The Cytoplasmic Membrane-Proximal Domain of the HtrII Transducer Interacts with the E-F Loop of Photoactivated Natronomonas pharaonis Sensory Rhodopsin II

The structures of the cytoplasmic loops of the phototaxis receptor sensory rhodopsin II (SRII) and the membrane-proximal cytoplasmic domain of its bound transducer HtrII were examined in the dark and in the light-activated state by fluorescent probes and cysteine cross-linking. Light decreased the accessibility of E-F loop position 154 in the SRII-HtrII complex, but not in free SRII, consistent with HtrII proximity, which was confirmed by tryptophans placed within a 5-residue region identified in the HtrII membrane-proximal domain that exhibited Förster resonance energy transfer (FRET) to a fluorescent probe at position 154 in SRII. The FRET was eliminated in the signaling-deficient HtrII mutant G83F without loss of affinity for SRII. Finally, the presence of SRII and HtrII reciprocally inhibit homodimer disulfide cross-linking reactions in their membrane-proximal domains, showing that each interferes with the others self-interaction in this region. The results demonstrate close proximity between SRII-HtrII in the membrane-proximal domain, and in addition, light-stimulation of the SRII-inhibition of HtrII cross-linking was observed, indicating that the contact is enhanced in the photoactivated complex. A mechanism is proposed in which photoactivation alters the SRII-HtrII interaction in the membrane-proximal region during the signal relay process.


INTRODUCTION
Archaeal sensory rhodopsins I and II (SRI and SRII) are membrane-embedded phototaxis receptors that modulate the motility apparatus of Halobacterium salinarum and related haloarchaeal species (1)(2)(3). The SRI and SRII proteins transmit signals to their cognate transducer proteins, HtrI and HtrII, respectively, and like their eubacterial counterparts in chemotaxis, the Htr transducers control a histidine-kinase and phosphoregulator protein that modulates motor function.
Biochemical and spectroscopic studies have shown that there is close interaction between the SR and Htr components of the signaling complex both in the light and in the dark (1); i.e. they are subunits of a molecular complex. Furthermore, chimera experiments demonstrated that the interaction specificities of SRI with HtrI and SRII with HtrII are determined by the transmembrane helices of the Htr subunits (4). Cubic lipid phase crystal x-ray structures of SRII have defined the membrane-embedded and membrane-external portions of the protein's transmembrane helices and periplasmic and cytoplasmic loops (5,6). Moreover, co-crystallization of SRII with an N-terminal 5 contacts between HtrII TM2 (and to a lesser extent TM1) and SRII helices F and G within the membrane domain. The presence of the cytoplasmic membrane-proximal domain, comprised of 32 residues beyond Leu82 at the membrane-cytoplasm interface, was necessary for high affinity binding of the HtrII fragment to SRII in detergent micelles used in the crystallization (7,8). The structure of the cytoplasmic extension was not resolved by the crystallographic refinement.
Site-specific mutagenesis suggests that the HtrII membrane-proximal domain is critical for signaling. Cysteine-scanning through the length of TM2 from Ala60 to Ala88 showed that no single residue in the membrane-embedded portion is critical for phototaxis function (9). Also, as reported here, mutation of SRII Tyr199 to Phe or to Ala, which eliminates a known hydrogen bond between the Tyr residue and an Asn residue on TM2 in the membrane domain, does not significantly impair signaling. In contrast, cysteine-substitution of the first cytoplasmic residue (mutants HtrII_G83C and 6 cluster of functionally important residues ranging from positions 53-96 in HtrI (Asn53 corresponds in HtrII to Leu82 at the membrane/cytoplasm interface), and also 3 residues in the cytoplasmic regions of helices F and G of SRI (10). Also a decreased phototaxis response occurs in HtrI I64C (11). Moreover, cytoplasmic extension of HtrI TM2 by 13 residues is necessary and sufficient for HtrI-dependent properties of SRI in the SRI-HtrI molecular complex, including the HtrI-inhibition of opening of its proton-conducting cytoplasmic channel during the photocycle ((12), Chen and Spudich, submitted).
Light-induced tilting of helix F (and to a lesser extent helix G), opens a cytoplasmic channel in BR (13) and this tilting has been shown to occur in SRII both when free and bound to HtrII (14,15), with HtrII inhibiting the proton uptake through the channel as in the SRI-HtrI complex.
This conformational change, which involves structural alterations mainly in the cytoplasmic end of helix F and in the E-F loop (13), has been proposed based on several lines of evidence to be responsible for activating the Htr transducers (16).
Our goal in this work was to use fluorescent probe accessibility, FRET, and cysteine Plasmid construction and site-directed mutagenesis: HtrII was expressed using the pCY8 plasmid while SRII was expressed using the pCY9 plasmid, constructed by PCR (17). pCY8 was constructed by using CY4 (9) as template DNA and two primers: a forward primer Nco-004-F, which encodes a NcoI site in the beginning of the HrtII gene and a reverse primer CY8-his6-R contained the first 159 residues of HtrII and added six histidines in the C-terminus followed by a HindIII site. The resulting NcoI-HindIII fragment was ligated with the large fragment of NcoI-HindIII treated pET21d vector (Novagen) to produce pCY8. The pCY9 was constructed by PCR using a pJS005 (18) as template DNA and a forward primer encoded the beginning of the SRII gene and a reverse primer encoded the last five SRII residues followed by six histidines and a HindIII site. The resulting NcoI-Hind III fragment was ligated with the large fragment of NcoI-HindIII treated pET21d to produce pCY9.
Single-cysteine substitutions or single-tryptophan substitutions in SRII and HtrII were constructed from the pCY8 or pCY9 plasmids by site-directed mutagenesis and Photocycle measurements. Nd-YAG laser-flash absorbance changes were acquired on a laboratory-constructed flash spectrometer as described in the companion article (Chen & Spudich, submitted).

Fluorescent probe labeling at S154-SRII for SRII-HtrII interaction measurements.
Labeling of SRII_S154C with (vinyl sulfone) was carried out by incubating SRII_S154C protein (in buffer M) at room temperature overnight at a molar ratio of protein:LY = 1:50. Disulfide bond formation assay. Cross-linking reactions were carried out as previously described (9). All SRII and HtrII samples were adjusted to 10µM with buffer M containing 0.1% DDM and cross-linking reactions were initialized with 3mM 1,10-phenanthroline (in ethanol) and 1.5mM CuSO 4 for 3 min before terminated by stop solution (2% SDS, 50mM NEM and 5mM EDTA). Photoactivation group were subject to continuously illumination with 500nm light for 3 min before and during the reaction.
All samples were then analyzed with 15% SDS-PAGE.
Protein sequence analysis used DNA Star V5.05 (Lasergen). 158, respectively, and 155Å 2 , 92Å 2 , and 49Å 2 for the same positions in HtrII-complexed SRII). Therefore, the reduced accessibilities suggest an effect of the unseen presence of HtrII residues in this region (i.e. residues present but unresolved in the crystal structure of the complex), which inhibit access to the probe.

Light decreases the accessibility of E-F loop position 154 in the SRII-HtrII complex
Probe accessibilities of the single-cysteine SRII mutants were then measured under 500-nm illumination in free SRII and HtrII-complexed SRII. In free SRII, none of the positions exhibit significant changes in accessibility upon photoactivation (Figure 2), confirming the logic of this approach since the residues were selected for the property of protruding outward into the cytoplasm in the crystal structure and therefore should be relatively insensitive to structural perturbations. However, in the presence of HtrII, photoactivation caused a decrease of 40% labeling efficiency in residue S154C in SRII and smaller but significant changes in S158C and L159C (Figure 2). We conclude that

FRET is eliminated by the signaling-deficient HtrII mutant G83F without loss of affinity for SRII. To test for a relationship of the FRET-detected proximity of SRII and
HtrII to function, we repeated the FRET measurements with the function-deficient mutant HtrII_G83F. FRET is eliminated in this mutant (Fig. 3), and therefore it became critical to determine whether the binding affinity, presumably determined by the extensive interactions between the SRII and HtrII transmembrane domains, was altered by the mutation.
To evaluate the SRII-HtrII binding constant in 0.1% DDM as used in the FRET assays, we applied two different methods. The first was based on an assay developed by Kamo and coworkers (20), which demonstrated that the photocycle of SRII_D75N was affected by the binding of HtrII and thus, flash photolysis of SRII_D75N can be used to determine its affinity for HtrII. The kinetics of the flash-induced absorption change at 570 nm in free SRII_D75N and that of SRII_D75N complexed with HtrII (Fig. 4A) shows that association with transducer resulted in changes in both the relative amplitudes and the rates of kinetic components in the photocycle of the SRII. We used the full amplitude of the differential signal (Fig. 4A, insert) to assess the extent of binding. Based on the concentration-dependence effect of HtrII on the photocycle of SRII_D75N ( Figure   4B), the Kd was determined to be 0.35 µM under our conditions. Similar measurement with HtrII_G83F produced a Kd of 0.22µM. These values are close to the affinities measured in previous reports (8,20), and show that the G83F mutation does not significantly inhibit HtrII binding to SRII_D75N.
Since the above measurement used SRII_D75N whereas the FRET measurements used SRII_S154C, we applied a second method using the fluorescent probe lucifer yellow vinyl sulfone (LY) attached to SRII_S154C as a fluorescent reporter group for the binding of HtrII. LY has been used successfully in several studies to report protein conformational changes (21) or protein association (22). LY-labeled SRII_S154C was divided into several different aliquots with equal volume and different concentrations of HtrII (Fig. 4C) or HtrII_G83F (Fig. 4D) were added. Both wild type HtrII and HtrII_G83F cause an increase in the LY fluorescence signal and the saturation curves are identical, confirming that the mutation does not inhibit binding. Therefore the loss of FRET between SRII_S154C-IAEDANS and Trp residues in HtrII_G83F is not due to loss of association and results instead from a change in HtrII structure evidently distancing the protein in the cytoplasmic region from SRII.

SRII and HtrII reciprocally inhibit disulfide cross-linking reactions in their membrane-proximal domains.
Two additional experiments were designed to test for physical interaction in the membrane-proximal domains of SRII-HtrII. First, disulfide cross-linking reactions of SRII_S150C and SRII_S154C were carried out with or without HtrII ( Figure 5A). The cross-linking efficiency at S150C was not affected by the presence of HtrII, whereas the cross-linking in S154C was abolished by association with HtrII. This result supports the fluorescent probe labeling data at S150 and S154, only the latter of which showed a significant decrease in labeling when complexed with HtrII, and provide further evidence for proximity of SRII and HtrII in the membrane-proximal region.
The converse experiment was performed to test whether SRII would inhibit disulfide cross-linking of the mono-Cys mutant HtrII_L90C. The presence of SRII reduced the cross-linking efficiency of L90C and it was further decreased when SRII was photoactivated ( Figure 5B). These results further indicate the existence of close proximity between SRII-HtrII in the membrane-proximal domain, and in addition, suggest that the contact is enhanced when the receptor is photoactivated, in agreement with the accessibility data.

DISCUSSION
The results above (i) provide compelling evidence for close proximity of E-F cytoplasmic loop residue Ser154 in SRII to residues 91-95 in the HtrII membrane proximal domain, (ii) show that SRII photoactivation alters the interaction in this region, and (iii) correlate the loss of interaction with the loss of function in HtrII mutant G83F.
The results suggest a model in which the crystallographically demonstrated (7) tight interactions between SRII and HtrII membrane-embedded domains hold the cytoplasmic domains of the receptor and transducer in close juxtaposition, and SRII photoactivation, which is believed to cause an outward displacement of helix F (14,16) that would in turn be expected to alter the structure of the E-F loop, alters the interaction in the membrane-proximal region during the signal relay process (Figure 6a).
Site-directed spin-labeling measurements revealed a light-induced transient decrease of nitroxide mobility in SRII S158C-R1 which faces HtrII TM2 (where R1 indicates the spin label attached to the Cys residue) and an increase in mobility of the probe attached at L159C, which faces away from TM2 on helix F. These results were interpreted as a movement of the helix toward TM2 of HtrII during activation (14,15). This movement fits well with the model suggested here. Ser154 is 2 helical turns below Ser158 facing in the same direction and the HtrII 91-95 region is on the cytoplasmic extension of TM2.
Assuming a similar structural change in SRII as in light-activated BR, the proposed motion of Ser154 toward HtrII 91-95 residues would entail tilting of helix F from Pro175 in the membrane-embedded portion of SRII, and would require the Ser158/Leu159 motion.
Cysteine-scanning through the length of TM2 from Ala60 to Ala88 showed that no single residue in the membrane-embedded SRI-HtrI interface is critical for phototaxis function (9). Furthermore mutation of SRII Tyr199 to Phe or to Ala, which eliminates a known hydrogen bond between the Tyr residue and an Asn residue on TM2 in the membrane domain, does not significantly impair phototaxis signaling assessed by computerized cell tracking (data not shown). In contrast to this permissive property of the SRI-HtrI interface within the bilayer, mutation of HtrII Gly83 to Cys or Phe causes a total loss of phototaxis responses in H. salinarum (9). Gly83 is the first residue beyond the membrane-cytoplasm interface in HtrII and begins the sequence Gly-Gly-Asp-Thr, which protein structure analysis ( Figure 6A) predicts to be a turn and a flexible region.
Also, a second turn and flexible region is predicted at positions 101-104, which would serve to bring the two post-turn helices of the HtrII monomers together in the coiled coil dimer motif expected from sequence algorithms for HtrII (11) and observed in the homologous Tsr transducer (23). In the model ( Figure 6B)