Several early studies using NMR, fluorescence and absorption spectroscopy predicted that Trp91 (or Trp104 in AppA) was likely buried in the hydrophobic FAD binding pocket (6
). However recent acrylamide quenching studies indicated that a Trp91 homolog in the native AppA is not located near the flavin in the dark state and that light excitation only promoted minimal movement of this aromatic residue (18
). Similar acrylamide quenching analysis of tryptophan in this study with PixD also indicates that Trp91 is partially solvent exposed under dark condition with a slight increase in exposure under lit condition. Thus, the PixD dark state also likely contains the hydrogen bonded triad of Tyr8-Gln50-Met93 that is present in 9 of 10 subunits of the PixD crystal structure where Gln50 is also hydrogen bonded to N5 of the flavin ring (6
). The observed slight increase in quenching of Trp91 by acrylamide under lit conditions is consistent with the observed increase in flexibility of the β4 and β5 connecting loop that contains Trp91 upon light exposure. Comparison of Trp91 in the wild type and Y8F crystal structures () also shows that this residue seems to sample more conformations in Y8F in comparison with two main orientations in wild type PixD dark state. This could explain why the Y8F acrylamide quenching curve is between the dark and lit state quenching curves of wild type PixD. Finally, the downward curve of acrylamide quenching observed with wild type PixD in the dark state can be explained by the presence of two main conformations of Trp91 under this condition. Dynamic sampling of different conformations by Trp91 could present a strategy for a ready response to even small perturbations of nearby H-bonds (), either by mutation, or light-induced electron/proton transfer. Finally, mutating Trp91 itself does not affect the output signal as measured by PixD interaction with PixE but it does affect rate of the photocycle (6
). This result is consistent with the likely location of Trp91 away from the Tyr8-Gln50-Met93 hydrogen bond network with the flavin yet located at the start of the β5 strand that is flexible and variable in length. Thus a mutation in Trp91 likely influences the dynamics of this region and ultimately the return to the dark ground state.
Tyr8 is thought to initiate the BLUF photocycle through donation of an electron and proton to the light excited flavin (11
). This is thought to be followed by either rotation (9
) or tautomerization of Gln50 (16
), resulting in the formation of a new hydrogen bond to O4 of the flavin that leads to a 10 nm spectral shift (20
). Our observation that mutations of Tyr8 and Gln50 do not exhibit a photocycle, while an Ala substitution of Met93 does lead to a productive photocycle, suggests that Met93 is not a key residue for the primary photochemistry. This is not the case for the output signal as an Ala substitution of Met93 does lead to a loss of PixD interactions with PixE. Met93, which is conserved among BLUF proteins, must therefore be considered a key residue in propagating the output light signal. This role of Met93 is also reflected in a report that M93A substitution does not undergo similar light-induced conformation change of the peptide as does wild type PixD (26
). Met93 is strategically located at a junction of the β4 and β5 loop that increases flexibility upon light excitation (). Presumably, light excitation of the flavin disrupts the hydrogen bond that holds Gln50 to Met93 thereby allowing this loop to become “untethered” and subsequently move. This conclusion is also supported by our Trp quenching study that shows dark-adapted M93A mutant having an identical amount of quenching as lit wild type PixD (). The loop that is “tethered” by Met93 contains Trp91 so the Met93 to Ala mutant should exhibit increased flexibility of this loop even under dark conditions, which is indeed the case. Finally, analysis of the crystal structure of the pseudo lit state Y8F mutant shows that Met93 exhibits significant variability from one subunit to another in its distance from Gln50 (). This also suggests that Met93 is likely untethered to Gln50 in the Y8F mutant.
The asymmetry unit of Y8F crystal contains a hexamer with subunits arranged in a double half circle () instead of a decamer comprised of two complete stacked circles as observed in wild type PixD (6
). An interesting interacting region between sets of AB dimer pairs involves an interaction between α3 from one AB pair with the β4 and β5 flexible loop in a neighboring AB pair (). The β4 and β5 loop contains both Trp91 and Met93, and as discussed above, is known to undergo increased dynamic motion upon light excitation (). Note that all these interaction faces are also present in the wild type PixD dimer, which has a full ring comprised of 5 AB dimers instead of the open ring comprised of 3 AB dimers in Y8F. Even though one cannot rule out the possibility of crystallization condition affecting the packing form, considering the fact that wild type PixD adopts the same decamer conformation from two different crystallization conditions and cannot crystallize in the same condition of Y8F, the difference of Y8F oligomerization state seen in the structure is likely biologically relevant. One significant difference in the Y8F structure is that neighboring AB dimers are somewhat tilted away from the center of the half circle relative to neighboring pairs observed in the wild type structure (). Thus, if additional Y8F dimers were added to complete a circular structure then the two ends would not align. An intriguing possibility is that this difference indicates that light illumination of wild type PixD could disrupt the higher oligomer of PixD by inducing a twist to the full circle that causes disruption of the pentameric arrangement. RMSD analysis of the Y8F structure also reveals changes in the β3 strand that contains Gln50 from where the light signal is propagated. Major changes are also present around β5 strand and α3 helix with the loop between α3 and α4 increasing flexibility. Interestingly, the loop connecting β4 to β5 in one AB dimer interacts with the α3 helix of a neighboring dimer (), suggesting that these regions may push against each other upon light excitation thereby inducing a twist that disrupts the interaction between neighboring AB dimer pairs leading to disassembly of the PixD decamer.
Figure 8 Superposition of Y8F structure (green) on PixD WT structure (wheat). Chain C and D of Y8F were superposed on Ca atoms of WT structure. A tilt away from the circle center was observed on Y8F AB dimer or EF dimer (indicated by an arrow). The interface between (more ...)
Recently, light-excited structural changes were also reported for the BLUF domain of Klebsiella pneumoniae
BlrP1, a light-dependent c-di-GMP hydrolase. This BLUF domain also possesses C-terminal a3 and a4 output helices akin to those observed in PixD (27
). Illumination led to significant changes in the position and intensity of NMR signals in residues within the β4-β5 loop, β5 strand and in the loop between the a3-a4 helices (27
). When viewed in the context of the dark state crystal structure of full-length BlrP1 (46
), these light induced structural changes of the BLUF domain appear to be integral to regulating the catalytic EAL domain. We find it interesting that the changes observed in BlrP1 closely match the major changes observed in the crystal structure of the PixD Y8F mutant suggesting that an output signal involving alteration of the flexibility of the β4-β5 loop and its propagation through the β5 strand to the a3-a4 helices may be a common output signal among BLUF domains.