Function, structure, and mechanism in bacterial photosensory LOV proteins

Abstract

LOV domains are protein photosensors conserved in bacteria, archaea, plants and fungi that detect blue light via a flavin cofactor. In the bacterial kingdom, LOV domains are present in both chemotrophic and phototrophic species, where they are found N-terminally of signaling and regulatory domains such as sensor histidine kinases, diguanylate cyclases/phosphodiesterases, DNA-binding domains, and σ factor regulators. In this review, we describe the current state of knowledge on the function of bacterial LOV proteins, the structural basis of LOV domain-mediated signal transduction, and the use of LOV domains as genetically-encoded photoswitches in synthetic biology.

Bacteria detect specific intra- and extracellular signals through sensory proteins^1,2 and RNAs^3–5, which then regulate an adaptive response. There is a detailed understanding of the regulatory response to signals such as nutrient status, oxidative state, osmolar state, and pH, in a handful of model systems. However, in typical natural environments there is tremendous species diversity, and a multitude of potential physical and chemical signals to which a bacterium could, in theory, respond. On this broader scale, we know very little of which signals are relevant to which bacterial species. Over the past decade, a surprising regulatory signal has emerged in the study of bacterial physiology: visible light. Although, the role of visible light in the regulation of photosynthesis and pigment production in phototrophic species is well studied and generally understood^6,7, there are decades-old reports of blue light-dependent behavioral^8,9 and developmental^10,11 phenomena in chemotrophic species that have yet to be ascribed a mechanism. With the discovery of several new classes of blue light photoreceptors ^12–14, and the realization that these photoreceptors are encoded across a broad phylogenetic cross-section of bacteria ¹⁵, there is a growing appreciation that blue light perception (i.e. light in the wavelength range of 440–480 nm) may be a common attribute.

Light is a ubiquitous signal in terrestrial and aquatic ecosystems, and can give important information about niche position. It is, therefore, not surprising that sensory systems that detect photons in the visible range are present in bacterial species. There is a long history of the study of light-dependent DNA damage, whereby photons in the near- and far-ultraviolet region of the spectrum alter the chemical structure of the genetic material. Many bacteria encode photoreceptors that regulate the synthesis of protective pigments or that directly repair photodamaged DNA in response to photon absorption ^16,17. There are also bacterial signaling proteins, namely the bacteriophytochromes, that detect red and far-red light and regulate varied cellular responses to these lower energy wavelengths ⁷. This review does not focus on these UV and red-light response systems but rather on recent developments in our understanding of bacterial cell signaling systems that detect blue photons via a class of flavin-binding photosensors known as Light, Oxygen, or Voltage (LOV) domains (Table 1). In particular, we concentrate on the reported functional roles for LOV protein photoreceptors in both Gram-positive and Gram-negative species, on structural mechanisms of light regulation of LOV protein activity, and on developments in the use of LOV domains in the area of protein engineering and synthetic biology.

Table 1.

Selected LOV proteins from plant, fungi, and bacteria

	Protein name	cofactor	Effector domain	organism	Function	Ref.
Plant	Phot1	FMN	Ser/Thr kinase	Arabidopsis thaliana, Avena sativa	phototropism, chloroplast movement, stomatal opening	50,126–128
Fungi	Vivid	FAD	-	Neurospora crassa	circadian pigment expression	28,129
	WC-1	FAD	Zinc finger	Neurospora crassa	circadian photoreceptor, transcription factor	26,27,30,130,
	WC-1 homolog	FMN	Zinc finger	Cryptococcus neoformans, Fusarium oxysporum, Histoplasma capsulatum	circadian photoreceptor, virulence regulation	29,131–133
Bacteria	YtvA	FMN	STAS	Bacillus subtilis	stress response, sporulation (?)	37,64,66
	Lov-HK	FMN	Histidine kinase	Brucella abortus	virulence regulation	39
	SL2	FMN	GGDEF-EAL	Synechococcus elongatus	cyclic-di-GMP phosphodiesterase	70
	LovK	FMN	Histidine kinase	Caulobacter crescentus	cell adhesion	38

The discovery of LOV domains

Early genetic work on non-phototropic hypocotyl (nph) mutants in Arabidopsis thaliana¹⁸ suggested that NPH1 was the photoreceptor for phototropism. NPH1 (subsequently named phototropin) was cloned, sequenced¹⁸, and demonstrated to be a photoreceptor serine/threonine kinase that binds a flavin-mononucleotide (FMN) cofactor in each of two N-terminal LOV domains¹². LOV domains have since been more strictly defined as a subset of the larger Per-ARNT-Sim (PAS) domain superfamily¹⁹, which bind a flavin cofactor (either FMN or FAD) and undergo light-dependent cysteinyl-C4(a) adduct formation ²⁰ via a triplet intermediate ²¹. In the context of phototropins, blue light absorption by LOV domains regulates autophosphorylation activity of the C-terminal serine/threonine kinase domain²². Phototropins have subsequently been shown to regulate various aspects of plant photomorphogenesis including leaf and stomatal opening, and chloroplast relocation ²³ (Table 1).

Since the initial discovery ¹⁸ and classification ¹² of LOV domains in plant phototropins, there has been an explosion in research on molecular and cellular aspects of light-dependent signal transduction mediated by LOV domains. Genes encoding LOV domains coupled to effector domains other than kinases were soon discovered in plants ^24,25, fungi ^26–30, and stramenopile algae³¹ and demonstrated to regulate a number of circadian processes and developmental phenomena via the same cysteinyl-C4(a) photochemistry. LOV domains have also been identified in bacterial genomes^32,33 where they are coupled to diverse signaling output domains ^23,34,35 (Figure 1) and regulate processes including general stress response ^36,37, cell envelope physiology ³⁸, and virulence ³⁹(Table 1).

Figure 1.

Domain architecture of select LOV proteins; this cartoon presents just a few examples of hundreds of LOV domain-containing proteins. Accession numbers and formal protein names (if available) are shown. The length of LOV proteins is specified.

LOV domain distribution and evolution

Bioinformatic analyses predict that somewhere between 3.5% ⁴⁰ to over 10% ¹⁵ of sequenced bacterial genomes contain at least one gene encoding a LOV domain. Among the bacteria, the genomes of the alphaproteobacteria and cyanobacteria have the highest number of encoded LOV proteins ^40,41. A phylogenetic tree constructed from LOV protein sequences across the three kingdoms of life reveals two distinct subtrees⁴⁰ (Figure 2). The first sub-tree includes LOV sequences from cyanobacteria, actinobacteria, proteobacteria, firmicutes and chloroflexi. Surprisingly, the sequences of the LOV-domain containing plant circadian regulators, ZTL/ADO/FKF1, are also present in this sub-tree. The second sub-tree consists of eukaryotic sequences from fungi (White Collar-1), plants and algae (phototropin, neochrome and aureochrome) and bacterial sequences predominately from alphaproteobacteria. From this result, it has been suggested that eukaryotic LOV domains present in the circadian plant photoreceptors ZTL/ADO/FKF1 have a cyanobacterial origin whereas the plant phototropin LOV domains, aureochrome LOVs, the fungal White Collar-1-LOVs and the animal LOVs have an alphaproteobacterial origin ⁴⁰. In accordance with endosymbiotic theory, ancestors of chloroplasts and mitochondria are internalized cyanobacteria and alphaproteobacteria^42,43, respectively, that may have transferred their genes to the genomes of a eukaryotic cell post endosymbiosis ⁴⁴.

Figure 2.

LOV domain sequences are conserved across multiple kingdoms. Tree adapted from Krauss and colleagues ⁴⁰. LOV sequences of different families are color coded: the archaea in turquoise, alphaproteobacteria (α) in pink, betaproteobacteria (β) and gammaproteobacteria (γ) in yellow, actinobacteria (actino) are in brown, chloroflexi (chloro) in dark green, firmicutes (firmi) in grey, cyanobacteria (cyano) in light blue, fungal sequences from the white-collar 1 (WC-1) protein are in green (fungi), plant phototropin and neochrome (Phot/Neo) LOV domains (LOV1 and LOV2) are in blue, algal aureochrome LOV sequences (aureo) in dark purple, the ZTL/ADO/FKF1-LOV family of plant regulators are in purple, and putative animal LOV domains are in red. Branches have been collapsed; triangle size is proportional to the number of collapsed branches. Numbers of collapsed branches are indicated at the base of each triangle.

Diversity in LOV signaling output

In bacteria, LOV proteins generally follow the domain organization described in other bacterial signalling proteins ⁴⁵, being arranged with a sensor domain (i.e. LOV) situated N terminally to a linker sequence and an effector/output domain. To date, several different classes of LOV domain and effector domain combinations have been described^34,35,46,47 (Table 1 and Figure 1). Among this diverse set, two major groups emerge. The first, LOV-histidine kinases, correspond to approximately 50% of bacterial LOV proteins. The second, LOV-GGDEF/EAL proteins, are predicted to regulate the synthesis and hydrolysis of cyclic-di-GMP and constitute ~20% of bacterial LOV proteins. Other less common LOV signalling proteins include LOV-STAS (Sulfate Transporter Anti-Sigma antagonist) proteins (~10%), LOV Helix-turn-helix (HTH) proteins (~3.5%) and the LOV-SpoIIE (SPOrulation stage II protein E) proteins (~2%). A small number of LOV proteins with a Globin domain, CheB/CheR domains or a Cyclase 4 domain have also been reported⁴⁷. LOV domains are also found as a single domain or are often associated with additional sensor domains (e.g. other PAS domain, GAF or CHASE domains). Such complex domain arrangements can allow integration of multiple environmental signals and provide mechanisms of light signal amplification or attenuation ⁴⁸.

LOV domain structure

The LOV core domain⁴⁹ has a classical PAS fold, consisting of a five-stranded antiparallel β sheet (Aβ, Bβ, Gβ, Hβ and Iβ), and helical connector elements (Cα, Dα, Eα and Fα). A flavin chromophore is bound non-covalently in a binding pocket delimited by the core (Figure 3A) ⁵⁰. As discussed above, an invariant cysteine residue forms a covalent bond with the flavin C(4a) carbon when illuminated with blue light (Figure 3C). This conserved cysteine is located on the Eα helix of the central core in a conserved GXNCRFLQ motif ¹² (Figure 3B) and can adopt two distinct conformations in the dark state ⁵¹. In most cases, the adduct thermally decays back to the ground state with a half life of minutes to hours, depending on the respective LOV protein ⁵²; in select cases adduct formation is irreversible (or extremely slow) ^39,53. Mutational studies have shown that the cysteine is not essential for flavin binding but is essential for the reversible photochemical reaction ²⁰. The dark recovery (i.e. rupture of the cysteinyl-C4a bond) of LOV domains can be modified by pH ^21,54,55, salt concentration ^54,55, or the presence of low-molecular-weight bases⁵⁶. These data point toward a mechanism in which structural and solvent effects on the protonation state of flavin N5 control the rate of dark recovery ^21,56,57.

Figure 3.

(A) Adiantum capillus-veneris phy3(neochrome)-LOV2 domain in the illuminated state (PDB code: 1JNU). α-helices are colored in blue; β-strands in yellow; loops in green. The FMN cofactor is colored in light gray. Helix Eα is shown as transparent to reveal the cysteinyl-FMN adduct. (B) Flavin-binding pocket of phy3-LOV2 domain. Residues that interact with the FMN cofactor are shown for the illuminated state (PDB code: 1JNU) (in orange), and for the dark state (PDB code: 1G28) (in gray). The glutamine (Q1029) of strand Hβ that rotates 180° on light activation is labeled as is the cysteine that forms the cysteinyl-FMN adduct (C966). (C) Schematic of cysteinyl-C(4a) covalent adduct formation in response to light absorption by the LOV domain.

The function of bacterial LOV proteins

Bacillus YtvA and the general stress response

In the Gram-positive soil bacterium Bacillus subtilis, the LOV-STAS protein YtvA (see Table 1 and Figures 1 and 4) binds a flavin cofactor and functions as a photoreceptor in vitro and in vivo. ytvA was initially reported as a positive regulator of the general stress factor σ_B ³⁶, and later shown to mediate blue light activation of the σ^B-regulated ctc promoter when overexpressed³⁷. YtvA may be involved in light-dependent control of sporulation as a result of cross regulation between the σ^B and cell sporulation pathways ^37,58. Though light regulation of sporulation has not been demonstrated in B. subtilis, it has been reported that blue light can inhibit sporulation in Bacillus licheniformis ¹⁰.

Figure 4.

Biochemical activities and cellular responses affected by LOV photoreceptor proteins in four bacterial species. Brucella abortus: on light activation BaLOV-HK is phosphorylated and positively regulates intracellular proliferation in a macrophage infection model. Caulobacter crescentus: light activation of the LovK protein induces autophophorylation. LovK, together with LovR, regulates cell adhesion through an unknown mechanism. Synechoccocus elongatus: regulation of the phosphodiesterase activity of the EAL domain is controlled by the LOV domain. Thus, light exposure controls the di-c-GMP level of the cell, an important second messenger. Bacillus subtilis: during stress conditions, illumination of the LOV domain of the YtvA protein is an important signal for σ^B activation and stress response. The associated STAS domain, maybe involved in energy level sensing, could also be regulated by the LOV domain as well as sporulation phenomenon.

YtvA and light regulation of B. subtilis σ^B occurs through a large protein signalling complex known as the stressosome, which contains several other STAS proteins ⁵⁹. Although the exact mechanism of this regulation is unknown, our understanding of stressosome signalling has expanded in recent years. Under stress conditions, the B. subtilis STAS domain proteins RsbS and RsbR are phosphorylated by a Serine/Threonine kinase (RsbT) and indirectly regulate σ^B activity through the stressosome ^36,60,61. Phosphorylation of YtvA has not been reported^36,62, suggesting that unlike other STAS proteins in the stressosome, YtvA activity is not phosphorylation dependent ^36,63. YtvA has been reported to bind GTP in vitro ^64–66; mutation of the putative GTP binding site abolishes blue light activation of σ^B-dependent transcription in vivo ⁶³. Based on these data, it has been proposed that YtvA is a light-dependent NTP recruiter for the RsbT kinase ⁶³. However, the hypothesis that YtvA explicitly requires GTP binding to function merits further investigation ⁶⁷.

Brucella LOV-HK: light and intracellular proliferation

The Gram-negative intracellular pathogen, Brucella abortus, encodes a LOV histidine kinase (LOV-HK) that exhibits an increase in histidine autophosphorylation on blue light absorption³⁹. Surprisingly, exposure of wild-type B. abortus to visible light results in a 10-fold higher level of cell replication in murine macrophages compared to a dark control³⁹. This light-dependent enhancement of virulence in a macrophage infection model requires LOV-HK and more specifically, the LOV cysteine residue that forms a C4(a) flavin adduct (see Table 1, Figures 1 and 4)³⁹. Once in the adduct state, photoactivated LOV-HK shows negligible decay back to the ground state in vitro, which suggests that once LOV-HK is illuminated in vivo, kinase activity will be similarly long lived³⁹.

Although there is a lack of data on the role of light in the B. abortus life cycle, it is interesting to consider possible links between the light environment and virulence. Light may function as an environmental signal that cues bacteria that have been ejected from a host to upregulate pathways that promote infection of a new host ^39,68. We note that alternative B. abortus LOV-HK signalling models, in which environmental cues such as oxidative state or redox potential affect the state of the flavin cofactor and the activity of the histidine kinase have not been explicitly tested. It is certainly possible that LOV-HK may function to integrate multiple environmental signals.

Caulobacter LovK and the regulation of cell adhesion

Phosphorylation of the Caulobacter crescentus LOV histidine kinase, LovK, is regulated by blue light in vitro ³⁸. In vivo, coordinated low-level overexpression of lovK and the adjacent receiver gene, lovR, results in a dramatic increase in cell adhesion that is accentuated by exposing cell cultures to blue light ³⁸ (see Figure 4). Blue-light-dependent enhancement of cell adhesion requires the presence of the conserved cysteine residue in the LOV domain that forms a cysteinyl-flavin adduct. A recent biochemical analysis of C. crescentus LovK has demonstrated that the reduction midpoint of the LovK flavin cofactor is −260 mV, which is poised near the redox potential of the cytosol⁶⁹. Given that the capacity of LOV domains to function as photosensors requires the flavin cofactor be in the oxidized state, cellular redox state could affect flavin redox state and accordingly influence the LOV photoactivity. Indeed, reduction of the LovK FMN cofactor in vitro ablates light-dependent regulation of kinase activity⁶⁹.

As with B. abortus, it is not immediately evident why blue light should affect C. crescentus cell physiology. C. crescentus is an oligotrophic bacterium that has evolved in dilute freshwater environments. Blue wavelengths efficiently penetrate the water column; thus light could provide an environmental signal that informs Caulobacter about its niche position. However, the observation that redox state can modulate the function of LovK as a photosensor suggests that signalling through LovK may be more complex than initially believed.

Control of cyclic di-GMP turnover in Synechococcus

A protein from the cyanobacterium Synechococcus elongatus, containing an N-terminal LOV domain and a C-terminal GGDEF-EAL domain (see Table 1, Figures 1 and 4) exhibits blue light-inducible phosphodiesterase activity in vitro⁷⁰. GGDEF and EAL are conserved domains that typically function as diguanylate cyclases and phosphodiesterases, respectively ⁷¹. The molecule synthesized and hydrolysed by these domains, cyclic di-GMP, is a second messenger involved in the regulation of biofilm, motility, and virulence. Blue light may therefore regulate multiple aspects of S. elongatus cell physiology by altering the concentration of cyclic di-GMP in the cell. Certainly, LOV-GGDEF-EAL proteins are among the most common LOV proteins in bacteria suggesting that light-dependent modulation of cytosolic cyclic-di-GMP may be a common phenomenon.

LOV domain signalling mechanisms

Clearly, LOV domains are versatile photoswitches. This raises an interesting mechanistic question: how do LOV domains regulate such a structurally disparate group of proteins? This question has been reviewed more generally for the PAS domain superfamily by Möglich and Moffat ⁷². Here we synthesize current data on LOV domain molecular structure and signaling mechanism. Biophysical and structural data on LOV domains centre on four very different proteins: plant/algal phototropin, the Bacillusσ_B regulator YtvA, Neurospora crassa VVD, and the light-dependent DNA-binding protein EL222 of Erythrobacter litoralis. We primarily focus on these systems.

Jα helix unfolding in phot-LOV2

Plant and algal phototropins contain two N-terminal LOV domains (LOV1 and LOV2) followed by a serine/threonine kinase (see Figure 1). Phototropin LOV1 and LOV2 have a high degree of sequence conservation but phylogenetically cluster into two distinct clades (Figure 2) and have unique kinetic properties ²⁰. LOV1 forms dimers in solution, suggesting it may have a role in full-length photoreceptor dimerization during light exposure ^73,74. There is also evidence that LOV1 modulates kinase photoregulation by LOV2 ⁷⁵. Nevertheless, the functional role of LOV1 in the regulation of phototropin kinase activity is not as well understood as LOV2, which is both necessary and sufficient for kinase activation ⁷⁶. Differences in lit state structure and dynamics between LOV1 and LOV2 almost certainly reflect the unique regulatory functions of each of these domains.

Gardner and colleagues were the first to demonstrate that structural elements outside the LOV domain core can undergo light-induced conformational changes ⁷⁷. They identified a C-terminal amphipathic helix (Jα) on Avena sativa (i.e. oat) phot1 LOV2, which is conserved in the phototropins and undocks and unfolds on illumination. The solution structure ⁷⁷ and crystal structure ⁷⁸ of A. sativa phot1 LOV2 domain show that Jα interacts with the β-sheet, burying hydrophobic surfaces on both (see Figure 5A). On photon absorption and adduct formation, the binding equilibrium between Jα and β-sheet is destabilized by approximately 4 kcal/mol ⁷⁹. This disruption of the Jα/β-sheet interaction occurs despite the absence of large changes in the overall structure of the LOV core^77,80–83. Mutational destabilization of the Jα/β-sheet interaction constitutively-activates the serine/threonine kinase⁸⁴ (Figure 6A). However, isolated phot1 LOV2 domain without the Jα helix can function as light-dependent inhibitor of the kinase activity in trans ⁷⁵, suggesting the kinase inhibitory function of phot1 LOV2 is encoded in the ≈100-residue LOV core and does not require Jα. The functional role of the structure N-terminal to the LOV core, which forms a helix-turn-helix motif that also docks against a hydrophobic region of the LOV β-sheet ⁷⁸, has yet to be determined.

Figure 5.

(A) Structural alignment of the dark-state structure of the YtvA LOV domain of B. subtilis (in green; PDB code: 2PR5) and dark-state phototropin 1 LOV domain of A. sativa (in red; PDB code: 2V0U). (B) Structural alignment of the dark state VVD LOV domain of N. crassa (in blue, PDB code: 2PD7) and the dark Phototropin 1 LOV domain of A. sativa (in red, PDB code: 2V0U).

Figure 6.

Structural models of signalling in LOV proteins (A) In phototropin-type signaling, cysteinyl-flavin adduct formation induces conformational change in the LOV2 domain (blue) that results in disruption of the interaction with the Jα helix (green). This leads to effector domain activation (yellow). Data from YtvA and bacterial LOV histidine kinases evidence models in which illumination of a LOV domain (blue) can induce conformational changes in an extended Jα helix (green), causing (B) tilting or rotational motion to activate the effector (in yellow). (C) In VVD, light activation leads to rearrangement of the N-terminal cap (in green) and a subsequent change in protein dimerization. (D) In the E. litoralis LOV-HTH protein, EL222, it has been proposed that light activation disrupts the interaction surface between the LOV domain and the HTH domain. This light-driven structural change leads to dimerization of the protein on DNA¹⁰⁶.

Although data demonstrating light-dependent undocking/unfolding of Jα are clear, the structural mechanism underlying the change in affinity between Jα and the LOV β-sheet is not firmly established. Mutational studies have revealed that residues located in the central β-scaffold are important for signal transduction and phototropin kinase activation ⁸⁵. In particular, a flavin-interacting glutamine residue of LOV β-strand 5 (Iβ) has been implicated in signal propagation from the β-sheet to Jα ^86,87 perhaps driven by glutamine side chain rotation on cysteinyl adduct formation ^32,82 (see Figure 3B). From molecular docking studies, it has been proposed that the phot LOV2 domain inhibits kinase activity by binding to the active cleft between the C- and N-terminal lobes of the kinase domain ⁸⁸. Light-induced unfolding of the Jα helix may result in dissociation of the LOV2 domain from the kinase catalytic region, and opening of the binding cleft to ATP ^75,88,89. However, it is unclear how conserved such a mechanism may be given the recent report that light-induced conformational change in the Jα helix varies substantially among phototropins ⁹⁰.

Another possible mechanism of phototropin kinase regulation by the LOV domain invokes a light-dependent change in the oligomeric state of LOV2. Such a model is supported by data showing an approximate 2-fold transient volume increase of LOV2 during light activation ⁹¹, which can be interpreted as a transient dimerization. One can thus envision a mechanism in which the Jα helix prevents dimerization of the LOV2 domains at the β-scaffold interface in the dark. Exposure of the β-scaffold on illumination could drive dimerization of the full-length protein and thus promote kinase activation. Light-dependent oligomerization in LOV domains is discussed further below in the context of the Vivid protein (Figure 5B and 6C).

Tilting and rotation – the YtvA-LOV model

The YtvA regulatory protein is composed of an N-terminal LOV domain linked by Jα to a C-terminal STAS domain ⁹². The photosensory LOV domain of YtvA was initially identified based on sequence homology to plant phototropins ^32,33. However, crystallographic, spectroscopic and biochemical data suggest it is unlike the phototropins in terms of signaling kinetics and light-dependent structural change ^66,93–96

The YtvA-LOV domain structure differs from phototropin LOV domains in both quaternary structure and in the orientation of Jα ⁹⁶, which has been suggested to form a coiled-coil ^59,63,97. Indeed, coiled-coil structural motifs are predicted to flank LOV domains in a range of LOV proteins. In both crystal and solution, isolated YtvA-LOV is reported to be dimeric^96,98,99 with Jα helices extending from the LOV-core dimer in a quasi-coiled-coil arrangement. As the YtvA Jα helix is connected to the LOV core domain by only a short loop and exhibits a more polar character, it has been proposed that it cannot readily bend and pack against the β-scaffold as described for the phot LOV2 domain. Consequently, the hydrophobic patch on the outer face of the β-scaffold is exposed to solvent and forms a surface for LOV-LOV dimerization^96,98.

Structural analyses under light and dark conditions have defined small structural rearrangements in YtvA-LOV on illumination that are dispersed across a large portion of the LOV domain. The Gln residue of β-strand 5 (Iβ) presumably undergoes a reorientation of its side chain ⁹⁶, consistent with phototropin LOV2 ^32,82 (Figure 3B). Light-induced structural change is proposed to drive scissor-like rotation of the two monomers by 4–5° relative to each other ⁹⁶ (Figure 6B). Additional biochemical and structural data are required to resolve how the LOV domain affects the STAS domain conformation and subsequent σ^B-dependent transcription.

Recent studies of artificial histidine protein kinases composed of the LOV domain of Bacillus YtvA fused to the kinase domain of Bradyrhizobium japonicum FixL provide evidence that histidine kinase activity is regulated by rotational movement in an α-helical coiled-coil (Jα) linking the domains⁹⁷. Although the extent of quaternary rearrangement in this engineered histidine kinase is not established, spectroscopic analysis of the natural LOV histidine kinase, LovK, of Caulobacter crescentus evidences a model in which small changes in overall tertiary/quaternary structure are sufficient to regulate protein activity¹⁰⁰. Additional structural and biochemical studies are necessary to determine the exact structural mechanism (or mechanisms) by which a LOV domain can regulate histidine kinases.

Light-driven dimerization: the VVD Ncap

Our review of the structural basis of signaling in phototropin LOV2 and YtvA-LOV has largely focused on conformational change in the β-scaffold and C-terminal Jα helix. Structural studies on the VIVID (VVD) LOV photosensor of the fungus Neurospora crassa reveal common features with phot-LOV2 and YtvA-LOV, and also provide insight into alternative modes of LOV signaling.

Proper regulation of N. crassa circadian rhythms requires VVD²⁸, which forms a feedback loop with the WC-1 protein to control diurnal gene expression. The VVD protein consists of a single LOV domain flanked by an N-terminal region of structure, termed Ncap (Figure 5B). Ncap is composed of an extended stretch of polypeptide containing an α-helix (aα), a β-strand (bβ) and a short hinge that connects the terminus to the LOV core ¹⁰¹, where it docks at a similar location on the β-scaffold as the C-terminal Jα-helix of phot-LOV2 domain ^77,78 (see Figure 5B), the αF helix in the Drosophila Period (Per) protein ¹⁰², and the N-terminal cap of photoactive yellow protein (PYP) ¹⁰³. These regions of structure have all been implicated in conformational switching during signaling ¹⁰¹.

On cysteinyl-flavin adduct formation and reorientation of the Gln residue in β-strand 5 (Iβ), the conformation of the hinge region connecting Ncap to the LOV core is altered ¹⁰¹. Specifically, illumination induces a shift in bβ of ≈2.0Å toward the LOV core, which disrupts packing of Ncap against the β-scaffold ^101,104, and results in transition from monomer to a rapidly exchanging VVD dimer ¹⁰⁴ (Figure 6C). Although conformational change in the Ncap is important for the function of VVD ¹⁰¹, the exact role of Ncap restructuring and oligomerization as it relates to WC-1 regulation are unknown. WC-1 has a very similar LOV domain to VVD, suggesting homo- and hetero oligomerization of these domains may have an important role in this process ^104,105.

Light-dependent DNA binding: Erythrobacter litoralis LOV-HTH

A recent study of the Erythrobacter litoralis LOV-HTH protein EL222 reveals a structural mechanism of regulation with features of both the phototropins and VVD ¹⁰⁶. EL222 is a light-regulated DNA-binding protein. In the dark, the protein is monomeric and the HTH domain is docked against the β-scaffold of the LOV domain, with Jα playing the role of a linker between domains (Figure 6D). The authors present evidence that light activation of the EL222 LOV domain disrupts the interface of interaction between LOV and HTH. This structural change ultimately leads to formation of a EL222 dimer on the DNA, the requisite oligomeric state required for HTH-DNA binding. Thus, light activation of a LOV domain is proposed to both promote undocking of HTH from the β-scaffold and to modulate the oligomeric state of EL222 in the presence of DNA.

LOV domains as a tool in synthetic biology

The versatility of LOV domains as a general light-sensing module has been demonstrated by recent successes in the engineering of synthetic LOV protein photosensors. As discussed above, a bacterial LOV domain has been swapped with a related (but non-photoresponsive) PAS domain to create a synthetic light-triggered histidine kinase ⁹⁷. The FKF1 LOV signaling protein of A. thaliana has been developed as a light-regulated switch that can initiate formation of heteromeric protein complexes in live cells that regulate processes including actin-remodeling and transcription ¹⁰⁷. Other groups have taken advantage of the large light-driven conformational change in A. sativa phot1 LOV2 ⁷⁷, and made photoactive fusions with the tryptophan repressor ¹⁰⁸, dihyrofolate reductase ¹⁰⁹, and the Rac1 GTPase ¹¹⁰ and demonstrated regulation of their activities in vitro or in vivo to varying extents. In these latter cases, there is no evidence that any of these proteins have ever been associated with, or regulated by, a PAS domain over the course of evolutionary history. The success of these engineered systems is even more surprising when one considers the relatively small number of constructs that were tested before the investigators obtained a synthetic light-regulated system. These results in LOV photoprotein engineering suggest there is a low barrier for the natural evolution of new protein function via domain swapping and help explain the great diversity of LOV proteins.

The recent successes of appending LOV domains to a broad range of protein platforms immediately suggest many possible biological and biotechnological applications of this tool ^111,112. For example, there are thousands of annotated sensor/effector proteins for which both the signal and physiological output are unknown. In a bacterial context, substitution of the sensor domain of a signaling protein of unknown function with a photo-inducible LOV domain can allow one to probe cellular function. Examples where this approach will be useful include sensor histidine kinases of two-component systems and ligand-binding transcription factors of one-component systems. In this scheme, a LOV domain (and light) could provide a generic signal that could be used to interrogate the physiological response downstream of any signaling protein. For the specific case of a PAS histidine kinase^48,97 this proof-of-concept has already been demonstrated. It is conceivable that LOV protein fusions could be engineered in bacteria to perturb dynamical subcellular processes regulated by the cytoskeleton, similar to what has been demonstrated for actin-based motility in mammalian cells using the synthetic LOV-Rac1 system ¹¹⁰.

For those wishing to use LOV domains to toggle the activity of their favorite protein, it may be the case that a LOV domain does not simply work “off the shelf”. In other words, the photochemical, structural or kinetic properties of LOV may need to be engineered to better affect the activity of the protein to which it has been appended. Recent studies have presented both rational design⁴⁹ and other mutagenesis/screening-based approaches^86,113–115 to generate LOV domains with more desirable structural or photochemical properties. Indeed, simple mutation of single residues or short sequences flanking the LOV domain can be sufficient to modulate the signaling lifetime by orders of magnitude ^69,111. The exchange of the natural FMN or FAD flavin cofactor for synthetic flavin analogs is an additional approach that may be used to engineer specific photochemical properties into a LOV domain ¹¹⁶.

Conclusions

The LOV domain is an ancient signaling module that has been adapted to serve a range of functional roles in eukaroytes, bacteria and archaea. Genetic and biochemical studies in plant and fungal systems in the mid-1990’s clearly established the role of LOV proteins in the regulation of photomorphogenesis and circadian biology. Since this time, detailed structural and biophysical studies of both isolated LOV domains and, in some cases, full-length LOV proteins have established models for the molecular basis of LOV signal transduction. The discovery of LOV domains in scores of bacterial genomes has motivated new biophysical analyses of LOV proteins and has offered the opportunity to explore novel functional roles for visible light in the bacterial kingdom. Although unexpected regulatory roles for LOV proteins and blue light have been documented in a handful of bacterial species, our understanding of blue light photobiology and LOV protein function in bacteria is in its infancy. Indeed, the sensory role of LOV domains in certain bacteria may be more complex, and integrate information about cellular stress, redox state and light.

Synthetic biology approaches have reinforced the idea that LOV domains provide a genetically-encodable and photoresponsive protein “switch” that can modulate the activity of a structurally-diverse range of signaling proteins and enzymes in both bacterial and eukaryotic systems. In addition, these engineered photosensors should permit investigators to probe nearly any biological or biochemical function with high temporal and spatial resolution. Future work in this exciting area of study promises to yield functional LOV fusions to new classes of proteins, as well as new LOV domains with modified photochemical and kinetic properties.

Text Box. Flavin photochemistry and photobiology: A brief history.

Flavin coenzymes absorb photons in the blue region of the visible spectrum, and can undergo dramatic chemical changes in response to light absorption. Well-known flavin photomodifications, first reported a half century ago, include the photolysis of the ribityl chain of flavin mononucleotide (FMN) to produce the N10 dealkylated species, lumichrome ^117,118, or the N10 methylated species lumiflavin ¹¹⁹. In both cases, ribityl photolysis is catalyzed by a highly-oxidizing triplet intermediate on the flavin isoalloxazine moiety formed via intersystem crossing from a singlet excited state. Triplet-dependent degradation of the ribityl chain can be quenched in the presence of alternative electron donors, resulting in simple photoreduction of flavin ¹¹⁹.

Although biological roles for flavin photolysis and photoreduction were not known at the time of these early photochemical studies, it had been postulated since the 1940’s that flavin coenzymes were involved in the regulation of blue light-dependent growth and developmental processes in plants ^120,121. It was not until 1993 that the first sequence for a plant blue-light photoreceptor, cryptochrome 1 (CRY1), was reported ¹²² and later shown to bind a flavin and pterin cofactors ^123,124. Although it was initially believed that the cryptochrome family of proteins would control many, if not all, blue light-dependent photoresponses in plants, a cry1 cry2 double mutant in Arabidopsis thaliana was shown to exhibit curvature toward unilateral light ¹²⁵ indicating another photoreceptor must be responsible for the phototropic response. This other photoreceptor was determined to be the serine/threonine kinase, phototropin 1 (phot 1), which employs LOV domains to sense blue light ^18,22.

Summary.

• LOV domains are a subset of the PAS domain superfamily that detect blue light through a flavin cofactor. These blue-light photosensory domains are conserved in all kingdoms of life.

• In bacteria, LOV domains are naturally found appended to a variety of signal transduction output domains including histidine kinases, diguanylate cyclases/phosphodiesterases, and DNA-binding domains.

• Bacterial LOV signaling proteins have been demonstrated to regulate processes including stress response, adhesion, cyclic-di-GMP synthesis, and virulence.

• LOV domains regulate the activity of their effector domains through a variety of structural mechanisms including light-dependent unfolding, tilting/rotation, and oligomerization.

• Recently, LOV domains have emerged as a powerful tool in protein engineering and synthetic biology. These modular photosensory domains have been used to artificially confer photoactivity on to a range of effector domains including transcription factors, actin remodeling proteins, and protein kinases.

Glossary definitions

Phototrophic

descriptor of an organism that utilizes solar energy (i.e. photosynthesis) to generate the energy required for cellular metabolism

Chemotrophic

descriptor of a non-photosynthetic organism that obtains energy for cellular metabolism by oxidizing organic or inorganic electron donors in its environment

Flavin

a common organic protein cofactor required for many biochemical redox transformations, which also functions as a blue light chromophore in LOV domains

Phototropism

response in which plants grow toward unidirectional visible light

Hypocotyl

the stem of a germinating plant seedling

Adduct formation

In the context of this review, adduct formation is the creation of a covalent bond between the conserved cysteine residue of the LOV domain and the 4a carbon of the flavin cofactor

Triplet intermediate

an electronic excited state intermediate of the flavin cofactor, through which the cysteinyl-flavin adduct is generated

Photomorphogenesis

a developmental process regulated by light

Histidine Kinase

a protein kinase, common in bacteria, which is phosphorylated at a single, conserved histidine residue

GGDEF/EAL proteins

Enzymes that function to synthesize (GGDEF) or degrade (EAL) the small signaling nucleotide, cyclic-di-GMP

Chromophore

a small-molecule cofactor or functional group that enables a protein to absorb light

Redox potential

A measure of the affinity of a particular chemical species for electrons; value is reported as an electrical potential (mV)

Oligotrophic

descriptor of an organism that lives in an environment that contains very low levels of nutrients

Biographies

Sean Crosson received his B.A. in biology from Earlham College in Richmond, Indiana in 1996. He completed his Ph.D. in biochemistry in the laboratory of Keith Moffat at The University of Chicago in 2002, and was a postdoctoral fellow in the laboratory of Lucy Shapiro at Stanford University School of Medicine from 2003 to 2005. Since 2006, he has been Assistant Professor at The University of Chicago in the Department of Biochemistry and Molecular Biology and the Committee on Microbiology. His research interests centre on signal transduction and cell regulation in bacteria.

After receiving his undergraduate degree in engineering from Université de Bretagne Occidentale, Julien Herrou began his graduate studies in 2006 in the laboratory of Dr. Camille Locht at Institut Pasteur de Lille, France. During this period, he studied the Bordetella pertussis virulence regulator, BvgAS, under the supervision of Dr. Françoise Jacob-Dubuisson and Dr. Rudy Antoine. Julien obtained his Ph.D. in Biophysics in 2008. Since 2010, he has been a postdoctoral fellow in Sean Crosson’s lab at The University of Chicago, USA where his research has focused on molecular mechanisms of general stress signaling in alphaproteobacteria.