Light is an important environmental signal for all organisms on earth because it is essential for physiological signalling and the regulation of most biological systems. Halophiles found in salt-saturated ponds encode various archaeal rhodopsins and thereby harvest various wavelengths of light either for ion transportation or as sensory mediators. HR (halorhodopsin), one of the microbial rhodopsins, senses yellow light and transports chloride or other halides into the cytoplasm to maintain the osmotic balance during cell growth, and it exists almost ubiquitously in all known halobacteria. To date, only two HRs, isolated from HsHR (Halobacterium salinarum HR) and NpHR (Natronomonas pharaonis HR), have been characterized. In the present study, two new HRs, HmHR (Haloarcula marismortui HR) and HwHR (Haloquadratum walsbyi HR), were functionally overexpressed in Escherichia coli, and the maximum absorbance (λmax) of the purified proteins, the light-driven chloride uptake and the chloride-binding affinity were measured. The results showed them to have similar properties to two HRs reported previously. However, the λmax of HwHR is extremely consistent in a wide range of salt/chloride concentrations, which had not been observed previously. A structural-based sequence alignment identified a single serine residue at 262 in HwHR, which is typically a conserved alanine in all other known HRs. A Ser262 to alanine replacement in HwHR eliminated the chloride-independent colour tuning, whereas an Ala246 to serine mutagenesis in HsHR transformed it to have chloride-independent colour tuning similar to that of HwHR. Thus Ser262 is a key residue for the mechanism of chloride-dependent colour tuning in HwHR.
Solar light is a crucial environmental signal for most life forms, as it is intricately involved in energy harvesting and many physiological signals, including circadian rhythm and phototaxic responses . Featuring biological systems with light-driven systems, halophiles are known to exist widely in salt-saturated ponds. They encode four types of archaeal rhodopsins that respond to different wavelengths of light either by transporting ions, as observed in BR (bacteriorhodopsin) and HR (halorhodopsin) [2–4], or by serving as illumination sensors that trigger different phototaxic responses typified by the SRs (sensory rhodopsins) I and II found in Halobacterium salinarium [5–7].
All four types of microbial rhodopsins are not simultaneously encoded in the genome of a single species as they are in H. salinarum . Only one HR and one SRII were found to exist in Natronomonas pharaonis , whereas one BR, one HR and one MR (middle rhodopsin) with ambiguous function were identified in H. walsbyi without any SR [9,10]. H. marismortui, by contrast, encodes six rhodopsins, including two BRs, one HR, SRI, SRII and SRM, which is the greatest number of microbial rhodopsins found in a single archaeon [11,12].
HR, however, is the most prevalent microbial rhodopsin in archaea. More than a dozen HRs have been either confirmed by protein studies or proposed in various genomic projects (Supplementary Table S1 at http://www.bioscirep.org/bsr/032/bsr0320501add.htm). The first HR was identified in HsHR (H. salinarum HR) and the second one was later found in NpHR (N. pharaonis HR). Previous studies have described HRs as light-driven inward anion/chloride translocators and have suggested that they facilitate BR-assisted energy production  and preserve the cytoplasmic osmolality equilibrium . HRs therefore assist cells in producing energy and maintaining their cellular environments. It is reasonable to expect to find HRs in almost any microbe residing in a hypersaline habitat, as HRs play such an essential role for cell survival in crystallizer conditions.
Various biological and biophysical properties of HsHR and NpHR have been reported [14–16]; both HRs have been shown to absorb light maximally at approximately 575 nm, and the light-activated HRs initialize inward transport of chloride and other halides (Br−, I−) . The molecular mechanism for this light-driven chloride transport has been investigated in several crystallography studies for both HsHR and NpHR [18,19]. Further studies have shown their similarity in chloride-binding affinities [20,21], pKa [20,22] and ms-ranged photocycle kinetics , as well as shifts in their λmax (maximum absorbance wavelengths) under low-chloride or low-salt concentrations. The environment-dependent λmax shift is one of the main reasons that an engineered HR such as NpHR 3.0 , a mutant based on NpHR that was designed to serve as a nerve de-activation signal, was developed for optogenetic technology.
The HR gene proposed in the H. walsbyi genome  is intriguing because H. walsbyi cells are one of the most widely distributed halobacteria, globally present from shallow ocean coastal areas to crystallized ponds , with a variety of chloride concentrations. All HRs from other species were shown to be sensitive to chloride concentrations [20,21]. The ability of this newly proposed HwHR (H. walsbyi HR) to survive in environments with great variation in salinity or osmolality is investigated in the present study.
The HwHR, together with HmHR (Haloarcula marismortui HR), was first overexpressed and then purified for in vitro characterization. Both HmHR and HwHR were found to share most of the conserved functions and properties that have been observed for other known HRs; however, HwHR showed unique salt-independent changes in λmax, a feature not observed in any of the other HR proteins characterized to date. A single residue substitution, which was chosen via structure-based sequence alignment, eliminated such chloride-independent spectral tuning.
MATERIALS AND METHODS
Bacterial strains and plasmids
Escherichia coli DH5α cells were used for cloning and E. coli C43(DE3) cells were used for protein expression. The genes of HmHR  and HsHR  were cloned from genomic DNA, and HwHR  was synthesized by Genomics with the addition of a NcoI restriction enzyme cutting site before the start codon and a XhoI restriction enzyme cutting site after the stop codon. The DNA fragment was treated with restriction enzymes and ligated into a NcoI–XhoI-treated pET-21d vector. As described previously , the expressed N- and C-terminal peptide sequences are as follows: HmHR: 1MTAAST… ….TPADD276KLAAALEHHHHHH and HwHR: 1MAQHLY… ….AVADD292LEHHHHHH. The site-directed mutagenesis was performed according to the instruction manual from the QuikChange™ Lightning Site-Directed Mutagenesis Kit.
Protein expression and purification
A single colony of transformed E. coli C43(DE3) cells was inoculated in LB (Luria–Bertani) medium supplemented with 50 μg/ml of ampicillin and incubated at 37°C overnight. For large-scale protein expression, a 1:100 (v/v) dilution of overnight culture was added to a fresh LB/ampicillin medium and incubated at 37°C. When the D600 of the culture reached 0.4–0.6, IPTG (isopropyl β-D-1-thiogalactopyranoside; final concentration 1 mM) and all-trans retinal (final concentration 5–10 μM) were added for induction. Following subsequent incubation for 4–6 h in the dark, cells were collected by centrifugation at 6750 g for 10 min at 4°C (Hitachi CR-21, R10A3). The collected cells were resuspended in buffer A (50 mM Tris/HCl, 4 M NaCl, 14.7 mM 2-mercaptoethanol and 0.2 mM PMSF, pH 7.8) and broken by ultrasonic processing (S-3000; MISONIC). For the separation of the membrane fraction, total cell-extract centrifugation was performed at 2330 g for 30 min at 4°C (KUBOTA 5910). Then, the supernatant was centrifuged at 169538 g for 1 h at 4°C (Beckman L-90, Ti-70). The sediment was dissolved in buffer B (buffer A supplemented with 1% DDM (n-dodecyl-β-D-maltoside) for at least 12 h at 4°C, followed by centrifugation at 32816 g for 45 min at 4°C (Hitachi CR-21, R20A2) to separate the detergent-soluble fraction. Solubilized HRs were purified by affinity purification using the Ni-NTA (Ni2+-nitrilotriacetate) method. The detergent-soluble solution containing 20 mM imidazole was incubated with Ni-NTA agarose at 4°C for 6–8 h of slow nutation. It was then transferred to a chromatography column and washed with buffer C (buffer A with 0.05% DDM and 50 mM imidazole). The target HRs were eluted with buffer D (buffer A with 0.05% DDM and 250 mM imidazole). Purified HRs were concentrated and exchanged into buffer E (50 mM Mes, 4 M NaCl and 0.05% DDM, pH 5.8) with a protein concentrator (Millipore, Amicon, cut-off size of 30 kDa).
Light-driven ion transporter assay
E. coli cells with overexpressed HR were prepared as described previously using protein overexpression and purification protocols [12,28]. Induced cells without all-trans retinal addition served as a control. The collected cells were washed twice using an unbuffered solution (10 mM NaCl, 10 mM MgSO4 and 100 μM CaCl2) and suspended with an adequate amount of fresh buffer. The suspended cells were illuminated with green laser light, and the extracellular pH change was monitored with a probe (Eutech Instruments, CyberScan pH 2100). After 100 s of illumination, the dark incubation of an additional 100 s was also recorded. The illumination followed by dark incubation procedure was repeated three times. To check for passive proton uptake, experiments using suspended cells with CCCP (carbonyl cyanide m-chlorophenylhydrazone; Sigma C2759; to a final concentration of 10 μM) were performed as described above. All pH changes were recorded with the CyberComm Pro program and exported as.csv files.
The purified proteins were diluted with buffers at different pH values (100-fold dilution). UV/Vis absorption spectra of 12 protein samples of pHs varying from 5.8 to 10.93 were recorded. To evaluate the wavelength that indicated the maximum absorbance change, all spectra were plotted together or subtracted. In HRs, the acidic–basic absorbance at 575 nm (HmHR) and 572 nm (HwHR) was calculated in relative units and plotted against the pH values. The sigmoid curve generated was fit to the following equation: Abs=Absacidic+Absbasic-acidic×(10pH−pKa). A similar equation named DoseResp in Origin® was also used: y=A1+(A2−A1)/1+10(LOGx0−x)p; the symbol LOGx0 was denoted as the pKa value.
Chloride titration assay
The purified HwHR was adjusted to different chloride concentrations to determine the chloride-binding affinity as described in previous studies with a slight modification [19,20]. The NaCl concentration was reduced from 4 M to 0.015 mM in a 50 mM Mes buffer, pH 5.8 (or in 50 mM Mops buffer, pH 7.0) At each chloride concentration, the UV/Vis spectrum was recorded and normalized using A280 as a standard. The amplitude changes of the absorbance in relative units were plotted against the chloride concentration (log10 scale), then the resulting curve was fitted to the Hill equation (R2>0.95). The following Hill equation was adopted for analysis: y=(Vmax×xn)/(kn+xn); k denotes the dissociation constant.
Flash-laser induced photocycle measurements
Nd-YAG laser (532 nm, 6 ns pulse, 40 mJ) flash-absorbance changes were measured by employing a laboratory-constructed laser cross-beam flash spectrometer as previously described . The purified proteins were dissolved in buffer E to reach 0.3 at a maximum-wavelength (λmax) (Amax), and transient-absorbance changes were recorded at their corresponding ground-state maximum. The curves represent the loss and recovery of absorbance at the target wavelengths upon green laser (λex=532 nm).
Sequence alignment of known and proposed HR genes from Halobacteria
HRs are prevalent in Halobacteria; more than 15 strains (Supplementary Table S1) of Halobacteria have been completely sequenced to date, and at least nine of these were proven or predicted to have hop genes. A phylogenic tree (Figure 1a) and sequence alignment (Figure 1b, and Supplementary Figure S1 at http://www.bioscirep.org/bsr/032/bsr0320501add.htm) of 17 annotated HRs showed high sequence identity (55–68%) among them, and the previously reported key residues involved in chloride transportation, i.e. Arg108/Thr111 and Arg200/Thr203 in HsHR , were found to be conserved in all HRs. Although the classification of HRs were not correlated with the evolution , HmHR and HwHR, which were the focus in the present study, showed higher sequence similarity with NpHR and HsHR respectively.
Bioinformatic analysis of microbial HRs and structural alignment of HsHR and HwHR
Purification and spectral characterization of HmHR and HwHR
To examine and compare the spectral characteristics of HR proteins, we overexpressed and purified His6-tagged HmHR and HwHR proteins using the E. coli C43(DE3) system according to previously described protocols . This procedure yielded several mg/l of culture.
The UV/Vis spectra showed the λmax of HmHR to be 576 nm (Figure 2a), whereas the λmax of HwHR was 573 nm (Figure 2b). Compared with all well-known HRs, all of them were within approximately 3 nm of 575 nm (summarized in Supplementary Table S3 at http://www.bioscirep.org/bsr/032/bsr0320501add.htm).
Spectroscopic characterization and functional determination of HmHR and HwHR
Light-driven chloride transport activity measurements
Light-driven inward chloride transport measurements for HmHR and HwHR were performed to examine their function as illumination-dependent chloride translocators, using previously described procedures . Light-dependent pH increases were recorded in both HmHR (Figure 2c) and HwHR (Figure 2d) samples with or without CCCP. The observed tendency of light-driven proton transport in both HRs were in fact passive processes accelerated by light-driven chloride inward transportation as explained in a previous study , indicating that light-driven inward chloride transport occurred. Thus both HmHR and HwHR proteins demonstrated light-driven inward chloride transport, confirming their conserved function as observed in two other HRs.
Flash laser-induced absorbance changes of HmHR and HwHR
To investigate the photocycle, the defining feature of the two HRs, time-resolved flash photolysis was measured. The data at different salt concentrations are shown along with time-dependent absorbance changes at 580 nm. The half-life of the ground state recovery is 4.17 and 3.23 ms of HmHR and HwHR respectively, under 4 M NaCl (Figures 2e and 2f). When exposed to a buffer of low ionic strength with 5 mM NaCl and 333 mM Na2SO4, the recovery times are slightly different in both HRs (Supplementary Table S2 and Figure S3 at http://www.bioscirep.org/bsr/032/bsr0320501add.htm).
Chloride-binding affinity of HmHR and HwHR
The H. walsbyi cells were shown to tolerate 2 M MgCl2, an unusual environment for any other currently known HRs. To investigate this observation, the chloride-dependent spectral shifts of HwHR and HmHR were measured (Figure 3). The chloride-dependent spectral shift indicates the binding affinity of chloride ions to HRs, which is analogous to the binding behaviour of a ligand to a receptor, and it can be described via the Hill equation . The results showed that, for HmHR (Figure 3a), an approximately 15 nm shift was observed in conditions of less than 1 M of chloride. The spectral shifts ceased at 591 nm when the concentration of NaCl fell below 1 mM, and the so-called ‘blue HR’ (Figure 3a, inset) was formed as observed in other studies [11,19]. HmHR had no spectral changes from 4 to 1 M NaCl. HwHR (Figure 3b), by contrast, showed no significant spectral shift throughout various chloride concentrations, with only a slight 2–3 nm of either redshifted or blueshifted changes observed in the range of 250 mM to 15 μM NaCl before finally settling at 576 nm, close to the ground state.
Chloride-dependent spectral changes in the absorption spectrum of HRs obtained from E. coli membrane
Furthermore, when both HmHR and HwHR were placed in a low-chloride or chloride-free buffer, HmHR became ‘blue HR’ as mentioned above, while HwHR remained purple without a significant λmax change (Figure 3b, inset). As previous studies showed that under very low chloride conditions, the λmax of HsHR  was reported to be 568 nm (a blueshift of approximately 10 nm from 578 nm), and the λmax was 599 nm, a redshift of approximately 20 nm, in NpHR , HwHR thus showed an exceptionally high tolerance to any chloride concentration changes.
Further analysis of the calculated difference spectra of HmHR (Figure 3c) and HwHR (Figure 3d) indicated that the decrease of chloride concentration made both HRs show a decreased fraction at approximately 550 nm and an increased fraction at approximately 650 nm. An isosbestic point was observed at approximately 600 nm in HwHR but not in HmHR, further suggesting that a one-step transition conformational change from the chloride-unbinding to binding state occurred in HwHR but not in HmHR. The amplitude changes (ΔA) at 650 nm were used to determine the dissociation constant Kd values, and they were 7.47 mM (Figure 3e) and 2.74 mM (Figure 3f) for HmHR and HwHR respectively. Those values were similar to the previously reported 10 mM for HsHR and 1 or 3 mM for NpHR [16,20,31], indicating the similarity in chloride binding affinity among all four HRs.
The structural-based sequence analysis of HRs for the unique chloride-independent spectra feature
Further investigating the retinal binding pocket at approximately 10 Å (where 1 Å=0.1 nm), we identified two unique residues that were not conserved only in HwHR (Figures 1b and 1c); one was Thr186 and the other was Ser262 of HwHR. Both residues are typically alanine in other known HRs (Figure 1b, and Supplementary Figure S1). According to a structural model created with Swiss Model, the Thr186 in HwHR points out towards the membrane region on the helix E, but Ser262 is orientated inward towards the Schiff base, the chloride-binding pocket on helix G (Figure 1c, right-hand panel).
The residue Ser262 of HwHR eliminated the unique feature of HwHR in the chloride-dependent λmax shift
As Thr186 and Ser262 are both typically conserved alanine residues in other known halobacterial HRs, single residue substitutions of T186A and S262A in HwHR were constructed. The S262A-HwHR eliminated the chloride-independent λmax change feature (Figure 4b), as a blueshift of approximately 8 nm under low chloride conditions was observed but T186A-HwHR (Figure 4a) behaved like the wild-type. However, the light-driven chloride-pumping capability and the photocycle of S262A were comparable with the wild-type (Supplementary Table S4 at http://www.bioscirep.org/bsr/032/bsr0320501add.htm). The chloride-binding affinity assay showed that the Kd was 10.40 mM for the S262A-HwHR, comparable with HsHR. The Hill coefficient (n) was 0.91, suggesting no significant change in chloride-binding from HwHR.
Chloride-dependent spectra of various mutants of HwHR and A246S HsHR
HsHR planted with corresponding Ser262 showed no chloride-dependent λmax change
The Ser262 in HwHR was engineered to HsHR, which was determined to be closer in sequence alignment, in order to create an A246S-HsHR, and the chloride-dependent λmax shift was eliminated (Figure 4c, and Supplementary Table S4); it showed a stable λmax at 568 nm from 4 M to 0.4 mM NaCl solutions.
HR proteins represent an effective biological mechanism that directly elevates the capability of early life forms to cope with extremely high salt environments, which is the case for most Halobacteria. In the present study, HwHR from H. walsbyi was shown to function as a light-driven chloride transporter with a stable λmax even under almost chloride-free conditions, a feature not yet observed in any other known HRs. The serine residue at position 262 of HwHR was found to act as a spectrum stabilizer, causing such chloride-independent λmax change.
Chloride binding is known to be important for maintaining a stable λmax, but the chloride-binding network in HwHR might have slight variations from other known HRs. According to the structural model of HwHR (Figure 1c), the Ser262 located at helix G faces the Thr66 at helix B, and together they can form a chloride stabilizer with their polar hydroxyl groups. Thr66 is known to be important in chloride translocation paths, and Ser262 is an alanine, a non-polar group, in all other annotated HRs except for Natrialba magadii and Salinibacter ruber. When A246S-HsHR was constructed to have this Ser262 corresponding to HwHR, the chloride-dependent λmax shift was eliminated, and it stayed at 568 nm from 4 M to 0.4 mM NaCl solutions.
It is mainly retinal and its surrounding chemical environment that contribute the colour, or absorbance, of a retinal-binding protein. Retinal divides the HRs into extracellular and cytoplasmic regions. Among them, residues Arg108/Thr111 locate in the extracellular region in HsHR and they are involved in transportation of an anion into the cytoplasmic region, which contains three conserved residues, Thr66/Arg200/Thr203 [32,33]. On the other hand, the Ser262 in HwHR and Ala246 in HsHR identified in the present study appear to locate near the cytosolic region containing conserved residues Thr82/Arg216/Thr219 in HwHR and Thr66/Arg200/Thr203 in HsHR that were known to be important for chloride translocation. Specifically, during the catalytic cycle of HRs, the chloride ion was proposed to move to Thr203 before being released into the cytosol [18,34]. Therefore a slightly different chloride translocation network in the cytoplasmic region formed via residues Thr82/Arg216/Thr219/Ser262 in HwHR can be thus proposed.
This explanation is consistent with previously reported observations that the λmax shifts occur in various HRs (Supplementary Figure S4) under 0–4 M NaCl environments: NpHR  showed a 22 nm blueshift, a 13 nm redshift in HsHR  and a 15 nm blueshift in HmHR, while either a red- or blueshift within 3 nm was detected in HwHR. In the chloride-free environment, HwHR was the only HR that showed no conversion into the so-called ‘purple HR’ Other results in the present study also support HwHR having a different chloride binding conformation. First, the lack of a clear isosbestic point in HmHR (Figure 4a) suggested that there are either multiple binding sites for the chloride or multiple conformations of the protein, but not in HwHR. Secondly, the Hill coefficient depicts the co-operation of ligand binding, and HmHR had a slightly larger Hill coefficient than that of HwHR (nHmHR=1.27, nHwHR=0.79), suggesting that there was most likely a slight variation in the ionic network for chloride binding among the residues between HmHR and HwHR.
HwHR might have a single chloride-binding site instead of two. Two chloride-binding sites were found in HsHR , and they occurred in the Schiff-base region, as in most HRs, with a three-residue binding motif composed of two positively charged arginine residues and a glutamate residue. Sequence analysis (Figure 1b, and Supplementary Figure S1) showed the second non-Schiff base chloride-binding site, which stabilized chloride with residues Arg24, Arg103 and Gln105 in HsHR, is not conserved in HwHR.
As the hydroxy groups exist in both side chains of serine and threonine, and both amino acids also occupy similar spaces in the side chains, the explanation for Ser262 to have such an impact in chloride-dependent λmax changes of HwHR requires further study.
In conclusion, HwHR had an unexpectedly stable λmax regardless of environmental chloride concentration, a feature not previously observed in any other HR. This feature is, at least partially, contributed by the residue Ser262, and the residue's arrangement that is determined by the structural conformation can explain this observation. This unique tolerance of HwHR for a chloride-free environment could further provide information in understanding the colour tuning in HR and the strategies adopted by early life forms to survive ever-changing environments.
Hsu-Yuan Fu, Yung-Ning Chang and Chii-Shen Yang contributed to writing the paper. Hsu-Yuan Fu and Chii-Shen Yang contributed to the organization of the paper. Yung-Ning Chang performed the wild-type HRs biochemical and biophysical analysis. Hsu-Yuan Fu performed the site-directed evolution approach including construction, mutant protein analysis and data arrangement. Ming-Jin Jheng performed the light-driven pump assay for mutant HRs.
This work was supported by Tai-He Yang (Giant Lion Know-How Co. Ltd, Taiwan).
These authors contributed equally to this work.