McLOV proteins are dimeric

All constructs readily purify from E. coli with a bound flavin cofactor and demonstrate spectra consistent with formation of a C4a adduct (Fig 2A–2C). Size exclusion chromatography (SEC) of LOV constructs reveals that McLOVn, McLOVr, and 337-TF are constitutively dimeric under dark state conditions (Fig 1C). Interestingly, in some McLOV purifications, a monomeric peak is present even at very high concentrations (Fig 1C). Such behavior has been observed in fungal LOV proteins that demonstrate sensitivity to oxidative stress [23]. We have been unable to determine the cause of a monomeric fraction in some preparations so far; however, conservation of key Cys residues that impart sensitivity to oxidative stress in other systems may be involved [23,25]. Thus, all Methylocystis LOV proteins are competent for dimerization and may partake in a model of competitive dimerization. Unfortunately, the photocycle lifetimes of all constructs were too fast to resolve any differences in oligomeric status following photo-excitation (see below). The limitations of the fast photocycles may be remedied through the design of rate altering variants, however that first requires a more thorough understanding of the LOV photocycle. Such knowledge may lead to the design of rate-altering variants for future structural studies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. McLOV photocycles and kinetics.

(A-C) McLOV proteins all demonstrate spectra consistent with C4a adduct formation. Dark-state spectra (black) are consistent with oxidized flavin. Illumination with blue light bleaches the 450 nm absorption bands (red). Subtle differences in degrees of photoactivation in McLOVn (A), 337-TF (B) and McLOVr (C) are indicated by residual presence of absorption bands centered around 450 with vibrational bands at 425 and 478 nm. (D-F) McLOVn (D), 337-TF (E) demonstrate first order kinetics as demonstrated by an absorbance trace at 450 nm and their respective residual plots. In contrast McLOVr (F) is best fit with a biexponential function.

https://doi.org/10.1371/journal.pone.0124874.g002

Photochemistry of LOV proteins from Methylocystis

All constructs purify with bound flavin and exhibit photobleaching in response to blue light (Fig 2). The dark-state spectra are consistent with an oxidized flavin with a broad absorption at 450 nm and shoulder peaks at 425 and 478 nm. The peak at 450 nm is the result of a S₀-S₁ electronic transition typical of the oxidized flavin. There is an additional absorption band in the UV-A region around 350 nm due to a S₀-S₂ electronic transition that facilitates adduct scission under illumination with broad spectrum light sources. Light excitation bleaches the 450 nm absorption band leading to a single peak centered around 390 nm, consistent with blue-light induced formation of a cysteinyl-flavin C4a adduct. In the Methylocystis LOV constructs, we are unable to completely form the light-state adduct; rather, illumination generates an equilibrium mixture between light- and dark-adapted states [7]. The different degrees of photoactivation likely result from different quantum yields of adduct formation as well as the rapid recovery observed in these proteins. To further clarify factors affecting LOV photocycles in Methylocystis species, we conducted a comprehensive analysis of the kinetics of adduct decay in all constructs.

Methylocystis LOV proteins are fast-cycling

Studies of LOV proteins have revealed that the rate of adduct decay under dark-state conditions can range from seconds to days [31]. In turn the variance in adduct lifetimes can affect sensitivity to light fluence that dictates the steady-state ratio between light and dark-adapted states [7]. Kinetic analysis of LOV proteins in Methylocystis reveals adduct decay proceeds with first order kinetics in all constructs tested (Fig 2). Rate constants and time constants for all constructs can be found in Table 1. The rate constants of these species are quite fast compared to fungal species, but consistent with those observed in fast cycling phototropins [2,31]. Specifically, the isolated sLOV protein McLOVn exhibited mono-exponential decay with a light-state lifetime of 38 seconds. McLOVr typically displays biexponential kinetics with a fast component (80 seconds) comprising 43% of the species and a slow component (180 seconds) 57% of the species. For Arrhenius and Eyring analysis we used the fast component (see below). The adduct lifetime of 337-TF was 130 seconds (Table 1), similar to the average time constant for McLOVr. Given the high homology to slow cycling fungal LOV proteins, we set out to examine what factors led to the fast photocycle. We specifically focus on two experiments. 1) Eyring and Arrhenius parameters for adduct decay. 2) The effect of small molecule bases on adduct recovery. Previous research in other LOV proteins indicates that solvent access to the active site can be probed using small molecule bases.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Kinetics of thermal reversion for LOV constructs and variants at 296 K.

https://doi.org/10.1371/journal.pone.0124874.t001

McLOV photocycle kinetics are entropically compensated

To better understand the photocycles of these sLOV proteins, we performed Arrhenius and Eyring analysis of dark state recovery. Typical energies of activation for fast cycling LOV domains (t1/2~30–1000 sec) are on the order of 70–100 kJ/mol [7]. In contrast, the temperature dependence of McLOVn kinetics reveals an uncharacteristically low activation energy of 52 kJ/mol (Fig 3). Eyring analysis indicates the low enthalpy of activation (51 kJ/mol) is compensated by a large unfavorable entropy of activation (-102 J/mol*K), allowing for a low energy of activation and weak temperature dependence of the rate of adduct scission (Table 1 and Fig 3). Such entropic compensation has been observed in circadian clock photoreceptors, where it may function to decrease the temperature dependence of photocycle kinetics [7]. Similarly, the McLOVn homolog McLOVr exhibits a low enthalpy of activation (53 kJ/mol) that is compensated by an unfavorable entropy of activation (-104 J/mol*K). Such similarities in the energetic parameters of adduct scission suggest evolutionary conservation of residues dictating the energy landscape during adduct scission. Given the high degree of similarity between McLOVn and the 337-TF (58% identical, 77% similar), we expected 337-TF to also exhibit a low enthalpy of activation that is entropically compensated; however, 337-TF demonstrates a markedly higher enthalpy of activation (71 kJ/mole) with a modest unfavorable entropy of activation (-44 J/mol*K). To examine whether these unique energetic parameters reflect additional alteration to the reaction landscape, we examined all constructs for the effect of small molecule bases on adduct decay.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Arrhenius and Eyring analysis of McLOV proteins.

Arrhenius (A,C,E) and Eyring (B,D,F) of McLOVn (A,B), 337-TF (C,D), and McLOVr (E,F) constructs. All measurements were carried out in triplicate, and error bars are shown as standard deviations relative to the mean. McLOVn and McLOVr both depict weak temperature dependence that indicates low entropies of activation with entropic compensation. In contrast 337-TF, has markedly increased enthalpies of activation (71 kJ/mole vs. ~50 KJ/mole) with a decrease in the entropic penalty (-44 J/mole*K vs. ~ -100 J/mole*K).

https://doi.org/10.1371/journal.pone.0124874.g003

External bases do not catalyze McLOVn and McLOVr

Scission of the Cysteinyl-C4a adduct has been shown to be catalyzed by small molecule bases such as imidazole [32]. These base catalysis studies in combination with solvent-isotope effects have led to a general model of adduct decay, where abstraction of the flavin N5 proton is rate limiting [31,32]. Currently, the intrinsic base is unknown. Notably, although LOV proteins are catalyzed by small molecule bases, the dependence of adduct decay on base concentration varies widely within the LOV family. Differences in sensitivity to base catalysis can often be explained via differences in solvent access to the active site, where an increase in solvent accessibility correlates to faster photocycles kinetics and sensitivity to small molecule bases [31,33]. We wanted to examine similar phenomena in McLOV proteins.

In contrast to other LOV domains, the dark state reversion process is not base catalyzed in McLOVn or McLOVr (Fig 4A and 4B). Addition of imidazole concentrations up to 150 mM has no effect on the rate of adduct scission at pH 8.0. Such behavior is surprising as most LOV proteins demonstrate remarkable base catalysis at 10–100 fold lower concentrations [7,34]. Even more surprising, despite high sequence conservation 337-TF demonstrates a large dependence of adduct decay on external bases (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. External bases do not catalyze some McLOV proteins.

(A,B) Both McLOVn (A) and McLOVr (B) show no dependence of adduct decay on imidazole concentration. In contrast, a Thr27Ile demonstrates spectra consistent with LOV chemistry (E) with a 100-fold decrease in the rate of adduct decay (F). The Thr27Ile variant causes an imidazole concentration dependent rate of adduct scission (C). Thus, Thr27 renders McLOV proteins insensitive to imidazole catalysis. (D) In contrast, 337-TF is imidazole sensitive despite containing a Thr at a position equivalent to 27. (G) A LOV structure (modeled from ENVOY PDB: 4WUJ) containing a Thr at a position equivalent to 27. Thr27 can coordinate a hypothetical water molecule between the N5 position on flavin, C61 and Q124 (McLOVr numbering). In all figures, error bars represent the standard deviation of the mean for experiments conducted in triplicate.

https://doi.org/10.1371/journal.pone.0124874.g004

Several factors could contribute to the divergent dependence of adduct decay on small molecule bases. These include: 1) No involvement of proton abstraction in the rate determining step. 2) A lack of solvent accessibility to the protein active site. We exclude this as a possibility, due to the fact that exclusion of solvent from the active site would lead to a slow rate of adduct decay, which is not observed. 3) The presence of an intrinsic base that renders external bases ineffective at accelerating adduct decay. To examine these factors further we conducted solvent isotope studies, as well as examined McLOV proteins for residues that could attenuate solvent accessibility or function as an intrinsic base.

Solvent isotope studies have been widely used in LOV proteins to examine the kinetics of adduct formation and adduct scission. SIEs have been reported of up to 5 for adduct formation and 2–4 for adduct decay [31,33,35]. Proton inventory studies confirm that a single proton transfer event, presumably involving N5 of the isoalloxazine ring, is rate limiting [31]. If such a proton transfer event is not rate limiting in McLOVn, McLOVr, and 337-TF, then we should observe little or no SIE. In contrast, McLOVn and McLOVr demonstrate SIEs of 3.4 and 2.5, respectively. Both values are consistent with a primary SIE, indicating that N5 deprotonation likely remains rate-limiting in these systems.

Since SIEs are consistent with N5 deprotonation being rate limiting, and the adduct decay rate is inconsistent with solvent exclusion from the active site, we presume that an intrinsic base may attenuate base catalysis in these systems. To probe such a possibility we examined McLOVn, McLOVr, and 337-TF sequences for residues that could either function as an intrinsic base or facilitate proton transfer from N5. A candidate residue was identified adjacent to the active site flavin and a conserved Gln residue required for LOV function. McLOVn, McLOVr, and 337-TF contain a Thr residue at position 27 (McLOVr numbering), where LOV proteins typically contain a hydrophobic residue (lle or Val) [31]. The site is well positioned adjacent to the N5 position, creating a small pocket lining N5 of the isoalloxazine ring, conserved Gln124 (McLOVr numbering), and Thr27 (Fig 4). Consistent with a role in affecting adduct decay, a Thr27Ile variant leads to a 100-fold decrease in the rate of adduct decay (Table 1 and Fig 4). In addition, a Thr27Ile variant modestly restores base catalysis in adduct decay (Fig 4C). Specifically, an increase of imidazole from 0 mM to 200 mM leads to a greater than 2-fold effect on the rate of adduct decay, consistent with a solvent-accessibility factor of 30. Such a value is consistent with weak base-catalysis as observed in other circadian clock LOV domains (ZTL, VVD) [7]. Notably, the high imidazole sensitivity of 337-TF indicates that a Thr at position 27 is not the sole factor influencing base catalysis, as 337-TF demonstrates a large solvent accessibility factor (a = 10000). We conclude that a unique residue substitution in the vicinity of the flavin N5 position greatly alters base catalysis, including rendering some sLOV proteins insensitive to base catalysis. Thus, examination of sLOV proteins in disparate systems indicates that mechanisms of LOV photocycles may not be universally conserved.

Implications for base catalysis

To the best of our knowledge, all LOV proteins that have been characterized demonstrate adduct decay pathways that are base catalyzed. In contrast, McLOVn and McLOVr are insensitive to external base concentrations. Examination of the LOV active site revealed an active site Thr that is well positioned to either activate a conserved glutamine as an intrinsic base, or to recruit and coordinate a base to the protein active site (Fig 4G). In McLOVn and McLOVr, the presence of Thr27 is essential to the abolishment of base catalysis and contributes dramatically to the fast photocycle present in these proteins. In contrast 337-TF exhibits strong base catalysis even in the presence of a Thr at the equivalent position. We propose that the Thr27 position directs the role of exogenous (imidazole) and endogenous bases in LOV catalysis. The divergent behavior in McLOV proteins in turn is likely coupled to nearby residues or alteration of N- C-terminal extensions to the LOV core. Such large effects on imidazole catalysis have been observed in other systems [34]. We further propose that the divergent dependence on imidazole allows McLOV proteins to function as model proteins for understanding the base-catalyzed mechanism present in all LOV proteins. We specifically address two proposed theories involving base catalysis in light of our data on McLOV proteins.

Small bases such as imidazole have been previously proposed to facilitate proton abstraction from the light state adduct [32]. These studies identified two plausible mechanisms for proton abstraction. The first suggests that an ordered solvent channel conserved in all LOV structures functions as a proton shuttle [41]. In such a model, imidazole acts as an external base that activates peripheral ordered water to facilitate proton transfer from N5. The second suggests that imidazole can act directly from the active site to abstract the N5 proton [31]. The second mechanism is supported by proton inventory measurements that implicate a single proton transfer event in the rate determining step in adduct decay. The mechanism is further supported by residue substitutions in the fungal LOV protein VVD that removes steric hindrance present in the LOV active site [31]. These substitutions enhance base catalysis in VVD and lead to a concentration dependence that fits a single site saturation model. Importantly, other LOV proteins demonstrate unsaturateable base catalysis. Even VVD variants that exhibit saturation must be modeled with a second unsaturateable component. In these regards, a direct site of action may not be a complete model.

Alternatively, the first mechanism proposes that ordered water directs base catalysis. Although known LOV structures conserve an ordered solvent channel leading to the flavin active site, no ordered water has been observed within H-bonding distances to N5 [41]. Further, such a mechanism contradicts proton inventory studies in several LOV systems. Thus, such a mechanism may not be accurate.

Based on our studies here, we propose an alternative hybrid mechanism. We propose that ordered water may function as an intrinsic base in LOV proteins. However, the ordered water near N5 is transient in nature and cannot be observed in existing LOV structures. The transient ordered water could then be activated by imidazole that acts in a single site of low affinity near the periphery of the conserved solvent channel. Such a mechanism is consistent with existing mutational data in VVD that requires mutation of an Ile residue near the periphery of the solvent channel to impose a saturateable model [31]. Moreover, introduction of an active site Thr in McLOV proteins may increase the occupancy of ordered water near N5 rendering base catalysis pathways comparatively inefficient. In such a system, water ordered by Thr27, Gln124, and N5 are directly activated by the surrounding protein, thereby out competing a very low affinity site for base catalysis. Importantly, such a function would also be dependent on the orientation of Gln124 as well as interactions to the N/Ccaps. Thus, Thr27 acts in a key position to have the capacity to alter imidazole catalysis in divergent manners. Importantly, a lack of crystal structures of McLOV proteins renders examination of ordered water in LOV proteins complicated. Future structural studies may shed light on the role of ordered water in base catalysis of LOV proteins. Combined, our studies highlight the importance of studying LOV proteins from novel environmental niches as they can facilitate our understanding of the global LOV photocycle mechanism as well as allosteric regulation of LOV function. Such studies can shed light on new mechanisms for optogenetic tool development and improvement of existing optogenetic devices.

Cloning

LOV constructs were either cloned from genomic DNA (Methylocystis Rockwall, kindly provided by Lisa Stein, University of Alberta) or synthetic DNA codon optimized for E. coli expression (Methylocystis rosea and Methylocystis Rockwall). Complete DNA sequences of McLOVn, McLOV-TF and McLOVr are provided in supporting information (S1 Sequence). Constructs were cloned by PCR amplification. Following amplification, PCR product was run on a 1% agarose gel to verify PCR product. PCR product was subjected to gel purification prior to digestion with Xho1 and Nco1 restriction enzymes. Digested DNA was ligated into a 6-His Parallel vector with ampicillin resistance. The resulting ligation reaction was transformed into DH5α E. coli cells and grown on LB-agar plates containing 200 μg/ml ampicillin. DNA was then isolated via miniprep (Promega). Isolated DNA was confirmed to contain insert by a restriction digest test prior to DNA sequencing (genewiz).

Expression

Proteins were expressed in E. coli BL21(DE3) cells grown at 37°C. After the cells reached a density (OD₆₀₀) greater than 0.5, the temperature was reduced to 18°C, and protein expression was induced with 200 μM IPTG. Cell pellets were harvested by centrifugation and stored in buffer (50mM Hepes, 100mM NaCl 10% Glycerol, pH 8) at -80 C.

Purification

Cell pellets were thawed at room temperature prior to sonication. Cell lysate was clarified via centrifugation at 16,000 rpm to remove cellular debris. Lysate was then applied to a 5 ml Ni-NTA affinity column to isolate the His₆ tagged protein; the protein was washed and eluted with 20 mM and 300 mM imidazole, respectively. The eluted protein was then incubated with TEV (Tobacco Etch Virus Protease) overnight to remove the His₆ affinity tag. Cleaved protein was buffer exchanged to eliminate residual imidazole, then subjected to another round of affinity purification to eliminate His₆-TEV. Finally, the desired protein was isolated with size exclusion chromatography (Superdex 75 GE-Healthcare) prior to concentration to 5–10 mg/ml characterization.

Photochemical characterization

All experiments were conducted in a quartz cuvette with a pathlength of 1.0 cm. The absorbance was measured at 450nm and 478nm to follow recovery from the light state back to the dark state after photobleaching. Prior to kinetic characterization dark-state samples were illuminated with a 5 mW 473 nm blue laser until a steady-state light/dark ratio was achieved. Recovery was measured at 0.5 s intervals, using the absorption values recorded at 450 and 478 nm. All kinetic traces were corrected for baseline drift by subtraction of the absorbance at 600 nm.

Eyring and Arrhenius analyses

All protein constructs were characterized with temperature dependent kinetics ranging from 15–36°C followed by Arrhenius and Eyring data analysis to extrapolate rate constants, activation energy, and entropy of activation. All kinetic studies followed procedures outlined above.

Base catalysis

Base catalysis and determination of solvent accessibility factors were conducted as outlined in Pudasaini et al. [7]. We briefly summarize the methodology here. Constructs were studied for their response to base catalysis by varying the concentration of imidazole within the range of 0–150 mM. We note that higher concentrations were tested, but in many cases led to protein instability. Kinetic parameters were extracted by measuring the absorbance at 450 and 478 nm as outlined above. A solvent accessibility factor (a) was calculated using methods published previously[34]. Briefly

(1)

The slope of a plot of k vs. base concentration is equivalent to:

(2)

The slope was calculated using the concentrations of both protein and imidazole in units of Molarity. All calculations were made with a protein concentration of 10 μM.

Solvent isotope effect measurements

All SIE experiments were conducted by first exchanging H₂O buffer (50mM Hepes, 100mM NaCl 10% Glycerol, pH 8) for an equivalent buffer prepared with D₂O. Buffer exchange was completed via 3 successive 10-fold dilutions into D₂O buffer followed by concentration to 5–10 mg/ml. The samples were than analyzed for their rate constants via the methodology defined above. SIE are reported as k_H/k_D.