Membrane-inserted complexes consisting of two photochemically reactive sensory rhodopsin (SR) subunits flanking a homodimer of a transducing protein subunit (Htr) are used by halophilic archaea for sensing light gradients to modulate their swimming behavior (phototaxis). The SR–Htr complexes extend into the cytoplasm where the Htr subunits bind a his-kinase that controls a phosphorylation system that regulates the flagellar motors. This review focuses on current progress primarily on the mechanism of signal relay within the SRII–HtrII complexes from Natronomonas pharaonis and Halobacterium salinarum. The recent elucidation of a photoactive site steric trigger crucial for signal relay, advances in understanding the role of proton transfer from the chromophore to the protein in SRII activation, and the localization of signal relay to the membrane-embedded portion of the SRII–HtrII interface, are beginning to produce a clear picture of the signal transfer process. The SR–Htr complexes offer unprecedented opportunities to resolve first examples of the chemistry of signal relay between membrane proteins at the atomic level, which would provide a major contribution to the general understanding of dynamic interactions between integral membrane proteins.
Phototaxis behavior by the halophilic archaeon Halobacterium salinarum is mediated by two membrane-embedded light-signaling complexes—SRI–HtrI and SRII–HtrII. Each consists of a homodimer of a transducer protein, the “haloarchaeal transducer” (Htr) subunits, flanked by two light-activated retinylidene receptors called “sensory rhodopsin” (SR) subunits (1–4). The SRI–HtrI complex mediates attraction towards orange light and, by a two-photon process, avoidance of near-UV light. The SRII–HtrII complex mediates avoidance responses to blue-green light.
Like the homologous chemotaxis receptors in numerous prokaryotes (5,6), the Htr subunits contain two transmembrane helices TM1 and TM2, with TM2 connected to a cytoplasmic (CP) helix-turn-helix motif (HAMP domain), followed by a CP domain which forms a four-helix coiled-coil bundle in the transducer dimer (Fig. 1a). As in chemotaxis receptors, adaptational methylation sites occur in the coiled-coil region and at the distal end the Htr proteins bind a CheW/A his-kinase which phosphorylates a regulator protein CheY that in turn modulates the flagellar motor. HtrII is a chemotaxis receptor as well as a phototaxis transducer; in HtrII from H. salinarum (HsHtrII), a binding domain for the chemoattractant serine is formed by the loop between the two TMs in the periplasmic domain (7).
A crystal structure of a complex of SRII (also called phoborhodopsin) and an N-terminal fragment of HtrII from Natronomonas pharaonis (NpSRII–NpHtrII) with truncation of NpHtrII at position 114 (22nd residue after the CP end of TM2) revealed the structure of the TM domain of NpHtrII, which associates with NpSRII helices F and G (Fig. 1b) (8). The structure in the CP part of NpHtrII was unresolved, but studies with fluorescent and spin-labeling probes (9,10) and peptide binding measurements (11,12) demonstrate that the receptor-transducer interface extends into this membrane-proximal region of the complex. From these observations two possible routes for signal transmission from the photoreceptors to the transducers could be envisioned—CP signal relay, in which conformational changes in the CP surface of the receptor are relayed to the transducer HAMP or CP domain of the transducer by direct interactions, versus membrane-embedded signal relay, in which the conformational changes in the helices of the receptor propagate to the transducer through the membrane-embedded interface. As discussed below, recent results provide compelling evidence for the latter, i.e. membrane domain signal relay, as the main mechanism.
Mechanism of activation of the photoreceptor subunits
Conformational changes in the receptor might be deduced from those of the light-driven proton pump, bacteriorhodopsin (BR), closely related to sensory rhodopsins and exhibiting 20–30% identity in their residue sequences (13–15) and sharing near-identical architecture as revealed by the crystal structures of BR (16) and NpSRII (17,18). Conservation is highest in the residues constituting the chromophore binding pocket, including Lys205 in SRII (Lys216 in BR) on helix G, which covalently attaches the chromophore through a protonated Schiff base (PSB), and the complex counterion of the PSB composed of Arg72, Asp75 on helix C and Asp201 on helix G (Arg82, Asp85 and Asp212 in BR). The similarity in the local environment of the chromophore results in very similar photochemical reactions of BR and SRII, in which photoisomerization of the chromophore from all-trans to 13-cis triggers a series of intramolecular chemical reactions recognized as a cyclic series (“photocycle”) of spectrally distinct intermediate states named K, L, M, N and O. While the shifts of the absorption maxima (λmax) to the red and slightly to the blue in the K and L states, respectively, reflect changes in the strength of the salt bridge between the PSB and the counterion complex (19,20), the much larger blueshift of the λmax upon M formation reflects deprotonation of the Schiff base by the Asp75/85 anion, thereby disrupting the salt bridge. In the proton pump BR, Asp85 protonation perturbs the hydrogen-bonding network in an extracellular (EC) channel causing proton ejection. The subsequent reprotonation of the Schiff base in the M-to-N transition, accomplished by proton transfer from Asp96 in the CP channel, results in vectorial translocation of one proton per BR photocycle. The following thermal isomerization of the chromophore from 13-cis to all-trans in the N-to-O transition reorients the PSB toward the protonated Asp85 partially restoring the interactions between them, as reflected by the redshifted λmax, until finally reestablishing the salt bridge by transferring the proton on Asp85 to the proton release group as O returns to the initial state (19,20). Thus, a net proton transfer from the CP side to the EC side of the molecule is accomplished during the photocycle.
In SRII, the photochemical events during the photocycle are essentially similar to those in the BR photocycle (21,22), although low proton conductivity in the CP channel makes SRII essentially nonfunctional as a proton pump (23). Some proton pumping activity is conferred on SRII by introduction of an Asp at position 87 corresponding to Asp96 of BR (24–26). In wildtype SRII, the Schiff base reprotonation in the M-to-O (N) transition appears to occur from elsewhere on the EC side during the photocycle resulting in an intramolecular proton circulation without significant net vectorial translocation (23,27).
The similarity in photochemistry between the two molecules suggests that SRII undergoes a conformational change similar to that of BR to switch the Schiff base accessibility from the EC to the CP side during the M state. As vectorial transport requires this switch, this suggestion is strongly supported by the proton pumping activity of the HtrII-free SRII_F86D/E mutant (24,26) as well as by the electrogenic proton transport by wildtype transducer-free SRI (28,29). In BR the accessibility switch involves the opening of the CP channel by an outward tilt of helix F and inward tilt of helix G detected by cryo-EM and electron paramagnetic resonance (EPR) spectroscopy (30–33), and a light-induced outward movement of helix F in transducer-free SRII has been shown by EPR analysis of spin labels (34). One of the motive forces inducing this conformational change in BR is indicated to be the salt-bridge disruption during the formation of the M state as a similar conformational change results from disrupting the salt bridge by mutating Asp85 to Asn in the background of Asp96Asn (35). Arguing for a critical role of salt-bridge disruption also in SRII, mutation of HsSRII Asp73 (Asp75 in NpSRII) to uncharged Gln or Asn was shown to activate the receptor to various extents as revealed by a higher frequency of spontaneous swimming reversals of the cells in the dark, diminished responses of the cells to the light stimuli and adaptive demethylation of the HsHtrII subunit in the dark mimicking the effect of photostimulation (36,37).
Constitutive activation studies also found that SRII mutants which were only partially activated (D73A, D73N, D73T, D73S or D73L) maintain light-elicited signaling function even though the photocycle does not involve deprotonation of the Schiff base as demonstrated by the absence of M formation. Therefore, photoisomerization of the chromophore generates further conformational change in the receptor that is not induced by the retinylidene proton transfer alone (36,37). Note that a disrupted salt bridge may be necessary for signaling, but in the partially activated Asp73 mutants it is not sufficient to fully induce the signaling conformation. Therefore, photoisomerization of retinal in addition to inducing the pKa changes in the Schiff base and its Asp counterion that lead to proton transfer has other effects on the protein contributing to signaling. In this regard, recently steric interaction of the isomerizing retinal with a Thr-Tyr hydrogen-bonded pair in the retinal pocket of wildtype NpSRII has been demonstrated to be essential for photoactivation of the NpSRII receptor. Elimination of this hydrogen bond by Ala substitution for Thr204 on helix G or Phe substitution for Tyr174 on helix F (Fig. 2) eliminates phototaxis but does not eliminate Schiff base deprotonation (38). The Thr-Tyr pair undergoes a large perturbation upon K formation (39) attributed to steric contact with the isomerizing retinylidene C13=C14 double bond of the chromophore (40).
The critical Thr-Tyr hydrogen bond is unique to SRII and is not present in BR (nor SRI). The steric interaction appears to be responsible for a greater distortion of the chromophore binding pocket in the K state in NpSRII compared to that in BR. Light-dark difference FTIR spectra show hydrogen-out-of-plane bending vibrational modes characteristic of twisted chromophores extending along the retinal polyene chain in NpSRII, whereas in BR the twist is localized to the Schiff base and C15 of the retinal (41,42). The difference in binding pocket perturbation between the two proteins is likely attributable to steric interaction of the isomerized retinal with the Thr-Tyr hydrogen bond because a similar distortion of the chromophore in BR was indeed obtained when Ala215 of BR was mutated to Thr, which forms a hydrogen bond to Tyr185 in a corresponding position as the Thr-Tyr bond in NpSRII (43).
SRI, like BR, contains an Ala at the corresponding position as Thr204 in SRII and therefore the Thr-Tyr hydrogen bond crucial for SRII activation is absent. Nevertheless a steric trigger during K formation has been demonstrated in SRI by retinal analog measurements showing that K does not accumulate and the photocycle and function are blocked unless the retinal 13-methyl group of the retinal interacts with the protein during primary photochemical events (44). Results from low temperature flash photolysis suggest a model in which the 13-methyl group steric contact with the protein functions as a fulcrum to permit movement of one or both ends of retinal to overcome an energy barrier against isomerization (45). As in wildtype SRII, the deprotonated Schiff base intermediate (S373) of SRI is a functional signaling state relaying a signal to its cognate transducer (46). Therefore, in the SRI–HtrI complex as in SRII–HtrII, both a steric trigger from retinal–protein interaction during photoisomerization and the later retinylidene proton transfer reaction are key chemical events in receptor activation.
Conversion of BR into a signaling photoreceptor
If SRII uses a modified form of the BR transport machinery with the additional feature of the retinal/Thr204-Tyr174 steric interaction for triggering conformational changes that activate HtrII, BR might be converted to a signaling molecule by introducing the Ala215Thr mutation combined with engineering BR to bind HtrII. Binding to NpHtrII with a Kd of ∼60 μm in detergent micelles was observed with the BR double mutant Pro200Thr/Val210Tyr (47), mimicking the residues Thr189 and Tyr199 of NpSRII, which hydrogen bond to Asn74 and Ser62/Glu43 of NpHtrII, respectively (8) (Fig. 1b). The BR triple mutant Pro200Thr/Val210Tyr/Ala215Thr (“BR-T”) was indeed found to mediate phototaxis in H. salinarum though NpHtrII (48). The efficiency of signaling was increased to levels approaching that of wildtype SRII–HtrII (36% of native HsSRII–HsHtrII) when BR-T was complexed with the native HsHtrII.
The BR-T results proved that BR and SRII share fundamental reactions in their mechanisms, and that the same machinery is used for transport and signaling with the Thr-Tyr hydrogen bond and HtrII binding converting the light-driven pumping process into photosignaling. The Thr-Tyr pair is adjacent to the mid-membrane SRII–HtrII interface and connected to SRII residues of this portion of the interface by van der Waals and hydrogen bonding. The most direct signaling mechanism would entail that the signal transfer to HtrII results from perturbation of this nearby membrane-embedded region of the interface by the Thr-Tyr steric interaction with the isomerizing retinal (the “membrane-embedded steric trigger” model, Fig. 3). However, the Thr-Tyr hydrogen bond connects helices F and G in the unphotolyzed state and perturbation of the bond upon K formation might result in structural changes in the Thr-Tyr pair that alter the later movements of helices F and G in BR-T compared to those of wildtype BR. Therefore, the BR-T results alone do not exclude that conformational changes distant from the Thr-Tyr site relay the signal, in particular changes in the extension of helix F on the CP surface of the photoreceptor which is known to interact with the transducer’s functionally important HAMP domain (see next section).
Association of the HAMP domain of the transducer with the photoreceptor
As discussed above, the SRII–HtrII interface extends beyond the TM region resolved in the crystal structure into the CP membrane proximal domain. Isothermal titration calorimetry (ITC) measurements identified the N-terminal 114-residue fragment of NpHtrII, which contains the TM domain and 22 residues of the CP extension, as sufficient for high-affinity binding to NpSRII in detergent micelles, whereas the N-terminal 100-residue fragment did not bind (11). By comparison of the sequence of NpHtrII in the membrane proximal region with that of the HAMP domain of Af1503 in Archaeoglobus fulgidus, for which an NMR structure shows two amphipathic α-helices AS-1 and AS-2 connected through a flexible linker peptide (helix-turn-helix motif) (49), residues 83-114 of NpHtrII contain AS-1 and the linker, while residues 83–100 contain only a part of AS-1. Thus high-affinity binding of NpHtrII to NpSRII requires a complete AS-1 and the linker portion of the HAMP. However, the affinity of the CP HAMP domain alone was shown to be weak (12) and mutation of Tyr199 of NpSRII to Phe/Val, which eliminates an intramembrane hydrogen bond to Asn74 of NpHtrII, reduces the affinity of NpSRII and NpHtrII(1-147) in detergent micelles (47,50). Therefore it appears that the TM and HAMP domains both contribute to receptor binding.
Signal relay through the TM domain of the transducers
Judging from the orientation of NpSRII with helices F and G facing NpHtrII, a good candidate for the CP binding region of SRII to the transducer HAMP domain is the portion of helices E and F (the EF loop) protruding into the cytoplasm. Strong support for EF loop binding to HtrII has been provided by fluorescent probe labeling and fluorescence resonance energy transfer (FRET) (9) and independently by EPR of spin-labels (34). Fluorescent probe labeling of SRII residue 154, which is at the tip of the EF loop, was inhibited by NpHtrII in the dark and even further inhibited in the photoactivated NpSRII–NpHtrII complex. Moreover, FRET from a probe at position 154 of NpSRII to Trp residues introduced in NpHtrII was maximal at position 93, which is in the AS-1 helix of the HAMP domain. These results indicate close proximity of residue 154 of the EF loop of the photoreceptor and residue 93 in the HAMP domain of the transducer, which become even closer when the photoreceptor is activated with light, presumably as a consequence of either the outward tilt of helix F of NpSRII or conformational changes in the HAMP domain of NpHtrII. A similar observation was made also at NpSRII position 158 (one helix turn toward the membrane from 154) labeled with an EPR probe, which became less accessible to CrOx when NpHtrII was bound and became immobilized upon illumination, indicating interaction with the transducer strengthened at this position by light activation (34).
The change in the interaction between the EF loop and HAMP domain upon photoactivation of the receptor raised the possibility of signal relay from NpSRII to NpHtrII taking place through these domains. However, an obligatory role of this interaction is ruled out because deletion of the EF loop in NpSRII does not eliminate robust signaling (37). Direct evidence for the TM domain of HtrII as the site of signal relay was obtained by localizing constitutive signaling by mutationally activated receptors in HtrII chimeras. Constitutive activity of NpSRII elicited by the Asp75Gln mutation is perceived by HsHtrII or a chimera of HtrII with the N-terminal membrane-inserted domain of HsHtrII joined to the CP portion of NpHtrII, but not by NpHtrII or a chimera of HtrII with the TM domain of NpHtrII joined to the CP portion of HsHtrII (37).
Recognition specificity and signal relay in the SRI–HtrI complex also reside largely or entirely in the membrane-embedded domain as swapping of the CP portion of the transducer below HtrI residue 60 with the corresponding portion of HsHtrII does not impair attractant and repellent phototaxis signaling by SRI. Therefore, the residues in the HtrI TM domain (residues 1–53) and the beginning part of HAMP (residues 54–60) appear to be sufficient for SRI–HtrI signal transfer (51).
These lines of evidence strongly argue that signal relay between the photoreceptor and the transducer subunits takes place through their helix–helix interactions in the membrane, i.e. membrane-embedded signal relay (Fig. 3). Supporting this mechanism, structural changes in the membrane-embedded SRII–HtrII interface upon photoactivation were shown by FTIR spectroscopy by the large light-induced perturbation of the hydrogen bonding between Tyr199 on helix F of NpSRII and Asn74 on TM2 of NpHtrII upon formation of the M intermediate (52).
If this perturbation destabilizes the complex, it would be in line with the suggestion from Sudo et al. (50) that a looser binding to SRII activates the transducer, based on the significantly higher dissociation constant they obtained from the kinetics of the M decay rate of NpSRII in the presence of NpHtrII compared to that obtained from ITC measurement of the dark state complex. If disruption (or partial disruption) of contacts in the interface occurs, it may cause the clockwise rotation of TM2 of HtrII (viewed from the CP side) deduced from light-induced changes in the distances between spin probes labeled on residues 78 and 78′ or 82 and 82′ in the TM2 helix of NpHtrII monomers (53). Such a rotation is supported by light-induced changes in TM2-TM2′ disulfide formation rates of cysteine residues engineered in the NpHtrII dimer (54) and by X-ray crystallography (55), and would input into the HAMP domain switch formed by the extension of TM2 in the cytoplasm.
Alternatively, the TM2 rotational motion may result from coupled movements of helices F and G of SRII without separation of SRII and HtrII. A mobility increase in a spin label attached to residue 159 on helix F on the CP surface facing helix G was attributed to rotation of helix F or the outward tilt of helix F (34). Coupled motions of the TM helices of SRII and HtrII would offer a mechanism of conformational transmission through steric contacts of these helices adjacent to the critical Thr-Tyr hydrogen bond.
The recent rapid progress on the SRII–HtrII complex offers an opportunity to resolve a first example of the chemistry of signal relay between membrane proteins at the atomic level, which would provide a major contribution to the general understanding of dynamic interactions between integral membrane proteins. The current results focus immediate attention on elucidation in the SRII–HtrII complex of the structural changes at the Thr-Tyr site caused by the steric conflict with 13-cis retinal, the chemical reactions propagating this change to the F-G/TM2 interface and the resulting structural change in the transducer TM2.
Acknowledgements— Work by the authors referred to in this review was supported by National Institutes of Health grant R37GM27750 and the Robert A. Welch Foundation.