Early Photocycle Structural Changes in a Bacteriorhodopsin Mutant Engineered to Transmit Photosensory Signals*

Bacteriorhodopsin (BR) and sensory rhodopsin II (SRII) function as a light-driven proton pump and a receptor for negative phototaxis in haloarchaeal membranes, respectively. SRII transmits light signals through changes in protein-protein interaction with its transducer HtrII. Recently, we converted BR by three mutations into a form capable of transmitting photosignals to HtrII to mediate phototaxis responses. The BR triple mutant (BR-T) provides an opportunity to identify structural changes necessary to activate HtrII by comparing light-induced infrared spectral changes of BR, BR-T, and SRII. The hydrogen out-of-plane (HOOP) vibrations of the BR-T were very similar to those of SRII, indicating that they are distributed more extensively along the retinal chromophore than in BR, as in SRII. On the other hand, the bands of the protein moiety in BR-T are similar to those of BR, indicating that they are not specific to photosensing. The alteration of the O–H stretching vibration of Thr-204 in SRII, which we had previously shown to be essential for signal relay to HtrII, occurs also in BR-T. In addition, 1670(+)/1664(-) cm-1 bands attributable to a distorted α-helix were observed in BR-T in a HtrII-dependent manner, as is seen in SRII. Thus, we identified similarities and dissimilarities of BR-T to BR and SRII. The results suggest signaling function of the structural changes of the HOOP vibrations, the O–H stretching vibration of the Thr-215 residue, and a distorted α-helix for the signal generation. We also succeeded in measurements of L minus initial state spectra of BR-T, which are the first FTIR spectra of L intermediates among sensory rhodopsins.

Rhodopsins, 7-transmembrane receptors using retinal as a chromophore, are widespread among microorganisms and animals (2). The microbial rhodopsin family is unusual in that many are light-driven ion transporters, whereas other homologous members of the family are sensory receptors that interact with transducer proteins to activate signal transduction pathways (3). The archaeal halophile Halobacterium salinarum contains four different rhodopsins in the membrane. Two, bacteriorhodopsin (BR) 3 and halorhodopsin (HR), are light-driven ion pumps, and the other two, sensory rhodopsin I (SRI) and sensory rhodopsin II (SRII, also called phoborhodopsin, pR), are phototaxis receptors controlling motility (3)(4)(5). Crystal structures of BR (6) and SRII (7) show close similarities in architecture, helix positions, and location of the retinal-binding pocket (see Fig. 1). However the detailed structures are very different, because 74% of their residues differ. SRII forms a 2:2 signaling complex with its cognate transducer protein HtrII in the membrane (8), and signals transmitted from SRII to HtrII modulate a phosphotransfer pathway controlling the flagellar motor switch. Light absorbed by SRII triggers trans-cis isomerization of the retinal chromophore (9). This photoexcitation results in the sequential appearance of photointermediates K, L, M, and O, followed by return to the unphotolyzed (i.e. dark) form of the protein (4). Protein structural changes in these intermediate states alter the structure of the bound HtrII, a process called signal relay. The mechanism of signal relay, i.e. the nature of these structural changes, is a central question in current studies of sensory receptors (3). The signal relay mechanism from SR receptors to their cognate Htr transducers has become a focus of interest in part because of its importance to the general understanding of communication between integral membrane proteins, about which little is known.
To determine the minimal core of the signaling mechanism, we worked on defining the minimal modifications necessary to convert BR into a protein that would mimic SRII and relay the photosignal to HtrII. We recently found that just three residues in BR replaced by the corresponding residues in SRII enable BR to efficiently relay the retinal photoisomerization signal to * This work was supported by grants from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (15076202) (to H. K.) and by National Institutes of Health Grant R37GM27750 and a Robert A. Welch Foundation endowment (to J. L. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence may be addressed. HtrII and induce phototaxis responses at 35% of the efficiency of SRII (10). The signal relay-competent triple mutant of BR (designated "BR-T") contains the mutations P200T, V210Y, and A215T (Fig. 1). The surface mutations P200T and V210Y are important for binding to the HtrII molecule (11), whereas on the basis of our previous measurements in SRII (12,13), we predict that the replacement A215T is required for an HtrII-dependent steric constraint between the retinylidene C 14 -H and the Thr residue that occurs upon photoisomerization. The resulting hydrogen-bonding alteration between the residues Thr-215 and Tyr-185 in BR-T upon formation of the earliest intermediate (K) is essential for signal relay (10), as is the corresponding hydrogen bond in SRII (13). This finding provides an approach for identifying the structural changes necessary to activate the HtrII transducer protein by comparison of light-induced changes in BR, BR-T, and SRII. Here we have applied Fourier transform infrared (FTIR) spectroscopy, a powerful tool for investigating protein structural changes of rhodopsins in atomic detail (14). Earlier, we reported structural changes in BR and SRII detected by means of low temperature FTIR spectroscopy (15)(16)(17)(18). In this study, the FTIR spectra of BR-T in the presence and absence of HtrII are compared with those of BR and SRII.

EXPERIMENTAL PROCEDURES
Sample Preparations-Expression plasmids of BR-T and BR-T⅐HsHtrII were constructed as previously described (10). Proteins were expressed in H. salinarum Pho81Wr Ϫ cells as described previously (19). Pho81Wr Ϫ lacks the four rhodopsins (BR, HR, SRI, and SRII) as well as the two transducer proteins (HtrI and HtrII). To obtain a high expression level, the strong bop promoter was used. Purple membrane was prepared using essentially the same method as previously described (20).
FTIR Spectroscopy-Low temperature FTIR spectroscopy was as described previously (15,21). The samples were washed three times with a buffer at pH 7.0 (2 mM phosphate) for spectral measurements of K-and L intermediates. 90 l of the sample was dried on a BaF 2 window with a diameter of 18 mm. After  hydration by H 2 O, D 2 O, or D 2 18 O, the sample was placed in a cell, which was mounted in an Oxford DN-1704 cryostat placed in the Bio-Rad FTS-40 spectrometer. The light-adapted state was obtained by illumination of the film with Ͼ500 nm of light for 1 min at 273 K.
The BR-T K minus BR-T difference spectra were measured at 77 K as follows. To convert BR-T to BR-T K , the sample was irradiated for 2 min with 500 nm light; subsequent illumination with Ͼ640 nm light for 60 s reconverted BR-T K into BR-T. The difference spectrum was calculated from the spectra constructed with 128 interferograms collected before and after the photoillumination. Twenty-four spectra obtained in this way were averaged for the BR-T K minus BR-T spectrum.
The BR-T L minus BR-T difference spectra were measured at 170 K as follows. Illumination with Ͼ580 nm of light at 170 K for 1 min converted BR-T to the L intermediate. The spectrum by photoreversion of BR-T L to BR-T was not a mirror image of the BR-T L minus BR-T spectrum (forward photoreaction). Therefore, four spectra obtained by illumination were averaged to obtain the BR-T L minus BR-T spectrum. Fig. 2 shows the K minus initial state infrared difference spectra of BR (21), BR-T, and SRII ( Fig. 2, a-c) (15) measured at 77K and pH 7. In Fig. 2, a and c, the negative and positive bands at 1530(Ϫ)/1514(ϩ) and 1554(Ϫ)/1544(ϩ) cm Ϫ1 correspond to ethylenic CϭC stretching vibrations of the retinal chromophore in BR and SRII, respectively. The bands at 1203(Ϫ)/1194(ϩ) cm Ϫ1 for BR and 1204(Ϫ)/1198(ϩ) cm Ϫ1 for SRII correspond to the C-C stretching vibration of the chromophore at position C14 -C15. It is well known that these bands indicate the formation of the K intermediate (15). The similar frequency shifts, 1533(Ϫ)/1524(ϩ) cm Ϫ1 for CϭC stretching vibration and 1202(Ϫ)/1195(ϩ) cm Ϫ1 for C-C stretching vibration, were observed in BR-T, indicating the formation of the K intermediate. The bands at 1533(Ϫ)/1524(ϩ) cm Ϫ1 in BR-T are relatively small. This appears to be due to the smaller frequency shift (9 cm Ϫ1 ) than in BR (16 cm Ϫ1 ), resulting in mutual cancellation of the positive and negative bands. Also in SRII, the bands at 1554(Ϫ)/1544(ϩ) cm Ϫ1 (frequency shift ϭ 10 cm Ϫ1 ) are smaller than in BR (frequency shift ϭ 16 cm Ϫ1 ). One of the differences between BR and BR-T is the appearance of a positive band at 1157 cm Ϫ1 in BR-T. The band was also observed in SRII at 1159 cm Ϫ1 , suggesting similar structural changes of some C-C stretching vibration between BR-T and SRII upon formation of the K intermediate, not occurring in BR.

Infrared Spectral Changes of BR-T upon Formation of the K Intermediate-
HOOP Vibrations of the Retinal Chromophore in BR-T-A previous FTIR study (15) showed remarkable spectral differences in the region of hydrogen out-of-plane (HOOP) vibrations of the retinal chromophore (1000-800 cm Ϫ1 ) between BR and SRII, which provide information about the chromophore  (22). The SRII K minus SRII spectrum exhibits more peaks (Fig. 3c), which are interpreted in terms of more extended chromophore distortion in the K intermediate of SRII (15). By using all seven monodeuterated retinal analogues, we have assigned the 966(ϩ)/971(Ϫ) and 958(ϩ)/961(Ϫ) cm Ϫ1 bands as originating from the C7ϭC8 and C11ϭC12 Au (irreducible representation of diene vibrations) HOOP modes, respectively (23). The positive bands at 1001, 994, and 979 cm Ϫ1 were assigned to the C15-HOOP vibrations of the K intermediate. Another positive band at 864 cm Ϫ1 was assigned to the C14-HOOP vibration. These bands were also observed in BR-T at 964(ϩ)/971(Ϫ), 958(ϩ)/961(Ϫ), 996(ϩ), 992(ϩ), 979(ϩ), and 861(ϩ) cm Ϫ1 (Fig. 3b), suggesting that the structural changes spread to the middle part of the retinal chromophore. In addition, a positive band at 1023 cm Ϫ1 appears in BR-T K . A similar band is also present at 1024 cm Ϫ1 in the SRII K , which was assigned to symmetric rocking of methyl groups connected to the C9 and C13 atoms (23). In the BR spectra, two intense positive bands at 811 and 803 cm Ϫ1 appeared in the 900-800 cm Ϫ1 region. We predicted that the bands originated from wagging vibrations of the retinal chromophore, because resonance Raman studies revealed that such vibra-tions appeared in the 900-800 cm Ϫ1 region (24). In contrast, the bands disappeared in BR-T as well as in SRII. Thus these results indicate that the structural changes of the retinal chromophore of BR-T are very similar to those of SRII. Fig. 4 shows spectral changes in the 1780-1560 cm Ϫ1 region, where most signals originate from protein vibrations. One exception is the CϭN stretching vibration of the retinylidene Schiff base that appears in the 1650-1600 cm Ϫ1 region (25). The overall spectral shape of BR-T looks similar to that of BR, not SRII. The CϭN stretching vibrations of BR are located at 1641(Ϫ)/1608(ϩ) cm Ϫ1 in H 2 O and at 1628(Ϫ)/1606(ϩ) cm Ϫ1 in D 2 O (Fig. 4a), whereas those of SRII are located at 1657(Ϫ)/1600(ϩ) cm Ϫ1 in H 2 O and at 1633(Ϫ)/1597(ϩ) cm Ϫ1 in D 2 O (Fig. 4c). The bands at 1641(Ϫ)/1612(ϩ) cm Ϫ1 in H 2 O and 1628(Ϫ)/1608(ϩ) cm Ϫ1 in D 2 O appeared in the case of BR-T, which are similar frequencies to BR, implying that structure and structural changes of the CϭN stretching vibrations in BR-T are almost the same as those in BR. Fig. 4 also shows vibrational bands of the protein. The intense peaks at 1704(Ϫ)/1700(ϩ) cm Ϫ1 (Fig. 4c) originate form the CϭO stretching vibrations of the side chain of Asn-105 in SRII (26). Because the Asn is replaced by aspartate both in BR and BR-T, the spectrum lack the bands, but contain the D 2 O-sensi-  tive bands at 1742(Ϫ)/1732(ϩ) cm Ϫ1 for BR (27) and at 1740(Ϫ)/1731(ϩ) cm Ϫ1 for BR-T, respectively (Fig. 4, a and b), showing similar structural changes of the aspartic acid (Asp-115) between BR and BR-T. The band pair at 1623(ϩ)/1617(Ϫ) cm Ϫ1 in BR (Fig. 4a) was assigned to the CϭO stretching vibrations of amide I of Val-49 (28). In BR-T, a band pair was also observed at 1623(ϩ)/1618(Ϫ) cm Ϫ1 , indicating the similar structural changes of Val-49 between BR and BR-T. In BR, the 1668(Ϫ)/1664 cm Ϫ1 bands correspond to the frequency region typical for the amide I vibration of the highly dichroic ␣ II -helix (29). In contrast, the intensity of the bands was markedly decreased in SRII and BR-T. Because the ␣ II -helix possesses considerably distorted structure (29), the structural changes of distorted ␣-helix did not occur upon formation of the K intermediates in both BR-T and SRII. Except for the changes in the ␣ II -helix region, the spectra of BR-T is likely similar to BR in the typical amide I vibrational region (1670-1600 cm Ϫ1 ), suggesting that protein structural changes in BR-T are similar to that in wild-type BR. Fig. 5 shows a spectral comparison in the 2780-1850 cm Ϫ1 region between hydration with D 2 O (red lines) and D 2 18 O (blue lines) for BR ( Fig. 5a) (21), BR-T (Fig. 5b) and SRII (Fig. 5c) (30), where isotope shifts were observed for many bands. We assign the green-labeled bands to the O-D stretching vibrations of water because of the isotope shift. In BR, negative peaks at 2690, 2636, 2599, 2323, 2292, and 2171 cm Ϫ1 and positive peaks at 2684, 2675, 2662, 2359, and 2265 cm Ϫ1 were earlier assigned to vibrations of water molecules (21) (Fig. 5a). Because the frequencies of the negative peaks at 2321, 2292, and 2171 cm Ϫ1 are much lower than those of fully hydrated tetrahedral water molecules (31), the hydrogen bonds of those water molecules must be very strong. The strong hydrogen-bonded water molecules were observed not only in BR but also in BR-T (at 2323, 2292, and 2195 cm Ϫ1 ) and SRII (at 2307 and 2213 cm Ϫ1 ). A band of the strongest hydrogen-bonded water molecule at 2171 cm Ϫ1 in BR, which was assigned to the O-D stretching vibration of water 402 hydrating Asp-85 (32), shifted to 2195 cm Ϫ1 in BR-T. In BR-T, the introduced Thr-215 probably forms a hydrogen bond with Tyr-185 (as discussed below). Because Tyr-185 also forms a hydrogen bond with Asp-212, the hydrogen bond between them becomes weaker. Moreover, Asp-212 directly forms a hydrogen bond with water 402 (33). Because the hydrogen bond between Asp-212 and water 402 becomes stronger, the hydrogen bond between Asp-85 and water 402 becomes weaker. As a result, the band at 2171 cm Ϫ1 in BR shifts to 2195 cm Ϫ1 in BR-T. Supporting this interpretation, the candidate  band for the O-D stretching vibration of water 402 hydrating Asp-212 (2636 cm Ϫ1 ) seems to be down-shifted to 2628 cm Ϫ1 . Recently Fututani et al. (34) proposed that the existence of strongly bound water molecules was required for proton pumping, because it was well correlated with proton pumping activity of archaeal rhodopsins. Sudo and Spudich (10) confirmed that BR-T in the absence of bound HtrII does pump protons with similar efficiency as wild-type BR. Therefore, this result is consistent with the correlation between the strongly hydrogenbonded water molecule and pumping activity.

X-D (XϭO,N) Stretching Vibrations in BR-T-
The frequency region shown in Fig. 5 also contains X-D stretching vibrations other than water molecules. In the BR K minus BR spectrum, the bands at 2506(Ϫ)/2465(ϩ) cm Ϫ1 labeled in purple and the underlined bands at 2171(Ϫ) and 2124(Ϫ) cm Ϫ1 were assigned to the O-D stretching vibrations of Thr-89 (35,36)  Although not assigned directly by use of the labeled protein, the bands at 2179 and 2128 cm Ϫ1 in BR-T (Fig. 5b) are likely to originate from N-D stretching of the Schiff base, whose frequencies are very similar to those in BR (2171 and 2124 cm Ϫ1 ). This fact indicates similar hydrogen bonding strengths between BR-T and BR. The frequency of the intense band (2179 cm Ϫ1 for BR-T and 2171 cm Ϫ1 for BR) corresponds to the results obtained for the CϭN stretching vibrations as shown in Fig. 4. Intriguingly, a new negative band at 1993 cm Ϫ1 appears in BR-T (Fig. 5b). The corresponding band at 2004 cm Ϫ1 was also observed in SRII spectra (Fig. 5c) and was not observed in BR spectra (Fig. 5a). It is assignable to an X-D stretching vibration, because there is no band in the corresponding region of the difference spectra measured in H 2 O. However, this band does not originate from O-D stretching vibrations of water, because no isotope shift was observed in the sample hydrated with D 2 18 O. In addition, the band at 2004 cm Ϫ1 in SRII was not affected by [ 15 N]lysine-labeled SRII as reported previously (38). It may originate from an N-D stretching vibration of the guanidium group of Arg, which is suggested to form a strong hydrogen bond because of its relatively low frequency. In the case of BR, the bands at 2292(Ϫ)/2266(ϩ) cm Ϫ1 and at 2579(Ϫ)/2567(ϩ) cm Ϫ1 were assigned to the N-D stretching vibrations of Arg-82 (39). Frequency Region-Fig. 6, a and b, shows BR-T K minus BR-T and SRII K minus SRII infrared difference spectra in the absence (dotted lines) and presence (solid lines) of HtrII. The right panels show the spectra in the 1320-900 cm Ϫ1 region. Both dotted and solid lines in BR-T look very similar in this frequency region, implying that complex formation of BR-T with HtrII has almost no effect on the spectral changes in the 1320-900 cm Ϫ1 region upon retinal photoisomerization. It should be noted that BR-T is almost completely in complex with HtrII, because the BR-T⅐HtrII complex does not show measurable photoinduced proton transport activity (10). Therefore the lack of effect in this spectral region is not due to lack of binding. Furthermore, in BR-T, a spectral difference was observed in the frequency region of the amide I vibration of ␣ II -helix (1664(Ϫ)/1670(ϩ) cm Ϫ1 ) (Fig. 6a, left panel). The new bands appear at 1663(Ϫ)/1671(ϩ) cm Ϫ1 in an HtrII-dependent manner. As described above, because the ␣ II -helix possesses considerably distorted structure, the results suggest that the structural changes of distorted ␣-helix are caused by association with HtrII both in the case of BR-T and SRII. Another difference due to HtrII is that the negative band at 1653 cm Ϫ1 in BR-T without HtrII completely disappeared in BR-T with HtrII, and a new band appeared at 1655 cm Ϫ1 (Fig. 6a), suggesting that amide-I vibration of the peptide backbone of BR and/or HtrII was greatly altered by association with HtrII.

Spectral Differences Caused by Association with HtrII in Vibrations in the Low
The O-H Stretching Vibrations of the Introduced Thr Residue into BR- Fig. 7, a and b, shows the BR-T K minus BR-T and SRII K minus SRII infrared difference spectra in the absence and presence of HtrII. Also shown is the BR K minus BR infrared difference spectrum. Previously we reported that the hydrogen bond between Thr-204 SRII and Tyr-174 SRII was greatly altered by formation of the K intermediate in a HtrII-dependent manner (Fig. 7b), and the O-H stretching vibration of the hydroxyl group of Thr-204 exhibited a frequency downshift of 110 cm Ϫ1 (40). This result suggested that a specific hydrogen bonding alteration between Thr-204 and Tyr-174 takes place upon retinal photoisomerization, because it has been reported (35,36) that the frequency shifts are very small in BR at 77 K (18 cm Ϫ1 for Thr-17, 13 cm Ϫ1 for Thr-121, and ϳ60 cm Ϫ1 for Thr-89 in BR). The BR-T used in this study has the Thr residue as A215T (Ala-215 corresponds to Thr-204 in SRII) as described above. Comparison of wild-type BR (Fig. 7a, ⅐⅐⅐) with BR-T (Fig. 7a, ---) revealed that enhanced bands at 3518(Ϫ), 3413(ϩ), 3390(ϩ), and 3368(ϩ) cm Ϫ1 appeared in BR-T without HtrII. The complex formation between BR-T and HtrII (Fig. 7a, solid lines) has essentially no effect on these bands, unlike in SRII. In BR, the bands at the 3480(ϩ)/3462(Ϫ) and 3415(ϩ)/3402(Ϫ) cm Ϫ1 were assigned to the O-H stretching vibrations of Thr-17 and Thr-121, respectively (35,36). Corresponding bands seem to be present in BR-T. We tentatively assigned the band at 3518(Ϫ) cm Ϫ1 as the O-H stretch of introduced Thr-215 in BR-T. Although we cannot identify which positive band originates from Thr-215, in all cases, the results suggest that the hydrogen bond of the introduced Thr residue strengthens by formation of the K intermediate due to the low frequency shifts. These frequency changes in BR-T (105, 128, or 150 cm Ϫ1 ) are similar to those in SRII (110 cm Ϫ1 ).
Infrared Spectral Changes of BR-T upon Formation of the L Intermediate-The illumination of rhodopsins elicits a linear and cyclic sequence of spectrally distinct transitions (photocycle), during which these pigments function (4). The photo-cycle intermediates in SRII are denoted as K, L, M, N, and O and are analogous to those of BR. We can trap and measure the BR L minus BR spectra at 170 K using FTIR spectroscopy (21). However, in the case of SRII, the high thermal stability of the K intermediate makes it difficult to trap and measure the SRII L minus SRII spectrum (15).
In this study, we succeeded in measurement of the L intermediate of BR-T, which is the first report among archaeal sensory-competent rhodopsins. Fig. 8b, solid lines, shows the L minus initial state infrared difference spectra of BR, transducerfree BR-T, and BR-T⅐HtrII complex measured in D 2 O condition. The CϭC stretching vibrational bands at 1555(ϩ)/ 1533(Ϫ) cm Ϫ1 , the C-C stretching bands at 1202(Ϫ)/1194(ϩ) cm Ϫ1 , and the C15-HOOP vibration at 986 cm Ϫ1 , which are characteristic of the BR3 L transition, appeared in the BR-T L minus BR-T spectra (Fig. 8a, solid lines), implying the formation of an L intermediate in BR-T. In contrast to the BR-T K minus BR-T spectra, the difference spectra for L intermediates of BR and BR-T are identical in the 1600-800 cm Ϫ1 region, indicating that the HOOP vibrations of the retinal chromophore are restored upon formation of L intermediate. The wagging vibrations of the retinal chromophore are also restored upon formation of L intermediate. In addition, the other bands in the 1600-800 cm Ϫ1 region were not perturbed by association with HtrII (Fig. 8b), implying no HtrII-effect on BR-T. On the other hand, the amide I vibrational region (1720-1640 cm Ϫ1 ) in wild-type BR (Fig. 9a, dotted line) was quite different from that in BR-T (Fig. 9a, solid line). The bands at 1688(ϩ) and 1679(ϩ) cm Ϫ1 and at 1674(ϩ), 1667(Ϫ), 1657(Ϫ), and 1648(ϩ) cm Ϫ1 disappeared and appeared in BR-T, respectively (Fig. 9a). Because the bands are nearly H-D exchange-independent (data not shown), this suggests the perturbation of the peptide backbone upon formation of the L intermediate of BR-T. In the complex of BR-T with HtrII, new positive and negative peaks appeared at 1672 and 1665 cm Ϫ1 , respectively (Fig. 9b, solid  line). This result suggests that a distorted ␣-helix is caused by association with HtrII upon formation of the L intermediate as well as the K intermediate. Intriguingly, the frequency is almost the same as that in the BR-T K minus BR-T spectra (1670(ϩ) and 1664(Ϫ) cm Ϫ1 ) (see Fig. 6a).

DISCUSSION
The light-dark difference FTIR spectra of the SRII⅐HtrII molecular complex is rich in information because myriad bond energy changes occur throughout the complex in response to light. The challenge is to sort out which of these vibrational changes are crucial to the signaling process and which are dispensable, i.e. of minor or even of no importance to signaling. The BR-T protein offers a way to help distinguish functionally important versus relatively unimportant light-induced changes in structure. Structural changes shared by the signal-competent BR-T and SRII (but not present in BR) are likely to be necessary changes for signal relay to HtrII. The following structural changes are in this category and therefore implicated in signal relay function: (i) hydrogen out-of-plane (HOOP) vibrations, indicating that steric constraints on the retinal chromophore in the binding pocket of BR-T are distributed more extensively along the retinal chromophore than in BR, as in SRII; (ii) HtrIIdependent bands near 1670(ϩ)/1664(Ϫ) cm Ϫ1 , indicative of a distorted ␣-helix; (iii) formation of a strong hydrogen bond in the earliest intermediate (K), between Thr-215 and Tyr-185 in BR-T, corresponding to the Thr-204 and Tyr-174 hydrogen bond in SRII, shown to be essential for signaling in the SRII⅐HtrII complex (13); and (iv) a negative band at 1993 cm Ϫ1 appearing in BR-T, corresponding to a band at 2004 cm Ϫ1 in SRII, and not observed in the BR spectra. This band may originate from an N-D stretching vibration of Arg.
Therefore the structural changes of the HOOP vibrations, the O-H stretching vibration of the introduced Thr residue (A215T), and a distorted ␣-helix appear to be important for signal generation. The results from several different methods show that light-induced structural changes occur all along the SRII⅐HtrII interface, which includes the region on the periplasmic side of the membrane (41), the membrane-embedded domain (42)(43)(44), and the cytoplasmic membrane-proximal domain (45,46). The structural changes detected in this study further suggest that the signal relay occurs within the membrane-embedded contact surfaces.
The vibrational bands of the CϭN stretching vibrations of the retinal Schiff base, the protein moiety, and X-D stretching vibrations in BR-T are more similar to BR than SRII, indicating that the spectral differences between BR and SRII in these regions are not important for the signal transfer to HtrII. In contrast, HOOP vibrations of the retinal chromophore in BR-T are very similar to those of SRII. It is notable that SRI, another phototaxis sensor in H. salinarum, also exhibits more vibrational bands in the HOOP region than does BR. 4 In BR-T, the changes of HOOP vibrational bands are presumably caused by the A215T mutation, which is located near the retinylidene Schiff base, because the spectrum of the K intermediate in BR-P200T/V210Y is almost identical to that of wild-type BR (data not shown). In SRII, by using all monodeuterated retinal analogues, we reported a local steric constraint between the C 14 -H of the retinal chromophore and protein caused by C 13 ϭC 14 double bond rotation (12). The protein contributor to this steric conflict of the C 14 -H group in SRII K may be Thr-204, because the structure of SRII K shows that the distance to Thr-204 is significantly reduced upon isomerization from 4.1 to 3.3 Å. A similar structural change appears to be induced in BR-T. In SRII, Thr-204 and its hydrogen-bonded partner Tyr-174 are crucial residues for photosignal relay (13). The hydrogen bond between them is greatly strengthened upon the formation of the earliest SRII photointermediate (SRII K ) only when SRII is complexed with HtrII (40)  SRII, i.e. Thr-204, Tyr-174), 24 cm Ϫ1 (T204S), 0 cm Ϫ1 (T204A, because of no O-H groups), and 28 cm Ϫ1 (Y174F) as reported previously (40). The lower values for T204S, T204A, and Y174F correlate with the lack of signaling function by these mutants (13). In BR-T, we observed D 2 O-insensitive new bands at 3518(Ϫ), 3413(ϩ), 3390(ϩ), and 3368(ϩ) cm Ϫ1 in the X-H frequency region. These bands are not HtrII-dependent as in SRII, but nevertheless the results indicate that a Thr-215 hydrogen bond becomes very strong upon formation of a K intermediate occurring in SRII. Thr-204 hydrogen bonds with Tyr-174 according to the crystal structure (8). Therefore the hydrogen bonding partner of Thr-215 in BR-T may be Tyr-185 in analogy to SRII. The structural changes of the O-H stretching vibration of Thr-215 in BR-T may be important for signal relay to HtrII, because the shift values upon formation of the K intermediate are well correlated with the function in the SRII⅐HtrII complex as described above.
In this study, we also measured the FTIR spectra of the L minus initial state. The high thermal stability of SRII K makes it difficult to measure the SRII L minus SRII FTIR spectra (15). BR-T mediates negative phototaxis as does SRII. Therefore, this is the first report of structural changes in L in sensory rhodopsins revealed by FTIR spectroscopy. The results show that steric constraints occurring upon formation of the K intermediate were restored upon formation of the L intermediate. Because the spectra of BR-T are almost identical to BR in the 1600-800 cm Ϫ1 region (Fig. 8), the stored energy in K is expected to induce the structural changes of peptide backbone structure appearing in the amide I vibrational region (1720-1640 cm Ϫ1 ) upon formation of the L intermediate (Fig. 9a).
Our results also demonstrate HtrII-binding effects on the spectra of BR-T. Both in the case of K and L intermediates, we observed new bands at 1670(ϩ)/1664(Ϫ) cm Ϫ1 for K and at 1672(ϩ)/1665(Ϫ) cm Ϫ1 for L, which originate from amide I vibrations of ␣ II -helix (Figs. 6a and 9b). Similar bands at 1671(ϩ)/1663(Ϫ) cm Ϫ1 were observed in SRII in a HtrII-dependent manner (Fig. 6b) (17). 13 C labeling of SRII or HtrII revealed that these spectral changes originate from the SRII helices and not those of HtrII (17). SRII binds to TM1 and TM2 of HtrII via the F and G helices of SRII (8,47). Therefore, the spectral changes probably originate from F and/or G helices.
The comparative study of BR, BR-T, and SRII reveal the functional importance of the structure and structural changes of HOOP vibrations, ␣ II -helix, and an O-H stretching vibration. The next steps in signal relay and signal propagation through HtrII require further study. According to data from several laboratories (3), these structural changes may induce outward F-helix tilting of SRII (48), structural changes of binding surfaces between SRII and HtrII (16,43,44), rotation of TM2 of HtrII, and structural changes in the membrane proximal domain (HAMP) of HtrII (45,49) to relay the signal to the phosphorylation cascade controlling the flagellar motor switch.