Critical Role of Transmembrane Segment Zinc Binding in the Structure and Function of Rhodopsin*♦

Zinc deficiency and retinitis pigmentosa are both important factors resulting in retinal dysfunction and night blindness. In this study, we address the critical biochemical and structural relevance of zinc ions in rhodopsin and examine whether zinc deficiency can lead to rhodopsin dysfunction. We report the identification of a high-affinity zinc coordination site within the transmembrane domain of rhodopsin, coordinated by the side chains of two highly conserved residues, Glu122 in transmembrane helix III and His211 in transmembrane helix V. We also demonstrate that this zinc coordination is critical for rhodopsin folding, 11-cis-retinal binding, and the stability of the chromophore-receptor interaction, defects of which are observed in retinitis pigmentosa. Furthermore, a cluster of retinitis pigmentosa mutations is localized within and around this zinc binding site. Based on these studies, we believe that improvement in zinc binding to rhodopsin at this site within the transmembrane domain may be a pharmacological approach for the treatment of select retinitis pigmentosa mutations. Transmembrane coordination of zinc may also be an important common principle across G-protein-coupled receptors.

Zinc deficiency and retinitis pigmentosa are both important factors resulting in retinal dysfunction and night blindness. In this study, we address the critical biochemical and structural relevance of zinc ions in rhodopsin and examine whether zinc deficiency can lead to rhodopsin dysfunction. We report the identification of a high-affinity zinc coordination site within the transmembrane domain of rhodopsin, coordinated by the side chains of two highly conserved residues, Glu 122 in transmembrane helix III and His 211 in transmembrane helix V. We also demonstrate that this zinc coordination is critical for rhodopsin folding, 11-cis-retinal binding, and the stability of the chromophore-receptor interaction, defects of which are observed in retinitis pigmentosa. Furthermore, a cluster of retinitis pigmentosa mutations is localized within and around this zinc binding site. Based on these studies, we believe that improvement in zinc binding to rhodopsin at this site within the transmembrane domain may be a pharmacological approach for the treatment of select retinitis pigmentosa mutations. Transmembrane coordination of zinc may also be an important common principle across G-protein-coupled receptors.
Several neurodegenerative disorders, Alzheimer disease, Parkinson disease, familial amyotrophic lateral sclerosis, and the transmissible spongiform encephalopathies, share a common pathogenesis involving the misfolding and aggregation of specific proteins. Mounting evidence has demonstrated the direct binding of zinc (Zn 2ϩ ) to the ␤-amyloid (Alzheimer disease), ␣-synuclein (Parkinson disease), superoxide dismutase (amyotrophic lateral sclerosis), and prion (transmissible spongiform encephalopathies) proteins, linking either the gain or loss of Zn 2ϩ binding to the progression of these severe protein misfolding disorders. Zn 2ϩ promotes aggregation of the highly fibrillogenic prion peptide, PrP106 -126 (1), and of the ␤-amyloid protein (2)(3)(4) into amyloidogenic aggregates. Clioquinol, a metal chelating agent, is currently being investigated as a potential therapeutic solution to inhibit ␤-amyloid neurotoxic-ity (5). In contrast, in mutant superoxide dismutase protein, loss of affinity for Zn 2ϩ results in diminished protein activity (6), reduced protein stability (7), and formation of amyloid-like filaments (8). Thus, both zinc deficiencies and excesses can lead to neurodegenerative diseases.
In the G protein-coupled photoreceptor rhodopsin, there have been over 100 distinct, heritable mutations identified, many of which induce an altered protein conformation, misfolding, aggregation, and cell death, followed by the clinical manifestation of the retinal neurodegenerative disorder retinitis pigmentosa (9). Interestingly, Zn 2ϩ deficiency is also known to cause retinal neurodegeneration and night blindness (10), symptoms reminiscent of retinitis pigmentosa. Furthermore, Zn 2ϩ has been shown to directly bind rhodopsin (11,12) and to reduce rhodopsin thermal stability and regeneration with 11cis-retinal at higher Zn 2ϩ concentrations (50 -200 M) (13). While Zn 2ϩ is a well known structural and catalytic cofactor for a number of metalloenzymes and transcription factors, recent studies have identified Zn 2ϩ as an allosteric modulator of structure and function for a number of G protein-coupled receptors, including the dopamine (14), adrenergic (15,16), melanocortin (17,18), and chemokine (19) receptors. Although a number of these studies have described the effect of engineered Zn 2ϩ coordination sites in the transmembrane domain of rhodopsin and other G protein-coupled receptors (20,21), and sites within intracellular and extracellular loops or at helical ends (14 -19), our goal was to confirm and analyze the role of a native Zn 2ϩ coordination site formed by residues within the transmembrane (TM) 1 domain, the first to be described among G proteincoupled receptors.
Upon analysis of the 2.8-Å resolution crystal structure of rhodopsin (22)(23)(24), we identified three putative zinc (Zn 2ϩ ) ions bound to each rhodopsin monomer (Zn 2ϩ ions 1, 2, and 3 on rhodopsin monomer A; Zn 2ϩ ions 4, 5, and 6 on rhodopsin monomer B) with a seventh Zn 2ϩ ion present at the interface between the two monomers of the rhodopsin dimer. We now report that the Zn 2ϩ coordination site formed by residues within the rhodopsin transmembrane domain is physiologically relevant and critical to the structure and function of rhodopsin. Furthermore, we have confirmed that this site is specific for Zn 2ϩ versus other similar divalent metals. This Zn 2ϩ coordination site in rhodopsin lies within the 11-cis-retinal binding pocket (in the TM domain), 8 Å from the ␤-ionone ring of 11-cis-retinal, coordinated by the transmembrane amino acids Glu 122 (TM III; side chain carbonyl), Met 163 (TM IV; backbone carbonyl), and His 211 (TM V; imidazole side chain). Through inductively coupled plasma mass spectrometry (ICP-MS), flu-orescent detection of Zn 2ϩ and site-directed mutagenesis, we confirm that Zn 2ϩ is bound to rhodopsin and that the coordination site formed by residues Glu 122 and His 211 is a highaffinity zinc coordination site. In contrast to a low-affinity, solvent-accessible Zn 2ϩ coordination site that destabilizes rhodopsin, the coordination of Zn 2ϩ within the transmembrane domain is critical for dark state rhodopsin stability, 11-cisretinal binding, and may play a vital role in transition to the active metarhodopsin II state.

EXPERIMENTAL PROCEDURES
Preparation of Rhodopsin Protein Samples-Amino acid substitutions were introduced into the synthetic rhodopsin gene through PCR mutagenesis, as described previously (25). COS-1 cells were transiently transfected with pMT4 vectors carrying the opsin genes. Cells were harvested and opsins regenerated with 11-cis-retinal and rhodopsin proteins purified by immunoaffinity chromatography. In experiments where rhodopsin was treated with EDTA (Sigma), CaCl 2 (Sigma), CoCl 2 (Sigma), CuCl 2 (Sigma), or ZnCl 2 (Fisher Scientific) or examined under various pH values, the respective phosphate-buffered saline buffer, with those changes, was prepared and utilized throughout the protein purification process. ROS (rod outer segments) from bovine retinas was a kind gift from Dr. Phillip Reeves (M.I.T.). The presence of rhodopsin protein, both from COS-1 cells and from ROS, was determined through UV-visible spectroscopy (500 nm absorbance) and Western blot analysis (25).
UV-visible Absorption Spectroscopy-UV-visible absorption spectra were recorded on a PerkinElmer Life Sciences -40 UV-visible spectrophotometer at 2 or 25°C (25). Rhodopsin thermal stability was determined by monitoring the decay of the 500 nm absorbance at a constant temperature of 50°C. Protein samples were allowed to equilibrate (2-5 min) to 50°C prior to acquiring absorption spectra. Absorption spectra were obtained at 5-min intervals, until complete disappearance of the 500 nm absorbance. Initial values of the 500 nm absorbance were normalized between experiments to account for base-line shifts and protein concentration.
Fluorescent Zn 2ϩ Detection-Purified rhodopsin protein or rhodopsin from ROS was incubated with a 1 M concentration of the fluorescent Zn 2ϩ indicator, FluoZin TM -3 (Molecular Probes). Samples were excited in a CytoFluor II spectrofluorimeter (Applied Biosystems) at a wavelength of 485 nm (bandwidth of 20 nm), and the emission was measured at a wavelength of 508 nm (bandwidth of 20 nm). This was within the active range of FluoZin-3 (optimal excitation at 494 nm and emission at 515 nm). A standard curve of Zn 2ϩ binding to FluoZin-3 was determined by the addition of 10-fold differing concentrations of Zn 2ϩ to FluoZin-3 (six Zn 2ϩ concentrations from 1.0 ϫ 10 Ϫ10 M to 1.0 ϫ 10 Ϫ5 M; n ϭ 6). To exclude fluorescence due to binding of alternate cations, we measured fluorescence of FluoZin-3 upon the addition of Ca 2ϩ (CaCl 2 ). To determine background Zn 2ϩ , the same concentration of FluoZin-3 was incubated with each solution in which rhodopsin was solubilized and purified. Rhodopsin was incubated with FluoZin-3 and assayed for fluorescence in either the dark or activated state. Three fluorescence scans were obtained and averaged for each cycle. For photoactivation, protein samples were illuminated for 30 s with a 150-watt fiber optic light. These samples were then returned to the dark for the remainder of the experiment.
ICP-MS-Purified rhodopsin protein was dissolved in a HNO 3 solution and digested in an MARS-5 microwave (300 watts for 10 min). ICP-MS analyses were performed on an Agilent 7500s ICP-MS equipped with an Agilent High Solids nebulizer. The solution was diluted 4-and 10-fold with deionized water and analyzed for zinc. Three isotopes of zinc ( 66 Zn, 67 Zn, and 68 Zn) were averaged for measurements and calculations. To check for contamination of the digestion procedure and sample manipulation, a blank solution was prepared and carried through each set of analyses. Gallium was used as internal standard for the determination of zinc.
Computer Modeling-The rhodopsin crystal structure (PDB ID: 1L9H) (24) was visualized with the Swiss PDB Viewer computer program (GlaxoSmithKline, Geneva, Switzerland) (26). Putative H-bonds were depicted in green. Zn 2ϩ ions were depicted as gray spheres.
Data Analysis-Values for all experiments are expressed as mean Ϯ S.E. All means and S.E. were calculated from at least three separate experiments using separate protein preparations. K rate constants, half-lives, and FluoZin-3 Zn 2ϩ binding rates were acquired with nonlinear regression using GraphPad Prism® software. A 95% confidence interval was used for all curve-fitting procedures using GraphPad Prism. Where applicable, data were determined to be statistically significant using the unpaired t test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).

RESULTS
Through examination of the rhodopsin crystal structure (PDB ID: 1L9H) (24), we identified four putative Zn 2ϩ coordination sites in each monomer of rhodopsin (Fig. 1A, Zn 2ϩ 1 , Zn 2ϩ 2 , Zn 2ϩ 3 , and Zn 2ϩ 4 ). Interestingly, while many metal binding sites occur in solvent-exposed regions, there was a predicted Zn 2ϩ 3 coordination site within the TM 11-cis-retinal binding pocket of rhodopsin, 8 Å from the ␤-ionone ring of 11-cis-retinal (Fig. 1B). Furthermore, we had previously identified a large clustering of retinitis pigmentosa mutations in this same region (9). As the rhodopsin crystallization procedure required supraphysiological concentrations of Zn 2ϩ (ϳ100 mM), it was crucial to determine which Zn 2ϩ coordination sites were physiologically and biochemically relevant and to determine the role played by the coordination of such Zn 2ϩ ions. Determination of Zn 2ϩ binding (ICP-MS and FluoZin-3) was followed by Zn 2ϩ treatment (pre-and postprotein purification), biochemical removal of the solvent exposed Zn 2ϩ binding sites (pH and EDTA), and mutagenic removal of the Zn 2ϩ from the transmembrane site. These assays in combination with the rhodopsin crystal structure confirmed a structurally important, buried, high-affinity Zn 2ϩ coordination site with the rhodopsin transmembrane domain and one or more low-affinity, solventexposed extracellular Zn 2ϩ coordination sites.

Rhodopsin Binds One High-affinity and
One Low-affinity Zn 2ϩ Using ICP-MS and fluorescent chelation of Zn 2ϩ , we confirmed that each rhodopsin molecule possessed at least one Zn 2ϩ binding site. ICP-MS detected a total binding of 1.33 Ϯ 0.27 Zn 2ϩ ions per molecule of rhodopsin ( Fig. 2A). We hypothesized complete occupancy of a high-affinity site (TM) and partial occupancy of low-affinity sites (extracellular). The Zn 2ϩdependent fluorophore, FluoZin-3, was used to quantitate lowaffinity Zn 2ϩ coordination in rhodopsin. FluoZin-3, a Zn 2ϩ chelator, removes Zn 2ϩ that are loosely bound to rhodopsin or free in solution. In Fig. 2B, we show a standard curve of Zn 2ϩ binding to FluoZin-3. 10-Fold differences in Zn 2ϩ concentrations were initially used (0.1 nM to 1 mM), followed by more detailed assessment of the linear portion of the curve to correlate fluorescence (508 nm) to Zn 2ϩ concentration (inset, Fig. 2B). The calculated linear regression was determined to be Y (A.U.) ϭ 428.7 X (log M) ϩ 3150. This binding was shown to be specific for Zn 2ϩ , as incubation of FluoZin-3 with Ca 2ϩ showed no significant fluorescence at equivalent concentrations of metal (Fig. 2B).
Wild type rhodopsin was incubated with FluoZin-3 and as-sayed for fluorescence in the dark or upon photoactivation. In the absence of added Zn 2ϩ , FluoZin-3 chelated 0.21 Ϯ 0.05 Zn 2ϩ in the dark state and 0.57 Ϯ 0.06 (p Ͻ 0.001) Zn 2ϩ per rhodopsin in the photoactivated state ( Fig. 2A). These data supported our hypothesis that one solvent exposed Zn 2ϩ coordination site was partially bound and solvent exposed (easily chelatable by FluoZin-3), while a second, higher affinity Zn 2ϩ coordination site (hypothesized to be the TM site) required photoactivation for chelation. We next examined Zn 2ϩ affinity in a His 211 3 Cys mutant, which was predicted to reduce the affinity of Zn 2ϩ at the Zn 2ϩ 3 site natively coordinated by His 211 and Glu 122 . FluoZin-3 chelated 0.40 Ϯ 0.03 Zn 2ϩ in the dark state, and the total zinc (1.37 Ϯ 0.12 Zn 2ϩ (p Ͻ 0.001)) was readily detected in the photoactivated state ( Fig. 2A). Data from the wild type rhodopsin proteins supported that the Zn 2ϩ 3 coordination site within the transmembrane domain was a physiologically relevant and high-affinity site; however, it became an exposed, lower affinity site upon photoactivation of rhodopsin. The His 211 3 Cys mutation reduced the affinity for Zn 2ϩ within the transmembrane domain, allowing rapid total chelation upon photoactivation.
To validate the physiological relevance of our findings, rhodopsin from ROS was analyzed using FluoZin-3. FluoZin-3 chelated 0.18 Ϯ 0.01 Zn 2ϩ in the dark state and 0.20 Ϯ 0.01 Zn 2ϩ per rhodopsin in the photoactivated state. While these data were not found to be statistically significant, a small light-sensitive trend similar to that observed with rhodopsin purified from COS-1 cells was consistently observed with rhodopsin from ROS. Nevertheless, these studies were able to confirm that the low-affinity portion of Zn 2ϩ bound to rhodopsin was less than half a Zn 2ϩ per rhodopsin molecule in both COS-1-expressed and ROS rhodopsin. As a result of the much higher stability of rhodopsin in ROS versus the dodecyl maltoside system used for rhodopsin expressed in COS-1 cells, the Zn 2ϩ ion in the TM domain may be less accessible and detectable by FluoZin-3 upon light activation.

Treatment of Rhodopsin with ZnCl 2 Reveals One Stabilizing and One Destabilizing Zn 2ϩ Coordination Site
Previous studies have used high concentrations of Zn 2ϩ after rhodopsin purification thus accessing only the solvent accessible site. The TM domain is buried in a highly hydrophobic region and only exposed either during protein folding or upon photoactivation, as observed with FluoZin-3 chelatability. To further support our hypothesis that the Zn 2ϩ 3 site and coordination of Zn 2ϩ within the transmembrane domain is physiologically relevant, we both pretreated cells immediately following transfection (to promote access to the Zn 2ϩ 3 coordination site during protein folding and processing) in addition to the conventionally used posttreatment procedure (after rhodopsin purification). Control samples were left untreated. UV-visible absorption spectra, obtained from the three rhodopsin populations (untreated, pretreated (50 M ZnCl 2 added to COS-1 cells), or posttreated (50 M ZnCl 2 added to purified protein)) show identical formation of a 500 nm peak (not shown). Upon photoactivation, all three populations also show identical formation of a 380 nm peak (not shown). Immunoblotting of rhodopsin showed that, at concentrations as high as 50 M ZnCl 2 in cell culture, no change in the expression or glycosylation of rhodopsin protein was observed (not shown).
Thermal stability of the rhodopsin dark state at 50°C was measured as a rate of decay of the 500 nm absorbance (i.e. release of 11-cis-retinal) over time. The wild type rhodopsin dark state decayed with a k rate constant of 0.024 Ϯ 0.001 min Ϫ1 (Fig. 3A and Table I). Pretreatment of COS-1 cells with 10 M ZnCl 2 consistently and significantly slowed the rate of rhodopsin decay to 0.021 Ϯ 0.001 min Ϫ1 (p Ͻ 0.05), while pretreatment with 50 M ZnCl 2 accelerated the decay rate to 0.030 Ϯ 0.001 min Ϫ1 (p Ͻ 0.001; Fig. 3A and Table I). Posttreatment of rhodopsin protein with 10 M ZnCl 2 increased the decay rate to 0.027 Ϯ 0.002 min Ϫ1 , although t test analysis confirmed these data as not statistically significant. Posttreatment of rhodopsin protein with 50 M ZnCl 2 markedly and significantly increased the decay rate to 0.052 Ϯ 0.006 min Ϫ1 (p Ͻ 0.01; Fig. 3A and Table I).
To evaluate the selectivity of the predicted Zn 2ϩ coordination site in the rhodopsin transmembrane domain, COS-1 cells were transiently transfected with wild type rhodopsin, pretreated with 10 M CaCl 2 , CoCl 2 , CuCl 2 , or ZnCl 2 , and 11-cis-retinal binding and rhodopsin thermal stability evaluated. Similar to ZnCl 2 , pretreatment with CaCl 2 , CoCl 2 , CuCl 2 did not alter the 11-cis-retinal binding affinity of rhodopsin (data not shown). While ZnCl 2 pretreatment and coordination of Zn 2ϩ in the TM domain improved rhodopsin thermal stability (Table I and Fig.   3E), none of the additional metals showed such an effect. These studies confirm that the metal coordination site in the TM domain of rhodopsin is selective for Zn 2ϩ . Interestingly, 10 M CuCl 2 pretreatment led to significant destabilization. This may be related to extracellular solvent-exposed binding.
These data suggested the presence of a stabilizing Zn 2ϩselective coordination site whose occupancy could be detected with a 10 M ZnCl 2 pretreatment and a destabilizing Zn 2ϩ coordination site whose occupancy could be detected with a 50 M ZnCl 2 posttreatment. These findings also indicated that only ZnCl 2 pretreatment, and not posttreatment of purified protein, allowed Zn 2ϩ coordination at the stabilizing site. As we initially hypothesized, only the highly hydrophobic Zn 2ϩ 3 coordination site (Fig. 1A) would be inaccessible to ZnCl 2 posttreatment of purified rhodopsin. Altogether, our data suggested that the Zn 2ϩ 3 coordination site within the transmembrane domain was a higher affinity, protein-stabilizing site, while one of the solvent-exposed (Fig. 1A, Zn 2ϩ 1 , Zn 2ϩ 2 , and Zn 2ϩ 4 ) coordination sites was a lower affinity, protein-destabilizing site. To further support our hypothesis we used biochemical and mutagenic techniques to remove these coordination sites.

Removal of Zn 2ϩ from a Solvent-exposed Zn 2ϩ Coordination Site Stabilizes the Rhodopsin Dark State
To support the presence of a low-affinity, solvent-accessible, destabilizing site, we treated purified rhodopsin with EDTA, a strong Zn 2ϩ chelator. We detected no fluorescence of FluoZin-3 in rhodopsin samples treated with 1 mM EDTA (pH 7.4), suggesting that EDTA chelated all Zn 2ϩ that were solvent-accessible (not shown). Removal of Zn 2ϩ from the solvent-accessible Zn 2ϩ coordination site stabilized rhodopsin at 50°C, as treatment with EDTA increased the thermal stability t1 ⁄2 of rhodopsin to greater than 60 min ( Fig. 3B and Table I). While untreated rhodopsin began to immediately decay at 50°C, EDTAtreated rhodopsin did not decay for the first 45 min. The protein then began to degrade at a rate constant of 0.025 Ϯ 0.006 min Ϫ1 , similar to untreated rhodopsin ( Fig. 3B and Table  I). As EDTA chelates other divalent cations we also examined the use of pH.
The solvent-exposed Zn 2ϩ coordination sites, Zn 2ϩ 1 and Zn 2ϩ 4 , each has a histidine residue as a critical coordinating ligand (Zn 2ϩ 2 does not). Since protonated histidine has a much lower affinity for Zn 2ϩ , we altered solvent pH levels near the pK a of histidine (pH 6.5) to determine whether coordination of Zn 2ϩ occurred at one of these two sites. When we purified rhodopsin at pH 5.5 and 6.0 (equilibrium shifted toward protonated histidine), rhodopsin exhibited no detectable decay in the dark state for a period of at least 120 min ( Fig. 3C and Table I). Purification of rhodopsin at pH 7.4, 8.0, and 9.0 (equilibrium shifted toward unprotonated histidine) destabilized rhodopsin in a pH-dependent manner, increasing the decay rates to 0.024 Ϯ 0.001 min Ϫ1 (p Ͻ 0.001), 0.040 Ϯ 0.001 min Ϫ1 (p Ͻ 0.001), and 0.049 Ϯ 0.002 min Ϫ1 (p Ͻ 0.001), respectively ( Fig. 3C and Table I). In Fig. 3D, a plot of thermal decay t1 ⁄2 versus solvent pH displays a half-point at pH 6.7 Ϯ 0.3, similar to the pK a of the histidine side chain (pH 6.5). These data suggest that rhodopsin thermal stability is directly dependent on the state of histidine protonation. These data further support our hypothesis that coordination of Zn 2ϩ at the solvent-exposed Zn 2ϩ 1 or Zn 2ϩ 4 coordination site destabilizes the rhodopsin dark state. Furthermore, it also indicated that coordination of Zn 2ϩ at the Zn 2ϩ 2 coordination site was most likely not physiologically relevant. This ascertainment is supported by the fact that Zn 2ϩ 2 has no histidine or cysteine residues, two amino acids commonly found to compose Zn 2ϩ coordination sites.

Mutation of the Zn 2ϩ 3 Coordination Site Transmembrane Domain Residues His 211 , Glu 122 , and Trp 126 Alters Spectral Properties, 11-cis-Retinal Binding, and Rhodopsin Thermal Stability
Mutagenesis studies were performed on the highly conserved residues that were predicted to form and stabilize the putative Zn 2ϩ coordination site within the transmembrane domain. Characterization of such mutations, utilizing a variety of experimental techniques, would provide critical information on the contribution of each residue to Zn 2ϩ coordination.
His 211 -A crystal structure-based model (Fig. 1B) suggested that His 211 was the primary Zn 2ϩ 3 coordinating residue. More interestingly, FluoZin-3 data of the His 211 3 Cys mutant indicated the important role of this residue to Zn 2ϩ 3 affinity. Sitedirected mutagenesis to His 211 3 Cys (a residue still able to coordinate Zn 2ϩ ) and His 211 3 Phe (a residue unable to coor-dinate Zn 2ϩ ) was performed. His 211 3 Cys and His 211 3 Phe mutants bind 11-cis-retinal with an absorbance ratio lower than wild type, at both 2 and 25°C ( Fig. 4A and Table II). Fig.  4A also shows a shift in the max from the wild type 500 to 493 nm, for both His 211 3 Cys and His 211 3 Phe. Such data suggest the destabilization and altered structure of the 11-cis-retinal binding pocket. Interestingly, the decrease in 11-cis-retinal binding was much more evident at the higher temperature (25°C), indicating a potential role of this site in thermal stability of the native protein. Mutation to His 211 3 Cys and His 211 3 Phe also increased the thermal decay k rate constant from the wild type 0.024 Ϯ 0.001 min Ϫ1 to 0.157 Ϯ 0.009 min Ϫ1 (p Ͻ 0.01; Fig. 4B and Table I) and 0.201 Ϯ 0.010 min Ϫ1 (p Ͻ 0.01; data not shown), respectively.
We hypothesized that each mutant had lowered the affinity for Zn 2ϩ 3 and that ZnCl 2 pretreatment of COS-1 cells, express- as a function of buffer pH. GraphPad Prism software was used to fit the curve and to identify the half-way point. All experiments were performed at least n ϭ 3 times, using separate protein preparations. E, to determine selectivity of the high-affinity Zn 2ϩ binding site, pretreatment of wild type rhodopsin with 10 M ZnCl 2 , CaCl 2 , CoCl 2 , and CuCl 2 were performed and thermal decay rate constants (k) determined (*, p Ͻ 0.05; **, p Ͻ 0.01) on the purified protein.
ing these mutant rhodopsin genes, might improve coordination of Zn 2ϩ at the Zn 2ϩ 3 site. His 211 3 Cys showed some improvement in thermal stability, with a 10 M ZnCl 2 pretreatment; however, t test analysis was not statistically significant ( Fig.  4B and Table I). His 211 3 Phe showed no improvement in thermal stability with either 10 or 50 M ZnCl 2 pretreatment (not shown). Posttreatment of purified protein with either 10 or 50 M ZnCl 2 did not improve the thermal stability of either His 211 3 Cys (Table I)  Glu 122 -Glu 122 3 Cys (a residue still able to coordinate Zn 2ϩ ) and Glu 122 3 Leu (a residue unable to coordinate Zn 2ϩ ) bind 11-cis-retinal with an absorbance ratio lower than wild type, at both 2 and 25°C ( Fig. 4C and Table II). In addition, there was a shift in the max from the wild type 500 to 482 nm (Glu 122 3 Cys) and 492 nm (Glu 122 3 Leu; Table II). As with His 211 mutants, the decrease in 11-cis-retinal binding was much more evident at the higher temperature (25°C), indicating a potential role of Glu 122 in thermal stability of the native protein. Glu 122 3 Cys and Glu 122 3 Leu mutants also increased the thermal decay rate to 0.041 Ϯ 0.002 min Ϫ1 (p Ͻ 0.01; Fig. 4D and Table I) and 0.038 Ϯ 0.001 min Ϫ1 (p Ͻ 0.001; data not shown), respectively. Pretreatment of cells expressing the Glu 122 3 Cys mutant, with 10 M ZnCl 2 , improved the thermal stability of this rhodopsin mutant from a k rate constant of 0.041 Ϯ 0.002 min Ϫ1 to 0.029 Ϯ 0.003 min Ϫ1 (p Ͻ 0.05; Fig. 4D and Table I). While 10 M ZnCl 2 pretreatment improved protein stability, a pretreatment of 50 M ZnCl 2 showed no statistically significant effect (Table I). Such data indicate that, at 50 M ZnCl 2 there is binding of Zn 2ϩ at the lower affinity, destabilizing site as well. This hypothesis is supported by similar treatments in wild type rhodopsin, where a higher concentration of ZnCl 2 also bound to the solvent-accessible site. Such a hypothesis may also explain the absence of improved His 211 3 Cys protein stability with additional ZnCl 2. Pretreatment of COS-1 cells expressing the Glu 122 3 Leu mutant, with 10 or 50 M ZnCl 2 , showed no improvement in the decay rate of that rhodopsin mutant (data not shown). Posttreatment with ZnCl 2 did not improve the thermal stability of either Glu 122 3 Cys (Table I)  Trp 126 -Although tryptophan is unlikely to directly coordinate Zn 2ϩ , hydrophobic amino acids typically surround the metal binding site to facilitate metal coordination and to sta-bilize the site. The Trp 126 3 Asp mutant binds 11-cis-retinal with an absorbance ratio lower than wild type, at both 2 and 25°C ( Fig. 4E and Table II). In addition, there is a shift in the max from the wild type 500 to 484 nm ( Fig. 4E and Table II). Mutation of Trp 126 3 Asp also increased the rhodopsin decay rate to 0.070 Ϯ 0.007 min Ϫ1 (p Ͻ 0.05; Fig. 4F). Pretreatment of COS-1 cells expressing the Trp 126 3 Asp mutant, with 10 or 50 M ZnCl 2 , showed no improvement in the decay rate of that rhodopsin mutant (Fig. 4F). Posttreatment of Trp 126 3 Asp with either 10 or 50 M ZnCl 2 also did not improve the thermal stability (data not shown).

Modeling of the Zn 2ϩ Coordination Site in the Rhodopsin Transmembrane Domain
His 211 and Glu 122 are necessary for the full affinity of this Zn 2ϩ 3 coordination site. The thermal stability of His 211 3 Cys and Glu 122 3 Cys could be rescued by pretreating with 10 M Zn 2ϩ (although the results for the more severe His 211 3 Cys had not reached statistical significance). Combined, these findings supported our FluoZin-3 and ICP-MS data showing a decrease in affinity for Zn 2ϩ with His 211 3 Cys. Molecular modeling of the Zn 2ϩ 3 coordination site (PDB ID: 1L9H) showed the structural basis for our results (Fig. 5). Computations performed by the Swiss PDB Viewer computer program proposed that mutation of Glu 122 3 Cys would result in a Cys 122 conformation that could not interact with Zn 2ϩ 3 (Fig.  5B). Such a conformation would account for the reduced Zn 2ϩ affinity and reduced thermal stability that we observed with this mutant. However, an alternate conformation of Cys 122 could coordinate Zn 2ϩ more tightly (Fig. 5C). Our data suggest that, through the addition of exogenous Zn 2ϩ , we favored the formation of this alternate and more stable Cys 122 conformation (Fig. 5C). Improvement in Zn 2ϩ 3 coordination correlated well with an increased thermal stability for this mutant. Pretreatment of Glu 122 3 Leu did not show an improvement in thermal stability, as observed with the Glu 122 3 Cys mutant. This was expected, since no conformation of leucine could participate in improved Zn 2ϩ coordination. DISCUSSION Zinc deficiency leads to night blindness, as do retinitis pigmentosa mutations. The role of Zn 2ϩ ions in the structure and TABLE I Thermal stability calculations for rhodopsin and rhodopsin mutant proteins Statistical significance for treatments with zinc, EDTA, and altered pH is compared with non-treated sample of same rhodopsin population (i.e. wild type versus wild type ϩ 1 mM EDTA). Statistical significance (for mutants at pH 7.4) is compared with wild-type rhodopsin at pH 7.4 (i.e. wild type versus Glu 122 3 Cys). ND, not detectable. ns, not statistically significant. function of rhodopsin is unclear with most studies, including the crystallization of rhodopsin utilizing supraphysiological doses. As we observed in this study, receptor destabilization through the addition of high Zn 2ϩ concentrations arises from binding to low-affinity, solvent-accessible, and perhaps nonphysiologically relevant sites. More importantly, we now confirm that a Zn 2ϩ coordination site in the transmembrane domain is selective for Zn 2ϩ and plays a critical physiological role in rhodopsin stabilization in the dark. We also hypothesize that this site may also serve a role in Meta I-Meta II active state transition. Based upon these findings, caution is required in interpreting studies using engineered Zn 2ϩ coordination sites (and supraphysiological Zn 2ϩ concentrations) as significant destabilizing effects may result from binding to native Zn 2ϩ sites.

Glu 122 and His 211 Form a Zn 2ϩ 3 Coordination Site within the Transmembrane Domain-Formation of the Zn 2ϩ
3 coordination site is critical for rhodopsin dark state stability. Loss of Glu 122 , Trp 126 , or His 211 lowers the affinity of rhodopsin for Zn 2ϩ 3 , resulting in a destabilized 11-cis-retinal binding pocket, altered spectral properties of the chromophore, and diminished retinal binding (Fig. 4 and Table II) structure (PDB ID: 1L9H) identifies the nearest group with which Glu 122 could hydrogen-bond, as the His 211 imidazole nitrogen, 4.26 Å away (24), suggesting a water-mediated hydrogen bond or metal coordination rather than a direct hydrogen bond. Mutation of His 211 3 Asn and His 211 3 Phe caused large aberrations in the Fourier transform infrared spectral bands assigned to the Glu 122 carboxyl side chain, as well as in bands that could not be attributed to the chromophore, His 211 , or Glu 122 , but rather to a "group X" (28). Additionally, the carbonyl group of Glu 122 interacted directly with this group X. These studies support our model that Glu 122 and His 211 form the Zn 2ϩ 3 (group X) coordination site in rhodopsin (Fig. 1B). The importance of this coordination is further highlighted by the high degree of conservation. Glu 122 , Trp 126 , and His 211 are 80, 86, and 100% conserved, respectively, in vertebrate rhodopsin pigments. Interestingly, upon analysis of 227 PDB files displaying a Zn 2ϩ coordination site, we found that a Zn 2ϩ coordination site from the ␣-toxin of Clostridium perfringens (PDB ID: 1CA1) displayed a resemblance to the geometry assumed by the Zn 2ϩ 3 ion (His side chain-Glu side chain-backbone carbonyl). Such an observation suggests that the coordinated Zn 2ϩ 3 ion may play an important and novel catalytic role in rhodopsin.
Rhodopsin Loses Affinity for the Zn 2ϩ Ion within the Transmembrane Domain in Meta II, Suggesting a Role in the Meta I-Meta II Transition-Past studies have concluded that more Zn 2ϩ bind rhodopsin upon light activation, due to an increase in detected levels of Zn 2ϩ (11,12). An early study using atomic absorption spectrophotometry and chromatography showed that the stoichiometry of Zn 2ϩ to rhodopsin from ROS was 0.28 in the dark state and 0.45 in the photoactivated state (12). Furthermore, Fourier transform infrared bands attributed to group X (discussed above) disappear in Meta II (28). We also report that the high-affinity transmembrane domain Zn 2ϩ binding site is not detectable in the dark state, becoming exposed and of lower affinity upon light activation. As a result, we conclude that Zn 2ϩ 3 is an important transmembrane structural cofactor for the rhodopsin ground state that may be released and detectable only upon photoactivation.
It has been suggested that repositioning of the ␤-ionone ring toward TM IV ultimately determines which activated state is assumed (27,29,30). Thus, structural changes in the region near Glu 122 and His 211 are required for proper repositioning of the ␤-ionone ring (27). Previous studies have show a direct interaction between 11-cis-retinal and Gly 121 in rhodopsin TM III (31)(32)(33)(34)(35). The increase in the size of Gly 121 correlated directly with the increasing dark state activity. Additionally, a Gly 121 3 Leu mutant was able to activate rhodopsin in the presence of ␤-ionone alone, with no Schiff base formation. Based on our studies, upon photoactivation Gly 121 is displaced along with its adjacent Glu 122 . We propose that such repositioning of Glu 122 (3 Å away from the ␤-ionone ring), would result in the association of the Glu 122 carboxyl with Zn 2ϩ 3 (Fig.  5D). In support of this model, during Meta I, group X (Zn 2ϩ 3 ) undergoes a change in hydrogen bonding with an OH-bearing residue (28). The nearest -OH group for Zn 2ϩ 3 to associate with is the Glu 122 carboxyl (4.28 Å). Zn 2ϩ has no energetic barrier to multiple coordination geometries and can therefore assume four-or five-membered coordination with relative ease (36). In metalloenzymes, the coordination number of Zn 2ϩ changes to stabilize high-energy intermediates and other transition states (37). An altered interaction between Glu 122 and Zn 2ϩ 3 could modify the local pH, reduce the charge on Zn 2ϩ 3 , and promote the Zn 2ϩ 3 -catalyzed protonation of a nearby residue. Interestingly, protonation of His 211 is coupled to the formation of Meta II (38). The midpoint of the Meta I-Meta II transition (pH 6.4) occurs near the pK a of the histidine side chain (pH 6.5) (39). As a result, we propose that protonated His 211 would lose the interaction with Zn 2ϩ 3 (Fig. 5E). Our data suggest that, in the ground (dark) state, the Zn 2ϩ 3 coordination site in the rhodopsin transmembrane domain is a stabilizing mechanism, retaining rhodopsin in the inactive conformation and 11-cis-retinal as an inverse agonist. Upon photoactivation, this Zn 2ϩ coordination is disrupted, promoting the dissociation of TM helix III from helices IV and V and allowing the unhindered formation of photoactivated intermediates. It has been hypothesized that protonated His 211 interacted electrostatically with another amino acid to drive rearrangement of TM helices (38); our studies suggest that protonated His 211 loses an interaction (with Zn 2ϩ 3 ) to drive this helical rearrangement. Interestingly, His 211 is 100% conserved in vertebrate rhodopsin pigments, and Cys 211 is 100% conserved in vertebrate cone pigments. The conserved nature of this site raises the possibility that the Zn 2ϩ 3 -mediated Meta II formation may be an acquired feature that accounts for the 100-fold difference in light sensitivity between rod and cone pigments (40). The fact that Zn 2ϩ deficiency results in abnormal dark adaptation and night blindness in humans and other animals (41) further supports the critical role of this site in rhodopsin structure and function.
Night Blindness and Protein Misfolding in Retinitis Pigmentosa-Naturally occurring mutations His 211 3 Pro and His 211 3 Arg give rise to severe opsin misfolding, leading to retinitis pigmentosa, a serious retinal degenerative disorder (reviewed in Refs. 9, 42, and 43). Interestingly one of the first manifestations of this disorder is night blindness. Moreover, a large cluster of retinitis pigmentosa mutations are located in the region near His 211 . We have recently shown that two of these mutations, Leu 125 3 Arg and Ala 164 3 Val, directly interfere with a Glu 122 -Trp 126 -His 211 interaction (25). We now propose that the disrupted Zn 2ϩ 3 coordination site in the rhodopsin transmembrane domain directly leads to the protein misfolding and night blindness associated with these mutations. A number of studies have shown that folding of the transmembrane domain is coupled to the formation of a specific tertiary structure and the native Cys 110 -Cys 187 disulfide bond in the intradiscal domain (42, 44 -46). Since loss of the Zn 2ϩ 3 coordination site and loss of the Cys 110 -Cys 187 disulfide bond promote rhodopsin unfolding, it is exciting to speculate that formation of the Zn 2ϩ 3 site could be coupled to the formation of this critical disulfide bond. Combined, our studies strongly support the importance of zinc in rhodopsin structure and function and also provide a possible molecular mechanism for night blindness and retinitis pigmentosa. Additionally, improvement in protein stability of the Glu 122 3 Cys mutant with the addition of Zn 2ϩ further highlights the importance of this site and suggests a Potential Mechanism for Neurodegeneration from Zinc Excesses and Deficiencies-Zn 2ϩ promotes aggregation of the highly fibrillogenic prion peptide, PrP106 -126 (1), and of the ␤-amyloid protein (2-4) into amyloidogenic aggregates, leading to cellular apoptosis and neurodegeneration. In contrast, in the mutant superoxide dismutase protein, loss of affinity for Zn 2ϩ results in diminished protein activity (6), reduced protein stability (7), and formation of amyloid-like filaments (8). The mechanisms for these observations are not well understood, particularly as both the loss and gain of Zn 2ϩ binding can induce neurodegeneration. We observe that both a loss of physiological Zn 2ϩ binding and a gain of non-physiological Zn 2ϩ binding can induce rhodopsin destabilization and misfolding through distinct mechanisms. Loss of a native high-affinity Zn 2ϩ coordination site, either through Zn 2ϩ deficiencies or induced by naturally occurring mutations, renders the rhodopsin protein unstable, misfolded, and eventually aggregated. Conversely, supraphysiological Zn 2ϩ concentrations binding to a non-native, low-affinity Zn 2ϩ coordination site in rhodopsin can also result in biochemical defects.
Selectivity of trace metals in neurodegeneration has recently gained increasing interest. Our studies confirm that the highaffinity metal coordination site in the TM domain is selective for Zn 2ϩ . Interestingly, at the solvent-accessible, low-affinity metal coordination site, recent studies suggest that high concentrations of either Zn 2ϩ or Cu 2ϩ can destabilize rhodopsin protein (13). This is similar to what has been observed with other proteins, where high-affinity and physiological metal coordination sites are selective, while low-affinity and non-physiological (or pathophysiological) metal coordination sites may be non-selective. High concentrations of Cu 2ϩ , similar to Zn 2ϩ , also promote aggregation of the highly fibrillogenic prion peptide, PrP106 -126 (1), and of the ␤-amyloid protein (2-4) into amyloidogenic aggregates. Our current study may help clarify the mechanisms by which a loss and/or gain of metal binding results in a wide variety of neurodegenerative diseases.