to promote protein association. Tyrosine kinase receptors, which are naturally activated by ligand-induced dimerization, can be induced to dimerize and signal using photodimerizers attached to intracellular domains, for example with TrkB receptor [45] and FGFR receptor [46], both using CRY2PHR/CIBN photodimerizers. The LOV domains of aureochrome-1 have also been used to dimerize receptor tyrosine kinases (FGFR1, EGFR, and RET) [47]. Downstream kinases involved in signaling have also been regulated using these approaches, such as CRaf and BRaf homodimerization and heterodimerization using CRY2PHR/CIBN [48].

Several groups have exploited light-dependent oligomerization of CRY2 to activate biological processes normally activated by protein oligomerization. CRY2-mediated clustering of TopBP1, a protein that activates ATR kinase, was used to induce the DNA damage response in the absence of DNA damage [49]. Stimulation of β-catenin or Rho GTPase pathways was induced by light-induced clustering of CRY2PHR-LRP6c or CRY2PHR-Rac1/RhoA, respectively [50]. In a follow up study, binding domains fused to CRY2PHR were found to show increased binding avidity upon light-induced clustering, providing a means to target and activate endogenous receptors with light [51]. CRY2olig, a mutant of CRY2PHR with increased oligomerization potential, was used to stimulate actin polymerization by clustering CRY2 fused SH3 domains of Nck [52]. Other mutants of CRY2 have also been found that enhance or even disrupt self-oligomerization properties [53,54]

Using photodimerizers to sequester/release proteins

Dimerization tools have also been used to sequester protein activity, allowing release at a user-specified time (Figure 2F-H). One of the first demonstrations of sequester/release with photodimerizers used UVR8/UVR8 interactions [55]. Tandem dimers of UVR8 were fused to an endoplasmic reticulum-processed protein, VSVG, which led to sequestration and entrapment of fused VSVG in the ER in dark. Light illumination disrupted UVR8 interactions resulting in release of ER sequestration and resumption of trafficking [55]. This work showed the power of using light to inducibly release proteins from a sequestered site.

In a different study, Yang et al. elegantly demonstrated use of phyB/PIF3 to sequester/release proteins [56]. They attached Clb2 to PIF6, allowing recruitment to phyB-tethered cellular locations such as the nucleus or spindle pole body. Sustained recruitment (with red light) resulted in alteration in Clb2-dependent pathways. The group used far-red light to release Clb2 from sequestration at specific times, for specific durations, testing whether recovery of Clb2 function at those times and durations was sufficient for rescue of altered pathways. The same approach was also used to sequester/release Bem1, a protein that regulates activity of Cdc42 in yeast [57]. In the LOVTRAP system, proteins of interest are fused to Zdk, which binds tightly to the dark state of AsLOV2 [58]. Localization of AsLOV2 to a subcellular location (for example, the mitochondrial membrane) allows sequestration of Zdk-fused cargo in dark, which can be released with light. This approach was utilized to regulate Rho GTPase signaling by sequestering constitutively active versions of Vav2, Rac1, or PI3K at the mitochondria in dark and releasing them with light [58].

The CRY2/CIB system has also been used to sequester proteins to block function using the LARIAT (light activated reversible inhibition by assembled trap) system [59]. This approach takes advantage of the propensity of CRY2 to cluster when associated with a multimeric protein, which can also occur through binding to a CIB-multimeric protein fusion. A variety of proteins including Vav2, Tiam1, Rac1, Cdc42, RhoG, and tubulin were clustered and regulated using this approach [59]. CRY2olig has also been used to disrupt function: attachment of CRY2olig to clathrin light chain resulted in disruption of endocytosis with light due to steric effects of clustering [52].

Another approach to sequestering proteins uses internal photodimerizer pair protein-protein interactions to block function (Fig. 2H). For example, one or two monomers of Dronpa were tethered on either side of an enzyme, blocking the catalytic site in dark [60,61]. Upon light-induced monomerization of Dronpa, the enzyme catalytic site was exposed and active. This approach was used on GEF proteins that stimulate Cdc42, the hepatitis C virus NS3-4A protease, Cas9, and various kinases [60–62]. These studies provide powerful examples of the potential of optogenetic tools for acutely modulating protein function by sequester/release approaches.

Conclusions

In summary, light regulated dimerization systems are powerful tools for regulating biochemical function with superb spatial and temporal control. A variety of dimerizer systems have been developed, each with diverse properties for different biological applications. While the blue light dimerizer systems are widely available and have been the most extensively used, new applications and optimized versions of these tools are continuing to emerge. Red and far-red responsive dimerizers have been less characterized and utilized, but emerging systems show significant potential and some advantages over blue light approaches. The studies highlighted here demonstrate some of the different approaches for using such tools to acutely manipulate protein function, altering localization, interactions, domain associations, or active site access. As this relatively recent field evolves, new approaches for manipulating protein activity using dimerizers will emerge, further enhancing the versatility of these powerful tools.

Highlights.

• Photodimerizer systems are a recently developed technology with similarity to chemical dimerizer approaches, allowing control of protein-protein interactions with light.

• A variety of photodimerizers have been developed, which can be used to induce and/or dissociate interactions with light or control oligomeric states.

• Dimerizer systems have primarily been used to manipulate protein localization, reconstitute split proteins, and manipulate dimerization or oligomerization states.

• These powerful tools can enable spatial, temporal, reversible, and dose-dependent control of biological processes.