A New Facet of Vitamin B₁₂: Gene Regulation by Cobalamin-Based Photoreceptors

Abstract

Living organisms sense and respond to light, a crucial environmental factor, using photoreceptors, which rely on bound chromophores such as retinal, flavins or linear tetrapyrroles for light sensing. The discovery of photoreceptors that sense light using 5′-deoxyadenosylcobalamin, a form of vitamin B₁₂ that is best known as an enzyme cofactor, has expanded the number of known photoreceptor families and unveiled a new biological role of this vitamin. The prototype of these B₁₂-dependent photoreceptors, CarH, is widespread in bacteria and mediates light-dependent gene regulation in a photoprotective cellular response. CarH activity as a transcription factor relies on the modulation of its oligomeric state by 5′-deoxyadenosylcobalamin and light. This review surveys current knowledge about these B₁₂-dependent photoreceptors, their distribution and mode of action, and the structural and photochemical basis of how they orchestrate signal transduction and control gene expression.

INTRODUCTION: LIGHT AND PHOTORECEPTORS

Sunlight is essential for life on Earth. Conversion of the energy contained in sunlight into chemical energy by photosynthetic bacteria, algae, and plants accounts for the majority of fixed biomass and molecular oxygen (1). A crucial environmental factor, the ability to sense and respond to light is vital for most living organisms. Human vision is based on the eye detecting visible light (or simply light), which corresponds to the 380-760 nm wavelength range of the electromagnetic spectrum (2, 3). Light can directly or indirectly signal diverse biological processes including DNA repair, circadian rhythms, taxis, development, morphology, physiology, and virulence, and also mediates biosynthetic reactions such as in vitamin D3 synthesis (4-13). The pervasive role of light in biology, however, comes with a price: light absorption by photosensitive biomolecules like porphyrins, chlorophyll, or flavins can generate highly reactive oxygen species (ROS) that cause photooxidative damage of DNA, proteins, membranes, and other cellular components, and ultraviolet (UV) light triggers formation of mutagenic thymidine dimer lesions in DNA (14-18). Consequently, various cellular strategies have evolved to avoid, minimize, or repair light-induced damage (11, 14-16, 19).

Light, beneficial or harmful, has to be detected and converted to a cellular signal to elicit the appropriate response. In all domains of life, this fundamental task is carried out by photoreceptor proteins (or photoreceptors) that directly sense light through a chromophore component. For example, specific tryptophans in the UVR8 (UV Resistance Locus 8) protein sense UV light (20, 21). Visible light photoreceptors, however, must rely on noncovalently or covalently associated nonprotein chromophores, since no protein component absorbs in this wavelength range. Known chromophores include retinal, flavin mononucleotide (FMN), flavin adenine dinucleotide (FAD), p-coumaric acid (4-hydroxycinnamic acid), and 3′-hydroxyechinenone ketocarotenoid, all of which absorb in the blue-green region (2, 3, 6, 22-28), and linear tetrapyrroles (biliverdins, phycocyanobilin, phytochromobilin, and phycoviolobilin), which absorb red/far-red light (29, 30). These unsaturated molecules have a conjugated π system with delocalized electrons, which favors light absorption (31). This type of system enables electron transfer, cis-trans isomerization about double bonds, or covalent bond formation-disruption to alter chromophore conformation and chemistry, which is then transmitted to the protein part of the photoreceptor to modulate function.

Photoreceptors have been classified into distinct families based on the chromophore and protein sequence/structure conservation. Rhodopsins have retinal as chromophore; cryptochromes, LOV (light-oxygen-voltage), and BLUF (blue-light utilizing FAD) sensors use FMN or FAD; phytochromes rely on linear tetrapyrroles; xanthopsins employ p-coumaric acid; and orange carotenoid protein has ketocarotenoid (12, 26, 28, 31, 32). Photoreceptors can be standalone proteins or part of larger proteins with multiple domains, each endowed with a defined activity, such as directing specific interactions (with proteins, DNA, RNA, membranes, small molecules) or enzymatic catalysis (kinases, diguanylate cyclases, phosphodiesterases). In multidomain photoreceptors, one or more modules sense the light input, undergo molecular changes, and transmit the signal to output effector domains to carry out a specific light-dependent function. The photosensory module frequently is an autonomous unit that can be swapped between different proteins, as is exemplified by the natural occurrence of a given type of photoreceptor module in functionally distinct proteins. In nature, this modularity allows for combinatorial module mixing as an effective strategy in the evolution of signaling and regulatory complexity (33). Biological engineers have also harnessed this modularity and created artificial genetic fusions of specific photoreceptor modules and proteins of interest, whose functions can then be optically monitored or controlled with high temporal and spatial precision (32). These advances have led to the vibrant field of optogenetics and the development of powerful tools and applications in cell biology and neurobiology (34-36).

The recent discovery of photoreceptors with a novel light-sensing chromophore, the vitamin B₁₂ derivative 5′-deoxyadenosylcobalamin (simply, adenosylcobalamin or AdoCbl), has enlarged the number of known photoreceptor families and unveiled a new biological facet of this vitamin (37). Significant progress has been achieved in understanding how this new type of photoreceptor works. High-resolution structures of the functionally relevant states have revealed the molecular architecture of the photoreceptor and provided detailed snapshots of its light-dependent mechanism of action (38). Some fascinating features in the photochemistry have also been observed (39, 40). These aspects will be covered here together with an overview of the discovery, functions, distribution, and evolution of these B₁₂-dependent photoreceptors.

VITAMIN B₁₂: CHEMISTRY AND BIOLOGY

Cobalamin Forms and Chemistry

Since its first discovery as the anti-pernicious anemia factor (41, 42), vitamin B₁₂ has seen a rich and storied scientific history. A decade after its first isolation in 1948 (43, 44), the description of its crystal structure by Hodgkin et al. (45) marked a heroic moment for small-molecule crystallography, revealing one of nature’s most complex cofactors in three dimensions, and its total chemical synthesis in the early ´70s capped a classic effort by Woodward, Eschenmoser, and coworkers (46). In parallel, extensive studies into the chemistry and biology of vitamin B₁₂ and its derivatives, collectively referred to as the cobalamins or often just “B₁₂”, revealed their intricate chemical structure, complex reactivity, and rich spectroscopic properties as well as their biosynthetic origins and roles in living systems (47-52). Their best-characterized biological functions in fatty acid and folate metabolism make them essential micronutrients for humans and other animals, although curiously not for plants or fungi. Fitting for B₁₂’s storied history, more recent studies have revealed new and unanticipated biological functions, for example in modulating the structure of microbial communities (53) and as a light sensor in light-dependent gene regulation, the main focus of this review.

Chemically, the principal feature of B₁₂ is its corrin ring, whose four pyrrolic nitrogens coordinate a central cobalt atom that is usually in the Co³⁺ or Co(III) oxidation state. The corrin pyrroline and pyrrolidine groups are adorned with methyl, acetamide, and propionamide groups, one of which links the ring to the 5,6-dimethylbenzimidazole (DMB) ribonucleoside tail characteristic of B₁₂ (Figure 1). In free B₁₂ at physiological pH, a nitrogen of this DMB base occupies the lower, or α, axial coordination site on the central cobalt, a conformation described as “base-on” or “DMB-on”. The octahedral coordination sphere of Co(III) is completed by an upper, or β, axial ligand, which can take a variety of forms depending on the B₁₂ derivative (Figure 1).

Figure 1.

Vitamin B₁₂ and derivatives. (Left) General chemical structure of B₁₂ in the base-on conformation with cobalt formally in the Co(III) state, the lower axial dimethylbenzimidazole ligand in blue, and the upper axial ligand denoted by “R” in red. (Right) Selected upper axial ligands and the corresponding B₁₂ forms are shown.

The two major biological forms of B₁₂, methylcobalamin (MeCbl) and AdoCbl (also known as coenzyme B₁₂), both have an alkyl group as the upper axial ligand: MeCbl has a methyl group (Me) and AdoCbl has a 5′-deoxyadenosyl group (Ado) that is bonded to the cobalt through its 5′-carbon (Figure 1) (54, 55). Thus, both MeCbl and AdoCbl feature a covalent Co-C bond, as revealed by their respective structures (56, 57), and represent rare examples of naturally occurring organometallic compounds. Not surprisingly, these Co-C bonds have intriguing chemical properties: chemically inert in the absence of light, their relatively low bond dissociation energies [reported as 32-40 kcal mol⁻¹ for MeCbl and 24-35 kcal mol⁻¹ for AdoCbl; (58-62)] allow them to be cleaved rather easily. This feature underlies the use of MeCbl and AdoCbl as cofactors in enzyme catalysis, although their exact chemical properties and biological functions are distinct. Other B₁₂ forms with different upper axial ligands are known (Figure 1). Cyanocobalamin (CNCbl or vitamin B₁₂) with a cyanide upper axial ligand is found frequently free in nature, but is nonfunctional and has to be converted in vivo to MeCbl or AdoCbl for biological use. In the absence of other ligands, B₁₂ in aqueous solution ligates a water molecule to form hydroxocobalamin or aquocobalamin (AqCbl), which predominates at physiological pH. Regardless of the upper axial ligand, free B₁₂ forms are generally found in the DMB-on conformation at physiological pH. Protonation of DMB at low pH leads to its replacement by water, yielding the “base-off” (DMB-off) conformation and a less rigid B₁₂ molecule with altered chemistry (63). Various proteins, including the B₁₂-dependent photoreceptors reviewed here and many B₁₂-using enzymes, bind B₁₂ with a His side chain replacing the DMB ligand, a B₁₂-binding mode known as “base-off/His-on” (64).

Under aerobic conditions, the central cobalt of the various B₁₂ forms is generally in the +3 oxidation state and relatively redox-inactive. Homolytic or heterolytic cleavage of the β-axial bond or reduction of AqCbl with a strong reductant can transiently generate cobalt in the Co²⁺/Co(II) or Co¹⁺/Co(I) states, the coordination number decreasing from 6 in Co(III) to 5 in Co(II) and 4 in Co(I). The one-electron-reduced Co(II) form of B₁₂, termed cob(II)alamin or cob(II), is paramagnetic and can be detected by electron paramagnetic resonance (EPR) spectroscopy. This form is relatively inert under anaerobic conditions but rapidly oxidizes in the presence of molecular oxygen. Reduction by an additional electron to Co(I) yields cob(I)alamin or cob(I), a supernucleophile and potent reductant that decomposes even under anaerobic conditions. The unique chemical properties of B₁₂ in each oxidation state are harnessed in various ways by different B₁₂-dependent enzymes.

Cobalamin Photochemistry

Cobalamins exhibit vibrant colors that originate from strong absorption in the UV-visible range, mostly from π-π* transitions. The spectral features of each form depend on the cobalt oxidation state and the nature of the upper and lower axial ligands (65-67). As a result of this absorption, cobalamins exhibit rich and complex photochemistry, which is again modulated by the axial ligands (61, 62). Although the photochemical properties of CNCbl and AqCbl have been studied (68, 69), the discussion here will center on the biologically relevant alkylcobalamins MeCbl and AdoCbl, whose light sensitivity has been known since their first isolation. In particular, photolytic cleavage of the AdoCbl Co-C bond has been regarded as a model system for its cleavage in enzyme active sites (70), and therefore has been studied extensively. Early studies focused on determining the products of light-induced decomposition (70-73). The underlying photochemical processes are controlled by electronic relaxation dynamics that occur on picosecond time scales. Paired ultrafast transient absorption spectroscopy and theoretical calculations have emerged as powerful approaches for the study of B₁₂ photochemistry and have provided detailed insight into the electronic processes and the intermediates following excitation (62, 74-78). Near-UV and visible light of wavelengths <530 nm (photon energies >40 kcal mol⁻¹) cleave the Co-C bond of both MeCbl and AdoCbl on a time scale of 10-100 ps. The quantum yields for MeCbl Co-C photolysis are wavelength-dependent, high at 400 nm but much lower at 522 nm, whereas near unit quantum yields are observed for AdoCbl over this wavelength range (61, 62). The initial cleavage events are generally homolytic, generating a caged alkyl radical:cob(II) pair that can either recombine to regenerate the Co-C bond or dissociate. Competition between geminate recombination and radical escape determines the ultimate photolysis yield (79). Net photolysis is reduced by solvent cage effects around Cbl and accelerated by compounds such as molecular oxygen that can intercept the radicals formed and suppress recombination pathways (74, 80).

The fates of the alkyl radical and cob(II) generated upon photolysis of alkylcobalamins depend on the environmental conditions. In the case of MeCbl, the highly reactive methyl radical rapidly reacts through complex pathways to form formaldehyde and smaller amounts of methanol, formic acid, and carbon dioxide under aerobic conditions, and under anaerobic conditions it forms a mixture of formaldehyde, methane, ethane, as well as smaller amounts of methanol and formic acid (72). Upon AdoCbl photolysis, the 5′-deoxyadenosyl radical (Ado•) rapidly reacts in the presence of molecular oxygen to form 5′-peroxyadenosine, which in turn decomposes to adenosine-5′-aldehyde and minor amounts of adenosine and adenine (39, 70). In the absence of oxygen, Ado• undergoes intramolecular addition of the radical to the adenine base and forms 5′-deoxy-5′,8-cycloadenosine (71, 73). The second major photolysis product of both MeCbl and AdoCbl, cob(II), is stable under anaerobic conditions, but is rapidly oxidized in the presence of oxygen to AqCbl, a reaction further enhanced by 5′-peroxyadenosine (70). The reactivity of these photolysis products underscores a central dichotomy of B₁₂ biology: although AdoCbl and MeCbl can mediate unique chemistry, their light sensitivity and reactivity require exquisite control to suppress inadvertent side reactions.

Cobalamins as Cofactors in Enzymes and Riboswitches

Although B₁₂ is essential in animals and in many prokaryotes, only some of the latter can synthesize it de novo (51, 52). Plants, fungi, and many prokaryotes bypass the need for B₁₂ by using only alternative enzymes or metabolic pathways that do not require this cofactor, and many other species contain both B₁₂-dependent and B₁₂-independent enzymes for the same reaction (48). Not surprisingly for such a complex molecule, the two B₁₂ biosynthetic pathways (aerobic or anaerobic) are amongst the most intricate known in nature. Each pathway involves more than 30 genes and steps subject to genetic and enzymatic controls. The earliest steps lead to the formation of a tetrapyrrole precursor, uroporphyrinogen III, which is also common to heme and chlorophyll biosynthesis (48, 50, 81-84). The two B₁₂ biosynthetic pathways then diverge, using different routes to convert the precursor into cobyrinic acid a,c-diamide, at which point they converge again to generate the B₁₂ corrin ring and to attach the DMB. Given the high genetic and metabolic cost, organisms that require B₁₂, even ones capable of its biosynthesis, have evolved mechanisms to acquire and assimilate trace amounts of exogenous B₁₂ and to salvage and regenerate the intracellular pool of this valuable cofactor (48, 85-89). Generally, external B₁₂ in diverse forms (including ones lacking DMB, the cobinamides) is captured and transported into cells using dedicated proteins and sophisticated mechanisms in bacteria and animals. In the cell, imported B₁₂ is usually processed (eg. decyanation of CNCbl, dealkylation) by specialized enzymes and escorted with the aid of chaperones (to prevent side reactions and dilution) to the AdoCbl or MeCbl generation/utilization pathways (48, 85-90). MeCbl is directly produced and used by methionine synthase/methyltransferase; AdoCbl is produced by ATP-dependent adenosyltransferases (ATRs) for use by other factors (86, 89, 91, 92). Three classes of sequence-unrelated ATRs are known in bacteria: CobA (BtuR or CobO), which acts in de novo AdoCbl biosynthesis, and EutT and PduO, which convert imported B₁₂ to AdoCbl (89, 92). The most widely prevalent and studied is PduO, whose human ortholog (MMAB) is associated with mitochondrial B₁₂ metabolism (89).

B₁₂-dependent enzymes have been extensively surveyed elsewhere (54, 55, 63, 93-97), but a few aspects relevant to the theme of this review are highlighted here. B₁₂ is used for three major types of reactions in biology: MeCbl is used for methyl transfer reactions, AdoCbl is used for radical-based transformations in mutases, dehydratases, deaminases and class II ribonucleotide reductases, and cobalamin without an upper ligand is used by reductive dehalogenases and by the tRNA-modifying enzyme epoxyqueuosine reductase, whose structures and mechanisms have only started to emerge (98-101). The activities of these enzymes, which bind to their cofactor with K_D in the nM-mM range (97), are based on Co-C bond cleavage and formation of highly reactive species. The modes of Co-C bond cleavage and the resulting reaction mechanisms, however, are very distinct.

MeCbl-dependent enzymes catalyze heterolytic cleavage of the Co-C bond to form highly reactive cob(I), which retains both bonding electrons, and a methyl carbocation that is transferred to a nucleophilic acceptor. The prototypical member of this class, MeCbl-binding methionine synthase (called MetH in bacteria), catalyzes transfer of a methyl group from 5-methyltetrahydrofolate (MeTHF) to homocysteine (Hcy) for methionine synthesis (54, 55). This large monomeric enzyme has four structurally and functionally distinct domains, one each to bind Hcy, MeTHF, B₁₂, and S-adenosylmethionine (102). The crystal structure of its isolated MeCbl-binding domain provided the first visualization of B₁₂ bound to a protein and of the base-off/His-on B₁₂-binding mode (64), which has since been observed in many other B₁₂-dependent enzymes. Structural and biochemical studies have provided a framework to understand the workings of this enzyme and the mechanistic safeguards that protect the reactive MeCbl cofactor (55, 103). MeCbl is sandwiched between a Rossmann-fold domain, which binds the B₁₂ lower face and the DMB tail, and a four-helix bundle, which caps the B₁₂ upper face and prevents inadvertent loss of the methyl group (64, 104). Thus, MeCbl is sequestered, requiring that the MeTHF and the Hcy binding domains of MetH displace the four-helix bundle to gain access during catalysis (55, 103).

In the presence of the corresponding substrate, AdoCbl-dependent enzymes catalyze homolytic cleavage of the AdoCbl Co-C bond to generate cob(II) and Ado•, which initiates a chemical transformation by abstracting a hydrogen atom from the substrate (93). At the end of the catalytic cycle, Ado• is regenerated and recombines with cob(II) to restore AdoCbl for another round of catalysis. AdoCbl is thus a radical reservoir in AdoCbl-dependent enzymes, allowing for reversible access to the working species, Ado•, as required. Examples of these enzymes include methylmalonyl-CoA mutase and its relatives, which interconvert branched and linear acyl groups through carbon skeleton transformations; aminomutases, which migrate the terminal amino groups of lysine or ornithine; and eliminases, which mediate migration and elimination of a hydroxyl or amino group. Not surprisingly, these radical-based reactions must take place under carefully controlled conditions to prevent side reactions and oxidative quenching of intermediates. To achieve these conditions, AdoCbl-dependent enzymes bind AdoCbl in a buried cavity between two domains: a substrate-binding domain, usually an (α/β)₈ triose phosphate isomerase barrel, and a B₁₂-binding domain that is either a Rossmann-fold domain homologous to that in MetH or a distinct domain that binds AdoCbl base-on (93). In this architectural context, Ado• generation and the ensuing chemical transformations occur in a controlled environment, thereby enabling the difficult radical-based chemistry of AdoCbl-dependent enzymes.

Beyond its function as an enzyme cofactor, B₁₂ can also bind to RNA-based regulatory elements called riboswitches (105, 106). The first such RNA element to be discovered, the AdoCbl riboswitch, spans the ≥200-nt 5′ untranslated region of mRNA from genes involved in B₁₂ metabolism (105, 107). Both the AdoCbl and the more recently discovered AqCbl riboswitches bind tightly (K_D≈10-250 nM) to the respective base-on B₁₂ form using similar RNA structural cores, with additional peripheral extensions in the AdoCbl riboswitch conferring cofactor specificity (108, 109). The less common AqCbl riboswitch was speculated to have evolved in marine bacteria due to their high light exposure, which would favor greater availability of AqCbl over the less light-stable AdoCbl (108). However, the photochemistry and photobiology of these B₁₂ riboswitches remain unexplored. Interestingly, B₁₂ riboswitches typically sense the presence of AdoCbl (or AqCbl) to regulate expression of proteins involved in B₁₂ uptake, biosynthesis, or use, whereas the photoreceptors reviewed here sense a light-dependent change in the state of B₁₂ to control expression of genes apparently unrelated to B₁₂ metabolism.

DISCOVERY OF CarA AND CarH, AND A ROLE FOR B₁₂ IN LIGHT RESPONSE

Even though the light sensitivity of AdoCbl and MeCbl was known for years, this property was thought to serve no physiological function. The light sensitivity is actually undesired in enzyme catalysis as it leads to cofactor inactivation. The discovery that B₁₂, specifically AdoCbl, can serve as a light sensor thus represented a major surprise. This new facet of B₁₂ biology emerged from studies in the Gram-negative soil bacterium Myxococcus xanthus, in which light induces a transcriptional response leading to carotenoid biosynthesis (11, 110). Carotenoids protect cells against photooxidative damage by quenching singlet oxygen (¹O₂) and other ROS (14, 16, 110, 111), and light has been shown to generate ¹O₂ and trigger the response in M. xanthus (112). Yet, despite its well-established light response, genetic and bioinformatic studies failed to identify conventional photoreceptors in M. xanthus (11).

A resolution to this conundrum and the earliest hints for a role of B₁₂ in a cellular response to light came from the identification of the first putative B₁₂-binding transcription factors: the paralogous M. xanthus repressors of carotenoid (car) gene expression CarA and CarH, encoded by the two most downstream genes of the light-inducible carB operon, which groups all but one of the structural genes for carotenoid biosynthesis (Figure 2) (11, 110, 113). Their N-terminal sequences resembled the DNA-binding domains (DBD) of MerR transcription factors (114-117), suggesting a direct involvement in regulating gene expression (118). Their C-terminal segments were noteworthy, as they resembled the MeCbl-binding domain of MetH in sequence, size, predicted secondary structure, and in the presence of the base-off/His-on B₁₂-binding motif Figure 2b (113), first described by Drennan et al. (64) and until then found only in enzymes that use B₁₂ as a cofactor. As predicted, CarA did bind to B₁₂ but, paradoxically, functioned independently of the cofactor. Whether or not B₁₂ was present, CarA could bind to DNA using its autonomous, N-terminal MerR-type DBD (119, 120), and dimerize via its C-terminal module (37, 121, 122). To repress transcription, the CarA dimer appears to bind to a bipartite operator in a stepwise cooperative manner, first to a high-affinity palindrome and then to a second similar but lower-affinity one downstream, which overlaps with the −35 element of the carB promoter (P_B), to block RNA polymerase access to P_B (Figures 2a and 3a) (123). Light abolishes CarA-mediated repression by inducing expression of CarS, an antirepressor that structurally mimics operator DNA to bind tightly to the CarA DNA recognition helix and physically sequester it from operator binding (Figure 3a) (122-125).

Figure 2.

Experimentally studied CarH and CarA proteins. (a) Genomic context of the carH/carA genes in M. xanthus (Mx), T. thermophilus (Tt) and B. megaterium (Bm). Other genes and their corresponding products are as follows (carotenoid synthesis genes are in blue): crtE (E), farnesyltransferase; crtI (I), phytoene desaturase; crtB (B), phytoene synthase; crtD (D), hydroxyneurosporene dehydrogenase; crtC (C), neurosporene hydroxylase; ?, putative carotenoid biosynthesis protein; crtYc (Yc), crtYd (Yd), components of a heterodimeric lycopene cyclase; white arrow, predicted acyltransferase domain-containing protein; phr, DNA photolyase; ldrP, CRP/FNR family transcriptional regulator. (b) Domain architecture of CarH/CarA proteins. Numbers delimiting the domains are from the crystal structure of CarH from Tt. Characteristic motifs (x, any residue) are shown below for each protein (size in residues is indicated in brackets).

Figure 3.

Light-dependent gene regulation mechanisms by B₁₂-binding transcription factors. (a) M. xanthus CarA and CarS. Cooperative binding of CarA dimers to a bipartite operator overlapping the −35 promoter region blocks access to RNA polymerase and represses transcription in the dark. Light induces production of CarS, which sequesters CarA, prevents operator DNA-binding, and activates transcription. (b) M. xanthus/T. thermophilus CarH. CarH monomers in the apo form bind to AdoCbl (filled blue asterisk) to form stable AdoCbl-bound CarH tetramers in the dark, which bind to operator DNA overlapping the −35 promoter region to block access to RNA polymerase and repress transcription. Light disrupts AdoCbl-CarH tetramers to monomers by photolysing bound AdoCbl (unfilled blue asterisks and blue circles correspond to products after photolysis described in the text), loss of operator-binding, and activation of transcription. (c) B. megaterium CarH. ApoCarH tetramers, which do not bind DNA, yield AdoCbl-bound tetramers that bind to operator DNA overlapping the −35 region of P_carH (another operator that overlaps the −10 region of P_crt is not shown) to repress transcription in the dark. Light disrupts AdoCbl-CarH tetramers to dimers (by photolysing bound AdoCbl), abolishing operator binding and relieving repression. (d) R. capsulatus CrtJ and AerR. CrtJ dimers repress transcription in the dark by binding to two sites overlapping the −10 and −35 promoter regions. On exposure to light, binding of AqCbl (produced by AdoCbl or MeCbl photolysis) to AerR enables its association with CrtJ to disrupt CrtJ dimers and DNA-binding, activating transcription.

Repression of P_B in the dark was abolished on deleting carA but, intriguingly, it was restored on supplying B₁₂ exogenously (M. xanthus cannot synthesize B₁₂ de novo but has the cellular machinery for its uptake and assimilation), as long as the CarA operator was intact (113, 121). Detailed genetic analysis in a carA-deleted genetic background demonstrated that downregulation of P_B by B₁₂ required CarH and its B₁₂-binding motif (121). These findings thus revealed that light-induced carotenogenesis in M. xanthus was regulated by two parallel and distinct pathways, orchestrated by a pair of paralogous factors, of which one (CarH), but not the other (CarA), required vitamin B₁₂ (121). The study also established a firm link between B₁₂ (the form used was unknown at this point) and CarH in light-dependent gene regulation and marked the discovery of a role for B₁₂ as a light sensor.

M. xanthus CarA and CarH were the only known transcription factors with a B₁₂-binding domain fused to a DBD until a later study reported a protein with a similar domain architecture acting in light-induced carotenogenesis in the Gram-positive soil bacterium Streptomyces coelicolor (126). Molecular details on how this protein, named LitR (for light-induced transcription regulator), functions and whether or not it requires B₁₂ remain unaddressed. The subsequent outpouring of microbial genome data revealed many bacterial species with genes for CarA/CarH homologs of unknown function, often amidst ones for carotenogenesis or light-related responses (37, 121). Studies of a selected few of these are now revealing the molecular basis for their distinct modes of action.

MOLECULAR MECHANISMS OF B₁₂-BASED PHOTORECEPTORS

Adenosylcobalamin in Light-dependent Gene Regulation by CarH

The mechanistic basis for the combined action of CarH and B₁₂, the link to light, and the specific form of B₁₂ involved emerged in a pioneering study reported in 2011 (37). Domain swap and bacterial two-hybrid analyses firmly established that the C-terminal domain of CarH conferred the B₁₂ dependence in vivo, with B₁₂ controlling the oligomeric state of this domain (37). In-depth analysis of the molecular interplay between CarH and B₁₂ was, however, thwarted by the inability to purify native M. xanthus CarH. The existence of a homolog of unknown function in Thermus thermophilus (a Gram-negative bacterium that can synthesize B₁₂ de novo) not only enabled comparative studies but also, since it was easily purified, its biochemical, hydrodynamic, structural, and photochemical characterization (37-40, 127). The T. thermophilus protein and a chimera with its DBD replaced by that of M. xanthus CarH (which could functionally replace M. xanthus CarH in vivo) exhibited, in vitro, B₁₂-dependent oligomerization and DNA-binding in the dark that was disrupted by light. Surprisingly, the form of B₁₂ required turned out to be AdoCbl, despite the similarity of the CarH B₁₂-binding domain to that in MeCbl-dependent MetH (37). Only AdoCbl rapidly transformed the monomeric apoprotein into a stable tetramer that has 1:1 AdoCbl:CarH stoichiometry and binds tightly to operator DNA (K_D≈70 nM) (37, 38). Photolysis of AdoCbl-CarH by exposure to near-UV, blue, or green light swiftly provoked disassembly of the tetramer to monomers, with concomitant loss of operator-binding (Figure 3b) (37, 127). Interestingly, T. thermophilus CarH retains the photolysed AdoCbl as a stable adduct, refractory to exchange with fresh AdoCbl (38, 40). That AdoCbl is also the B₁₂ form required by M. xanthus CarH was evident when the deletion of the only gene encoding an ATR (of the PduO type) in this bacterium, which would cause an abrogation of intracellular AdoCbl generation, resulted in the loss of B₁₂-dependent repression of carotenogenesis. Moreover, even in the absence of the CarS antirepressor, light still relieved repression by M. xanthus CarH. These studies thus established thatM. xanthus CarH also directly senses light using AdoCbl to regulate gene expression (37).

The location of the gene for T. thermophilus CarH next to one for carotenogenesis, as in M. xanthus (Figure 2a), suggested a function in light-regulated carotenoid synthesis, which was indeed demonstrated (128). However, while M. xanthus CarH acts at a promoter that drives expression of the genes for carotenoid synthesis as well as its own, the one in T. thermophilus represses its own promoter and that of LdrP, a transcriptional activator of the carotenogenic gene transcribed in the opposite direction (Figure 2a) (128). In another variation recently found in Bacillus megaterium (a Gram-positive bacterium that can synthesize B₁₂ de novo), the CarH homolog regulates its own expression and that of target carotenogenic genes present at an unlinked genetic locus (Figure 2a) (129). Here, the active “dark” state AdoCbl-bound repressor is again a tetramer but its inactivation by light yields a dimer rather than a monomer (Figure 3c) (129).

Another Mode of Gene Regulation Dependent on B₁₂ and Light: AerR

A new twist to B₁₂-dependent light-regulated gene expression was reported with AerR, a small, standalone B₁₂-binding protein found in Rhodobacter capsulatus and related α-proteobacteria (130, 131). AerR (also termed PpaA) is an antirepressor of CrtJ (the aerobic repressor of carotenoid (crt) gene expression; also termed PpsR), a dimeric, redox-regulated repressor of genes for carotenoid, heme, and bacteriochlorophyll biosynthesis and for structural proteins of the light-harvesting complex II (130, 132, 133). Like CarA, the CrtJ dimer binds cooperatively to two tandem palindromes in the promoter region to repress transcription, and requires an antirepressor (AerR) for derepression in the light (132). But unlike the CarA antirepressor CarS, AerR is not transcriptionally activated by light but instead binds to AqCbl, which is produced in the light by photolysis of AdoCbl (or MeCbl). AqCbl-bound AerR then targets CrtJ to disrupt its oligomerization, DNA-binding and repressor activity (Figure 3d) (130, 131). As in T. thermophilus CarH, photolysed B₁₂ is tightly bound to AerR. Two His in AerR were implicated in this tight binding, one corresponding to the base-off/His-on B₁₂-binding motif and another that is not conserved in its homologs (130).

The CarH Tetramer: Structural Comparisons with Known DNA- and B₁₂-binding Proteins

A series of crystal structures of the functionally relevant states of T. thermophilus CarH and structure-based mutational analysis by our groups recently provided detailed insight into the architecture and the light-dependent conformational changes of the CarH photoreceptor (38). The “dark” AdoCbl-CarH tetramer is a dimer of two dimers, each with a striking head-to-tail arrangement of two monomers. The N-terminal domain of each monomer structurally matches the winged-helix DBD found in CarA (Figure 4a) (119) and MerR-type transcription factors (114-116). It has the canonical DNA recognition α-helix, with a conserved RxWERRY motif in CarH, its homologs, and CarA (Figure 2b), and the β-hairpin wing. The CarH DBD conserves most of the residues that contact DNA in CarA (119) and in MerR proteins (114-116), but employs a distinct DNA-binding mode described in the next section. The DBD adopts different orientations relative to the C-terminal domain due to a flexible linker of ~6-residues observed in the T. thermophilus CarH structure. Remarkably, the C-terminal AdoCbl-bound light-sensing module is structurally more similar to the MetH MeCbl-binding domain (64) than to any known AdoCbl-binding protein (Figure 4a). As in MetH, this CarH module has a four-helix bundle, which contacts the upper face of the Cbl, followed by a five-stranded α/β Rossmann domain that binds the lower face in the base-off/His-on mode using the conserved His of the Glu/Asp-x-His or E/DxH (where x is any residue) B₁₂-binding motif (Figures 2b and 4a). But in contrast to the MetH module, in which only the Me of MeCbl can fit snugly as the upper axial ligand, the CarH B₁₂-binding pocket accommodates the far bulkier Ado group of AdoCbl. Also, AdoCbl-CarH is a tetramer whose B₁₂-binding modules act as light sensors (37, 127), whereas the large multidomain MetH is a monomer (102) whose B₁₂-binding module suppresses undesired light-induced side reactions (104).

Figure 4.

Dark state AdoCbl-bound CarH compared to structurally similar DNA- and B₁₂-binding domains. (a) CarH protomer structure [Protein Data Bank (PDB) accession code 5C8D] showing the modules for DNA-binding (compared with that in CarA, top left; PDB accession code 2JML) and B₁₂-binding (compared to that in MetH, bottom left; PDB accession code 1BMT). (b) Close-up of the B₁₂-binding site showing residues contacting the upper axial ligand (in cyan) of MeCbl in MetH and cobalt-coordinating His (H759). (c) Close-up of the B₁₂-binding site showing residues contacting the upper axial ligand (in cyan) of AdoCbl in CarH and cobalt-coordinating His (H177). For a better view, the orientation in b and c is slightly different from that in a. (d) Comparison (in stereo) of the orientations of the AdoCbl Ado group in CarH, in free AdoCbl, and in selected enzymes whose structures have been determined with intact AdoCbl: IcmF (PDB accession code 4XC6) and related mutases; ornithine-aminomutase (OAM) in its resting state (PDB accession code 3KP1).

The CarH structure reveals how an enlarged B₁₂-binding pocket and the substitution of four hydrophobic residues in the four-helix bundle, Phe708, Leu715, Val718, and Val719 in MetH to Trp131, Val138, Glu141, and His142, respectively, in CarH (Figure 4b,c), allowed for the repurposing of the MetH MeCbl-binding module into one that binds AdoCbl (38). These changes produce a larger cavity, a more polar environment, and a H-bond between Glu141 and the Ado ribose group. Moreover, the Phe in MetH is directly above the Me group of MeCbl to protect it from photolysis (104), whereas the larger side chain of the equivalent Trp in CarH is to the side of the Ado group in AdoCbl, contacting its ribose moiety. Although the AdoCbl Co-C bond length in CarH (2.2 Å; (38)) is similar to that in the free form or in the substrate-free resting state of AdoCbl-based enzymes (93), the relative orientation of the Ado group differs (Figure 4d). Consistent with their importance, the aforementioned Trp, Glu, and His form a conserved W-9(x)-EH motif in CarH homologs, and mutating any of these residues impairs AdoCbl-binding and tetramerization (38). Tellingly, M. xanthus CarA, which binds to B₁₂ but does not require it for oligomerization or function (37), retains the Glu and His but lacks the Trp and four other adjacent residues (Figure 2b).

The presence of the Ado group maintains the AdoCbl-bound protomer in an extended conformation, which allows the helix bundle of one monomer unit to pack against the Rossmann fold of another unit to form a head-to-tail dimer. The dimer interface between the head-to-tail monomeric units is extensive, with over twenty H-bonds and salt bridges, some involving the Ado group (Figure 5a). Tetramers form by packing two head-to-tail dimers in a staggered manner (Figure 5b). Assembly of the tetramer from apoprotein monomers upon adding AdoCbl is so rapid and favorable that dimers are detected only if the dimer-dimer interface is disrupted by mutation (37, 38, 127). Residues at the dimer and tetramer interfaces are not highly conserved, but key interactions appear to be retained. For example, an Arg-Asp salt bridge between Arg176 in one monomer and Asp201 in the other is crucial for dimerization (Figure 5a), as its disruption produces monomers, but can be swapped for an Asp-Arg or a Glu-Arg pair (38). In M. xanthus CarH, this pair is already swapped in the form of a Glu-Arg pair, but is also indispensable for oligomerization, and can be replaced by Arg-Glu in vivo (J. Fernández-Zapata, M.C. Polanco, S. Padmanabhan & M. Elías-Arnanz, unpublished findings). It is intriguing that CarA, unlike CarH, forms dimers independently of B₁₂ or light (37), and some AdoCbl-dependent CarH homologs, such as that in Bacillus megaterium, are active as repressors in the tetrameric form but become dimers instead of monomers in the light (Figure 3) (129). Structures of these homologs can provide insights into why their oligomerization behavior differs from that observed for T. thermophilus CarH.

Figure 5.

The CarH tetramer and its unexpected DNA binding mode. (a) The head-to-tail packing of the two protomers in a CarH dimer, with the helix bundle colored yellow, the AdoCbl-binding domain in green, and the two DBDs in cyan. On the right is a close-up of the extensive interface of the head-to-tail dimer with several H-bonds and ionic interactions indicated. D201 and R176 indicate residues whose interaction was shown to be crucial by mutational analysis. (b) The CarH tetramer of two head-to-tail dimers with the four DBDs (in purple, cyan, light cyan, and dark blue) [Protein Data Bank (PDB) accession code 5C8D]. Two alternative views of the tetramer are shown to better appreciate the complex quaternary structure of CarH. The view on the left shows the distribution of the DBDs on the protein surface (DBDs shown in pink and dark blue correspond to the head-to-tail dimer at the back, whose B₁₂-binding domains are shown in gray). (c) CarH tetramer (colored as in panel b) in complex with DNA (PDB accession code 5C8E) is shown. Three DBDs are reoriented and contact three direct repeats (DR) in the DNA sequence, depicted schematically below (structures in b and d are redrawn versions of those previously reported in (38)). (d) Direct repeats recognized by CarH at P_carH in T. thermophilus (Tt) determined from the structure of the complex, and those inferred in M. xanthus (Mx) and B. megaterium (Bm) from footprinting data and sequence inspection. Base pairs covered by the recognition helix, as deduced from the CarH-DNA structure, are shaded. The −35 promoter region is shown in red. In Mx P_B, the two CarA binding site palindromes are underlined. The repeats in Bm P_crt and its −10 region (in purple) correspond to the noncoding strand. The DNA sequence logo for a probable consensus CarH direct repeat recognition site from the sequences above is shown (bottom).

CarH´s architecture of a tetramer formed by two head-to-tail dimers is rather unusual for a transcription factor. This arrangement results in neighbouring DBDs being pointed away from each other (Figure 5b). Very few structures of transcription factors with such a head-to-tail assembly are known. One example, Bacillus subtilis GabR, has an N-terminal winged-helix domain connected by a long, flexible 29-residue linker to a C-terminal domain with an aminotransferase (AT) family fold that binds to pyridoxal 5′-phosphate (PLP), the vitamin B₆ coenzyme (134, 135). With or without bound PLP, two AT-domains pack head-to-tail to form a stable GabR dimer (134). The GabR-DNA complex structure is not known but it has been proposed that a GabR dimer binds to two direct ATACCA repeats, separated by a 29-bp AT-rich spacer that is bent in the complex and includes the −35 promoter region (134, 135). The DNA-binding mode of CarH, however, is distinct.

The Surprising Mode of DNA-binding by CarH

The crystal structure of the CarH-DNA complex revealed a surprising mode of DNA-binding, with three out of the four DBDs in the CarH tetramer contacting three adjacent sites (Figure 5c). This binding mode was corroborated by hydroxyl radical footprinting and systematic analysis of CarH-binding to mutant operators (38). The CarH tetramer, alone or DNA-bound, has the same overall architecture except for a reorientation of three DBDs to contact three adjacent direct 11-bp repeats, the central one containing the TTGACA of the −35 promoter element (Figure 5c, d) (38). H-bonds and electrostatic interactions position the recognition helix of each DBD (involving the conserved RxWERRY motif) in the major groove, and the β-hairpin wing in the minor groove. Mutating DNA contacts in any two repeats or all three, but not in just one, abolishes binding by CarH. Tight DNA binding requires a tetramer, since mutants that form only dimers bind DNA with reduced affinity and cooperativity, and light-induced monomers do so poorly (37, 38).

It is intriguing that the T. thermophilus CarH tetramer uses three out of four DBDs to contact three adjacent DNA sites and, accordingly, only three direct repeats can be identified in its operator segment (Figure 5d) (38). Curiously, within the ~50 bp M. xanthus CarH operator (which also spans that of CarA) mapped using the chimera with the M. xanthus CarH DBD fused to the T. thermophilus CarH AdoCbl-binding domain mentioned earlier (37), five direct repeats similar to those of T. thermophilus can be identified (Figure 5d). Future studies will help establish the actual number of repeats and DBDs required for DNA-binding and function in this case. Interestingly, AdoCbl-bound B. megaterium CarH has been reported to bind to two DNA sites located at unlinked genetic loci (Figure 2a). One is an ~45-bp DNA site that overlaps with the −35 element of its own promoter (P_carH), and the other is about 10 bp shorter and includes the −10 element of the promoter for the carotenogenic operon (P_crt); in both cases, an imperfect interrupted palindrome with two 6-bp half-sites separated by 16 bp was proposed as the binding site (129). However, four and three similar direct repeats can be discerned in the longer and shorter sites, respectively (Figure 5d). Thus, the B. megaterium CarH tetramer could conceivably target direct repeats, as does T. thermophilus CarH, rather than a palindrome as proposed, although this possibility remains to be established experimentally. The activities of signaling proteins such as photoreceptors often depend dramatically on the properties of the linker between the receptor and effector modules (136). The intrinsically unstructured linker between the CarH DBD and the light-sensing oligomerization module, whose length and sequence vary in different homologs (~6 residues in T. thermophilus, ~20 in M. xanthus, ~14 in B. megaterium), could underlie the flexible DNA-binding mode of CarH, an aspect that needs to be explored. The finding that the B. megaterium homolog can bind to a site that overlaps with the −10 promoter element (Figure 5d) suggests that CarH can target the −35 as well as the −10 promoter elements to achieve transcriptional repression. The CarH proteins that have been studied, although from distantly related bacteria, nonetheless appear to recognize similar 11-bp repeats, with highly conserved bases at positions 5 (T), 8 (A), 9 (C) and 10 (A) (Figure 5d). Occurrence of these conserved bases in adjacent repeats could thus help identify other CarH-regulated promoters.

Interestingly, in binding as a tetramer to direct repeats, CarH conserves most of the DNA contacts of CarA and MerR proteins, even though the latter bind as dimers to (pseudo)palindromic sequences (114-117, 119, 123). But unlike MerR factors, CarH and CarA are only known to repress transcription by binding to sites overlapping a promoter with optimal spacer length (38, 119, 121, 123). Whereas CarA does this as a dimer, and independently of B₁₂, by stepwise cooperative binding, T. thermophilus CarH only represses as an AdoCbl-bound tetramer using a DBD from one head-to-tail dimer to bind the −35 promoter element in the central repeat, and the DBDs from the other dimer to bind the outer repeats (Figure 5c) (38).

Structure of Light-exposed CarH Suggests Mechanism of Photoregulation

The structure of light-exposed AdoCbl-CarH (Figure 6a) clearly revealed bound Cbl without an Ado group, and provided molecular insights into how light triggers tetramer disassembly (38). In the dark, the Ado group functions as a “molecular doorstop” by stacking against Trp131 (of the W-9(x)-EH motif) (Figures 4c, 6b). This interaction maintains the CarH protomer in the extended conformation required to assemble the head-to-tail dimer and thereby the tetramer. Once the Ado group dissociates upon exposure to light, Trp131 moves into the void caused by the loss of the Ado group, leading to a sizable shift (>8 Å) of the four-helix bundle relative to the Rossmann fold (Figure 6a). The resulting bent conformation of the protomer disrupts the head-to-tail dimer interface, causing tetramer collapse and loss of DNA binding. The helix bundle shift repositions Trp131 and the adjacent His (His132), the latter of which ends up as the upper axial ligand of the Co in Cbl, forming a very stable bis-His ligation (Figure 6b,c) (38). The structure of photolysed CarH provided the first visualization of a bis-His ligation involving Cbl. The very stable bis-His bond explains why fresh AdoCbl cannot replace Cbl in photolysed T. thermophilus CarH (38, 40). The biologically expensive Cbl is thus securely retained for recovery and reuse in vivo, although the details of the recovery process are not yet known. The lower axial His is the most strictly conserved residue of the canonical B₁₂-binding motif and is indispensable for cofactor binding and activity (37, 121, 129). By contrast, the upper axial His is not strictly required for tetramer assembly or its light-induced collapse. This His is conserved in thermophilic (38) and many other CarH homologs, but not all (for example, it is a Glu in M. xanthus and B. megaterium CarH; Figure 2b). A similar bis-His cobalt ligation was proposed in AerR, but the expected upper axial His is neither conserved nor is it in the putative four-helix bundle (130). Whether other residues can coordinate the Co on the upper site or another ligand such as water is involved remains to be explored.

Figure 6.

Molecular basis of photoregulation by CarH. (a) Structure of light-exposed CarH [solid, Protein Data Bank (PDB) accession code 5C8F] with the arrow indicating the major shift in the helix bundle relative to the dark structure (transparent, PDB accession code 5C8D). (b) Close-up of the AdoCbl-binding site highlighting the role of the upper axial Ado group (in cyan) of AdoCbl as a molecular doorstop in the dark state. The arrow indicates the helix bundle shift that occurs on exposure to light leading to relocation of W131 and the other indicated residues. The nonconserved E129 was suggested to be involved in forming the bis-His adduct by deprotonating His132 based on homology modeling and molecular dynamics studies (40), but this may be unlikely given its positioning in the crystal structure. (c) Close-up of the bis-His Co coordination in light-exposed CarH. (d) Scheme of the proposed mechanism for CarH photolysis adapted from (39), with additional details from the high-resolution structures (38) and apparent rates, k_app, from transient kinetics data (40). The chemical structure of 4′,5′-anhydroadenosine, the product of AdoCbl photolysis in CarH (39), is shown in the middle (and to the right) of the scheme.

Photochemical Basis of CarH Function

The use of AdoCbl as a light sensor posed additional interesting questions regarding the photochemistry involved in the process. In particular, it seemed highly counterintuitive that a cell response to mitigate photooxidative damage by highly reactive ROS would rely on AdoCbl photolysis, as this process usually generates Ado•, which can itself produce ROS and trigger radical-induced cell damage. This apparent paradox prompted studies aimed at characterizing the products of CarH photolysis and the underlying photochemistry. In one study, the products of AdoCbl-CarH photolysis formed under aerobic or anaerobic conditions were analyzed using liquid chromatography-mass spectrometry and UV–Vis, EPR, and NMR spectroscopies after sample workup on the time scale of minutes to hours (39). The other study involved transient kinetics analysis over femtoseconds to seconds using ultrafast spectroscopy, in which photolysis triggered by laser pulse excitation was followed by time-resolved acquisition of absorption spectra (40). These studies indicated that CarH orchestrates a distinct photochemistry to safeguard the use of AdoCbl as a light sensor and set the stage for further studies to elucidate the details.

Photolysis of free AdoCbl generates Ado•, which then forms a set of well-characterized products described earlier (5´-peroxyadenosine, adenosine-5′-aldehyde, and minor amounts of adenosine and adenine in the presence of oxygen, and 5′-deoxy-5′,8-cycloadenosine in the absence of oxygen) (71, 72). However, photolysis of CarH-bound AdoCbl did not yield any of these products but instead generated 4′,5′-anhydroadenosine (Figure 6d), which had not been previously detected as a photolysis product of AdoCbl (39). 4′,5′-anhydroadenosine had been observed as an inactivation product of some AdoCbl-dependent enzymes, the relevance and mechanism of which remain unclear (137, 138). More importantly, it was known to be a minor product of AdoCbl thermolysis via β-H elimination from the ribose C4′ of Ado•, primarily in viscous solvents when radical escape is slowed by solvent cage effects (139). The observation of 4′,5′-anhydroadenosine as the sole photolysis product thus directly suggested that CarH alters the photochemistry of AdoCbl and that the reactive Ado• is not released free into solution. In principle, two mechanistic routes could lead to formation of 4′,5′-anhydroadenosine (Figure 6d). Photolysis could occur by Co-C bond homolysis to form cob(II) and Ado•, followed by β-H transfer from Ado• to cob(II) to form 4′,5′-anhydroadenosine and cob(III) hydride (hydridocobalamin), or by Co-C bond heterolysis to form cob(III) and an Ado⁻ anion, which could undergo β-hydride elimination to again yield 4′,5′-anhydroadenosine and cob(III) hydride (39). Under aerobic conditions, cob(III) hydride would get oxidized to the final bis-His-ligated cob(III) species observed in the crystal structure. Thus, either pathway yields the same end products, requiring characterization of the intermediates to distinguish between the pathways.

Analysis of the photochemical mechanism (over fs to s) of CarH-bound AdoCbl using ultrafast spectroscopy led to the conclusion that Co-C bond photolysis occurs primarily by a heterolytic mechanism (Figure 6d), providing further evidence that CarH alters the photochemistry of AdoCbl (40). Although a small amount of homolytic cleavage was also observed, the resulting cob(II):Ado• radical pair underwent quantitative recombination. Thus, this study suggested that the photochemistry of CarH-bound AdoCbl is altered in two ways: formation of Ado• is suppressed by activation of a heterolytic Co-C bond cleavage pathway, and any Ado• formed through homolytic Co-C bond cleavage is not released into solution (40).

The mechanistic basis for this altered photochemistry is still largely unclear. The UV-visible spectrum of CarH-bound AdoCbl, compared to free AdoCbl, exhibits some changes in the α and β absorption bands, including a shift of the α band to higher energy (38-40). Although these changes could point to a specific effect of the binding environment on the photochemistry, more studies are needed. In addition, CarH might exert a cage effect following photolytic cleavage of the AdoCbl Co-C bond to suppress dissociation of Ado• and Ado⁻, and to instead favor β-H elimination. Similar cage effects have been observed in MetH and in the AdoCbl-dependent enzyme glutamate mutase, although in both cases these cage effects favor radical pair recombination and not a β-H elimination (104, 140, 141). Additionally, the observation that cob(II) is formed upon anaerobic photolysis of CarH seems at odds with heterolytic Co-C bond cleavage, although formation of cob(II) could be explained as a downstream consequence of cob(III) hydride decay (39). The contributions of specific amino acids are similarly unclear. The lack of conservation of His132, whose side chain occupies the B₁₂ upper face following photolysis, suggests that its role may be specific to formation of bis-His-ligated cob(III). Instead, the conserved Trp, Glu, and His of the W-9(x)-EH motif, which interact with the Ado group in the dark-state CarH tetramer and orient the Ado group differently than is usually observed in enzyme-bound or free AdoCbl (Figure 4d), could be involved in reprogramming AdoCbl photochemistry (37-40, 142), but again more work is needed. Detailed mechanistic studies using experimental as well as quantum mechanics/molecular mechanics approaches will help further elucidate the novel and promisingly rich photochemistry of CarH.

DISTRIBUTION AND EVOLUTIONARY ASPECTS OF B₁₂-DEPENDENT PHOTORECEPTORS

Genome searches yield hundreds of proteins with a MerR-type DBD fused to a B₁₂-binding domain in bacteria, only some of which are capable of de novo B₁₂ biosynthesis. We could identify ~600 proteins with the characteristic motifs found in experimentally studied CarH proteins, and hence likely to be AdoCbl- and light-dependent: RxWxxR in the DBD and W-(9)x-EH (or W-(10)x-EH in many actinobacterial homologs, notably of the genus Streptomyces) preceding the E/DxH motif in the B₁₂-binding domain (Figure 7). Various other surrounding residues in the DBD and the B₁₂-binding domain are also conserved (Figure 7b), hinting at their structural and functional importance. However, the His observed as the upper axial ligand in the bis-His cobalt of light-exposed T. thermophilus CarH (His132) is conserved only in some phyla, like Deinococcus-Thermus, suggesting a restricted functional role. One noteworthy feature is that the His of the W-(9)x-EH motif (His142 in T. thermophilus) and the Glu/Asp of the E/DxH of the B₁₂-binding motif (Glu175 in T. thermophilus) are typically spaced 32-39 residues apart (33-36 being most common). Another is the rather wide distribution of linker size (~6-120 residues) between the DBD and the B₁₂-binding domain, which possibly underlies, as noted before, a rather flexible DNA-binding mode. Altogether, the motifs shown in Figure 7b may serve as a signature for an AdoCbl-binding light-sensor module. The various putative CarH proteins are distributed across many bacterial phyla, being most abundant in Actinobacteria, Bacteroidetes, Chloroflexi, Deinococcus-Thermus, firmicutes, and β- and δ-Proteobacteria (Figure 1a). In most species, the corresponding genes are present as a single copy that, consistent with a light-response function, frequently occurs in the vicinity of genes for carotenogenesis or DNA photolyase. However, as observed in B. megaterium (129), carH can be involved in light-dependent regulation even when spatially decoupled in the genome from its target genes.

Figure 7.

Phylogenetic distribution of CarH and of its B₁₂-binding domain in other proteins. (a) Distribution of the indicated domain architectures across different bacterial phyla, with the number of genomes for each phylum indicated (from ~8400 bacterial genomes available at http://www.ncbi.nlm.nih.gov/genomes; for simplicity, only Candidatus phyla with relevant B₁₂-binding domains are shown). The number of proteins in each phylum (H, CarH; A, CarA) correspond to the color heat map on the bottom right. InterPro (protein sequence analysis & classification, https://www.ebi.ac.uk/interpro/) was used to search for proteins with a B₁₂-binding domain (IPR006158) alone, or in combination with a MerR-type DNA-binding (IPR000551) domain or other domains [IPR007024, globin-like; IPR005467, histidine kinase; IPR029016, GAF (cGMP-specific phosphodiesterases, adenylyl cyclase, FhlA)]. Protein sequences retrieved from UniProt (Universal Protein Resource, http://www.uniprot.org/) were aligned using MUSCLE in MEGA7 (Molecular Evolutionary Genetics Analysis, http://www.megasoftware.net/). Those with the E/DxH B₁₂-binding motif preceded by a W(9/10)xEH motif were selected by manual curation. The ones fused to a MerR-type DBD with the RxWxxR motif were classified as CarH or, if they lacked the Trp but not the EH of the W(9/10)xEH motif, as CarA. The 200-260 residue size range was used to select for standalone proteins. (b) Sequence logo created using WebLogo (http://weblogo.berkeley.edu/) for 498 CarH proteins with the W-(9)x-EH motif (those with a W-(10)x-EH motif, mostly in actinobacteria, were omitted). Shown are segments around signature residues (asterisks above) used for curation in the DBD (top left, cyan border), the four-helix bundle (top right, gold border) and the Rossmann fold (bottom, green border). In the sequences (x, any residue) below each logo, conserved residues (≥95%) are highlighted in bold and larger font.

Proteins with CarH domain architecture but which lack the Trp of the W-(9)x-EH motif, like CarA, and hence probably B₁₂-independent, are less abundant (~60) and more narrowly distributed (Actinobacteria, Firmicutes, and δ-Proteobacteria) (Figure 1a). The predominance of CarH proteins and their wider distribution suggests an ancestral protein that was more like CarH, a hypothesis in line with the distribution of CarH and CarA in the Myxococcales order of δ-Proteobacteria (25 and 10 proteins, respectively). Members of its Sorangiineae and Nannocystineae suborders have just one carH gene. M. xanthus and related species in the Cystobacterineae, however, have two tandem paralogous genes (carA and carH), as well as CarS and the full repertoire of other factors required in the B₁₂-independent CarA-CarS pathway, with Cystobacter fuscus standing out as the only one lacking CarA (M.C. Polanco, J. Fernández-Zapata, S. Padmanabhan, M. Elías-Arnanz, unpublished data). Taken together with the ancient origin proposed for B₁₂ (48), the most parsimonious hypothesis is that the gene for an ancestral AdoCbl-dependent CarH duplicated and functionally diverged in Cystobacterineae to evolve into carA. Presence of carH, alone or in tandem with carA, in the neighbourhood of genes for carotenogenesis in myxobacteria, suggests a common AdoCbl-dependent mode of regulation of this light-induced process, the alternative B₁₂-independent pathway co-existing in species with CarA. Having both pathways likely confers an evolutionary advantage, since the photoprotective carotenogenic response is maintained even in the absence of available B₁₂. It mirrors the coexistence of B₁₂-dependent and independent isozymes in many organisms that ensures crucial enzyme activities on limited B₁₂ availability (48).

Besides B₁₂-dependent enzymes and CarH/CarA, genome data reveal the B₁₂-binding domain in a large number of proteins, standalone or linked to other effector/output domains (37, 143). In particular, a CarH-type AdoCbl-binding module with the characteristic W-(9)x-EH motif (again, W-(10)x-EH in Streptomyces) can be found in ~170 widely distributed standalone proteins (200-260 residues) (Figure 7a). Notably, B₁₂-binding AerR/PpaA antirepressors, which are chromosomally coupled to CrtJ/PpsR and restricted to purple nonsulfur bacteria, appear to be distinct from CarH, lacking the characteristic motif (130, 131). In some cases, a CarH-type AdoCbl-binding module is associated with domains typically involved in signal transduction (Figure 7a), such as globin (heme-based oxygen sensor), histidine kinase, or GAF (cGMP-specific phosphodiesterases, adenylyl cyclase, FhlA) (144). Their functions and modes of action remain uncharted, but it is tempting to speculate that these exhibit AdoCbl- and light-dependent protein-protein interactions with themselves (like CarH) and/or with other effector proteins. The stage has thus been set for future studies of the function and evolution of these putative B₁₂-binding proteins.

CONCLUDING REMARKS AND OUTLOOK

The discovery that AdoCbl is the chromophore of a new family of photoreceptors changed the perception that a basic property of AdoCbl, its light sensitivity, has no direct biological function. Major landmarks have been reached in understanding the mechanism of action of the first known photoreceptor of this class, CarH, including characterization of its distinct photochemistry and high-resolution structural descriptions of its functionally relevant states. These studies have illustrated how nature has assembled a light- and B₁₂-dependent transcription factor by repurposing structural modules of enzymes and DNA-binding proteins and repurposing AdoCbl from an enzyme cofactor to a light sensor. Genome data reveal similar B₁₂-binding domains in a plethora of organisms, standalone or linked to other modules that could allow for integration of signals and regulation of diverse biological processes. Their study will surely be a rich source to identify new modes of B₁₂- and light-dependent regulation and possible cross-talk between distinct signaling pathways. Future surprises are certainly in store, given the precedent from the studies thus far. Undoubtedly, innovative experimental and theoretical approaches will address the various remaining questions on the molecular mechanisms of B₁₂-based photoreception. Whereas light induces reversible molecular changes in the chromophore of other photoreceptors (31, 32), it alters the chromophore irreversibly for the AdoCbl-dependent photoreceptor. The latter suggests that specialized cellular salvage and repair pathways may exist that remain to be discovered. Another exciting research avenue concerns possible applications of these photoreceptors. In initial studies, CarH has been exploited for light- and B₁₂-dependent conditional expression of essential bacterial genes (145). Whether it can be optimized for additional optogenetic applications remains to be seen. In any case, the study of these photoreceptors has already uncovered fascinating new biology and chemistry, with many more surprises likely ahead.

SUMMARY POINTS.

1. A new family of photoreceptors uses 5′-deoxyadenosylcobalamin (AdoCbl) or coenzyme B₁₂ as a chromophore to sense near-UV, blue, and green light.

2. The first characterized AdoCbl-dependent photoreceptor, the CarH transcription factor, is a tetramer that represses transcription of genes involved in carotenoid biosynthesis in the dark. Light causes the CarH-bound AdoCbl to photolyse, leading to a large protein conformational change that disrupts the tetramer, impairing DNA-binding and abolishing transcription repression.

3. The photochemistry of the AdoCbl-based photoreceptor appears to be distinct from that of free and enzyme-bound AdoCbl. The altered photochemistry provides a means to safeguard the use of AdoCbl as a light sensor. The underlying molecular basis for the difference in photochemistry remains an open question.

4. In another mode of light- and B₁₂-dependent gene regulation, light induces cobalamin-binding to an antirepressor, which then binds to its target repressor to activate transcription.

5. Genome data reveal B₁₂-binding modules similar to the one in CarH associated with various other domains or existing as standalone proteins, the functions of which are unknown.

FUTURE ISSUES.

1. The structural basis for the variations in the oligomerization behaviour of AdoCbl-dependent photoreceptors is not known. For example, it is not clear why some tetramers disassemble to dimers whereas others disassemble to monomers.

2. More evidence is needed to support the proposal that the variation in linker length between the CarH DNA-binding domains and light-sensing domains explains the apparently flexible modes of DNA-binding of CarH homologs.

3. The exact role(s) of the protein residues around the Ado group in CarH remain to be determined. It is not clear which residues, if any, are responsible for the distinct photochemistry of CarH.

4. Theoretical (quantum mechanics, molecular mechanics) approaches will be helpful in understanding the photochemistry of AdoCbl-bound CarH.

5. The intracellular fates of the photolysed photoreceptor and its associated chromophore remain to be established, as do the cellular machineries involved in determining these fates.

6. Whether other physiological processes are regulated by AdoCbl-based light-sensing is an open question.

7. It remains to be determined if B₁₂-based photoreceptors can be exploited as a useful optogenetic tool.

Photoreceptor: a protein that directly senses light via a chromophore component, which is an organic noncovalently or covalently bound molecule that undergoes photochemical transformation and triggers conformational changes in the protein for signal propagation. Various classes of photoreceptors have been established based on the chromophore and the protein component. A “sensor” or “input” domain in a photoreceptor senses light producing molecular changes that are transmitted to an “effector” or “output” domain that has a defined activity, which can be binding to DNA, RNA, or proteins, or be enzymatic.

Optogenetics: a technology that combines genetic and optical methods to achieve rapid and precise control of biological processes. Optogenetics uses natural or genetically-modified light-sensitive proteins or domains to spatiotemporally modulate protein-protein interactions, cell signaling, enzyme activities, or gene expression.

Adenosyltransferases (ATRs): enzymes that convert cob(II)alamin to AdoCbl. Three classes of sequence-unrelated ATRs are known in bacteria: CobA (BtuR or CobO), which acts in de novo AdoCbl biosynthesis, and EutT and PduO, which convert imported B₁₂ to AdoCbl. The most prevalent and well studied is PduO, whose human ortholog (MMAB) is associated with mitochondrial B₁₂ metabolism.

Riboswitches: RNA-based elements, found mostly in bacteria and archaea (a few in plants), which mediate a mode of gene regulation based on RNA conformational changes upon binding to ligands like vitamins, amino acids, nucleotides, aminosugars, or metals. Riboswitches can regulate transcription or translation, and generally act in cis, although some that act in trans are also known. The first discovered riboswitch and the second most-widespread is the AdoCbl riboswitch.

MerR proteins: Transcription factors that regulate responses to heavy metals, drugs, and other stresses in bacteria. They are dual-function regulators that can repress or activate a promoter by binding as dimers to a site with (pseudo)palindromic sequences, which is located within a spacer of suboptimal length between the −10 and −35 promoter elements.