The Role of the Native Lipids and Lattice Structure in Bacteriorhodopsin Protein Conformation and Stability as Studied by Temperature-dependent Fourier Transform-Infrared Spectroscopy*

We report the effect of partial delipidation and monomerization on the protein conformational changes of bacteriorhodopsin (bR) as a function of temperature. Removal of up to 75% of the lipids is known to have the lattice structure of the purple membrane, albeit as a smaller unit cell, whereas treatment by Triton monomerizes bR into micelles. The effects of these modifications on the protein secondary structure is analyzed by monitoring the protein amide I and amide II bands in the Fourier transform-infrared (FT-IR) spectra. It is found that removal of the first 75% of the lipids has only a slight effect on the secondary structure at physiological temperature, whereas monomerizing bR into micelles alters the secondary structure considerably. Upon heating, the bR monomer is found to have a very low thermal stability compared with the native bR with its melting point reduced from 97 to 65 °C, and the pre-melting transition in which the protein changes conformation in native bR at 80 °C could not be observed. Also, the N–H to N–D exchange of the amide II band is effectively complete at room temperature, suggesting that there are no hydrophobic regions that are protected from the aqueous medium, possibly explaining the low thermal stability of the monomer. On the other hand, 75% delipidated bR has its melting temperature close to that of the native bR and does have a pre-melting transition, although the pre-melting transition occurs at significantly higher temperature than that of the native bR (91 °C compared with 80 °C) and is still reversible. Furthermore, we have also observed that the reversibility of this pre-melting transition of both native and partially delipidated bR is time-dependent and becomes irreversible upon holding at 91 °C between 10 and 30 min. These results are discussed in terms of the lipid and lattice contribution to the protein thermal stability of native bR.

The importance of membrane proteins in nature is paramount. They are involved in vision (1), photosynthesis (2), phototaxis (3), and cellular ion pumps (4). Bacteriorhodopsin (bR) 1 is the only protein found in the purple membrane of Halobacterium salinarum (5) and is one of the most widely studied of these proteins because it can be easily isolated and purified. The ability of the protein to transfer a proton unidirectionally from the cytoplasmic side to the extracellular side of the membrane has stimulated the interest of both the biochemical and biomaterials communities (Ref. 6, and references therein). A retinal molecule is covalently bound to the Lys-216 residue and, after absorbing a photon of light, isomerizes from the all-trans to the 13-cis form. The energy stored in the bR at this point then drives a series of thermal reactions, forming spectrally distinct structural intermediates that result in proton pumping from the cytoplasmic side to the extracellular side of the membrane and, subsequently, reforming the groundstate bR (for reviews see Refs. 2 and 7). The pH gradient is used by the bacteria to drive the synthesis of ATP and is a much simpler photosynthetic system than the electron pumping mechanism in chlorophyll (8,9).
There are ϳ10 lipids per bR molecule that hold the membrane structure in a two-dimensional hexagonal close-packed lattice with the bR monomers held together as trimers that form the unit cell (10). These lipids are composed of various diether lipids, ϳ80% of which are acidic. 70% are phospholipids and 30% are glycosulfolipids (11). The wide interest in bacteriorhodopsin has pushed forward the boundaries of structure determination of membrane proteins. Over the years, various techniques such as electron diffraction (10,12), neutron diffraction (13), and x-ray structure analysis (14 -16) have been used to elucidate the structure of bacteriorhodopsin. The fact that the native purple membrane forms two-dimensional crystals has rendered it difficult to elucidate the three-dimensional structure. Landau and Rosenbusch reported a novel method to form such three-dimensional crystals by changing the lipidic phase of the bR (17), and subsequently the high resolution x-ray structure of bR to 1.55 Å was published (16) leading to a burst in the intensity of research in structure-function relationships of membrane proteins.
One of the questions that the x-ray structure could not answer is the location of bound cations in bR. Well-washed native bR contains ϳ4 mol of Mg 2ϩ and 1 mol of Ca 2ϩ per mol of bR (18). Upon removal of cations by acidification, deionization, or chelation (19 -21), bR no longer functions as a proton pump and the absorption maximum shifts from 568 to 603 nm (purpleblue). The effect of pH and cations on the protein structure and * This work was supported by the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Science, Office of Sciences, U. S. Department of Energy (Grant DE-FG02-97ER14799). 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  thermal stability was recently studied (22,23). It was shown that, upon cation removal, there is a relatively large change in the protein secondary structure, which is reversed upon regeneration with a number of different divalent cations. The thermal stability was only partially regenerated, and the extent is dependent on the cation identity (23).
In this report, we have studied the effect of removal of native lipids as well as removal from the bilayer into micelles on the protein structure and thermal stability in an attempt to understand the role of the lipids and the lattice on bR protein structure and its unusually high thermal stability. This effect is also compared with the effect of pH and cations on the protein, which was recently studied (22,23) in an effort to understand the interaction of cations, protein, and lipids.

MATERIALS AND METHODS
bR is grown and purified by a standard procedure described previously (24). To remove 75% of the lipids, the native bR was washed in 18 M⍀ doubly distilled water and added to 50 mM CHAPS in acetate buffer (pH 5) as described previously (25). The solution was allowed to equilibrate overnight, and the excess detergent was removed by centrifugation and washing in doubly distilled water three times. To monomerize the bR, the washed native bR was added to 10 mM Triton X-100 in 0.1 M Tris buffer (pH 7.4) (26). Monomerization was shown to be complete by the visible absorption spectrum as well as the observation that no sediment formed upon centrifugation of the sample at 19,000 rpm.
The visible absorption spectra of the samples were recorded on a Shimadzu UV-3101PC UV-visible spectrometer. To perform FT-IR spectral measurements the samples were re-suspended into D 2 O (Sigma Chemical Co., 99.9%). The CHAPS-treated sample was centrifuged at 19,000 rpm and re-suspended into D 2 O. This was repeated four times to ensure complete exchange for H 2 O and labile amide II N-H for N-D. The Triton-treated sample was exchanged into D 2 O by rotary evaporation of the aqueous detergent under high vacuum at 25°C to a very concentrated solution, followed by addition of D 2 O. This evaporation was repeated six times to ensure complete exchange. Repeated washing in D 2 O as purchased ensures that all samples are at the same pH upon sample measurement, because the buffer is washed out, and the pD of the D 2 O is ϳ7. The FT-IR spectra at each temperature were measured as previously described on a Nicolet Magna 860 FT-IR spectrometer (22). All measurements were repeated three times to ensure reproducibility and to calculate standard errors in transition temperatures.

RESULTS
The UV-visible spectra of native, CHAPS-treated, and Triton-treated bR are shown in Fig. 1. The CHAPS bR peak is blue-shifted by ϳ8 nm compared with the native, indicating that there is a small change in the retinal environment upon removal of the first 75% of the lipids. However, upon monomerization, there is a blue shift of ϳ20 nm in the peak showing that the retinal environment changes considerably. These results are in agreement with those previously reported (25,27).
To investigate conformational changes in the protein at 20°C upon 75% delipidation and monomerization into micelles, the FT-IR spectra in the amide I region are shown in Fig. 2. The spectra of native bR at neutral pH and upon acid blue formation is taken from Ref. 22 and shown for comparison. Native bR has its amide I stretching frequency centered at 1665 cm Ϫ1 . Upon removal of 75% of the lipids, there is a slight change in the amide I band. Two convoluted peaks are observed but are much more separate than in either native or monomeric bR. This indicates that there are two conformations present but in different proportions to the other samples. Compared with native bR, there is a slight increase on the higher energy side (ϳ1665 cm Ϫ1 ) and a slight decrease on the lower energy side (ϳ1658 cm Ϫ1 ) upon 75% delipidation. Upon solubilizing bR into micelles, it is shown that large changes in the amide I band have occurred. The band is centered at 1655 cm Ϫ1 and is much narrower. This shows that removal of the first 75% of the lipids only slightly affects the secondary structure of bR, but removal of the extra lipids that hold together the lattice affects the secondary helical structure greatly. Upon removal of these lipids and bR monomerization, many protein-protein interactions as well as protein-lipid interactions are lost that are important for the protein secondary structure. We had found previously that removal of cations by acidification or deionization has an effect of shifting the amide I band toward lower energy (22,28,29), but the shift is not as extensive as micelle solubilization. There is also a much higher degree of heterogeneity in the acid blue bR than in either native, partially delipidated, or monomerized bR as shown by the broader peak (22).
The effects of lipid removal and solubilization on the thermal transitions in bR are studied by investigation of the FT-IR spectral changes with temperature. This is shown for CHAPStreated bR in Fig. 3 and for Triton X-100-treated bR in Fig. 4. The spectral changes of each sample are considerably different, indicating a strong effect of lipids and lattice on the thermal stability and melting mechanism. Previous results (22,30,31) have shown that, as the temperature is raised, native bR shows a shift in the amide I ␣-helical frequency from the unusually high 1665 cm Ϫ1 to the more usual 1652 cm Ϫ1 at the pre-melting temperature of 80°C. This is followed by denaturation of the protein, detected by an increase in the intensity of the 1623 cm Ϫ1 band at 97°C. The increase in temperature is also accompanied by a decrease in the intensity of the amide II N-H band at 1545 cm Ϫ1 and an increase in the amide II N-D intensity at 1437 cm Ϫ1 due to the D 2 O solvent being exposed to the protein backbone N-H and exchanging to N-D. By analysis of Fig. 3, similar results are seen for the 75% delipidated sample, but the temperatures at which the transitions occur are different from the native bR. The amide I shift due to the pre-melting transition occurs at 91°C, a higher temperature than in the native, whereas the 1623-cm Ϫ1 band begins to rise at 94°C, a slightly lower temperature than the native. However, from Fig. 4, the bR monomer is considerably different in temperature dependence than either the native or the 75% delipidated bR. There is no shift in the amide I band, being already at the lower frequency, whereas the 1623-cm Ϫ1 melting band increases in intensity at 65°C showing a much reduced thermal stability for the bR monomer.
Quantitative analysis of the spectral changes with temperature is shown in Fig. 5 (A-E). The changes in absorbance with temperature from that at 20°C of the amide I and amide II peaks are shown in Figs. 3b and 4b. The corresponding spectral changes for native bR is taken from a previous study (22) and shown for comparison. The decay of the absorbance of the 1665-cm Ϫ1 band (1655 cm Ϫ1 for monomer), the rise of the 1652-cm Ϫ1 band (if present), and the rise of the 1623-cm Ϫ1 band of the melted protein show the amide I changes as a function of temperature during the conformational transitions. The pre-melting transition temperature is found from the derivative of Fig. 5b for each sample, and the melting transition similarly was derived from Fig. 5c.
The melting transition for the 75% delipidated bR is 94 Ϯ 1°C but at 65 Ϯ 3°C for the monomer, whereas it is at 97 Ϯ 1°C for the native bR. The N-H to N-D exchange is probed as shown in Fig. 5, d and e, respectively. It is interesting that there is little reduction in the N-H band for the bR monomer with temperature and that the intensity of the N-D band is significant even at lower temperatures. This suggests that no hydrophobic regions protecting the protein from the aqueous medium exist in the monomer to be exposed upon heating and may partially explain the low thermal stability of the monomer.
The pre-melting transition for 75% delipidated bR is determined to be 91 Ϯ 1°C, whereas it is at 80 Ϯ 1°C for the native bR and there is no such transition observed for the bR monomer. For the native bR the pre-melting transition has been shown to be reversible (32), and it is important to determine if this is the case for the CHAPS-treated bR. Fig. 6 (a and b) shows the visible absorption and FT-IR spectra as the CHAPS- 3. a, FT-IR spectra as a function of temperature in the amide I and amide II regions between 1700 and 1350 cm Ϫ1 for 75% delipidated bR. There is a pre-melting transition at which the amide I band shifts from 1665 to 1652 cm Ϫ1 at about 90°C. This is followed by an increase in the 1623-cm Ϫ1 band at the melting temperature of 93°C. There is a decrease in the amide II N-H stretch at 1545 cm Ϫ1 and a corresponding increase in N-D at 1439 cm Ϫ1 with temperature due to the enthalpic process of N-H to N-D exchange as D 2 O is exposed to parts of the protein. This exchange is more pronounced at the transition temperatures. b, FT-IR difference spectra of 75% delipidated bR taking 20°C as the background and subtracting from the higher temperatures. This allows the spectral changes to be seen more clearly. FIG. 4. a, FT-IR spectra as a function of temperature in the amide I and amide II regions between 1700 and 1350 cm Ϫ1 bR monomer. The amide I band is centered at the lower frequency of 1652 cm Ϫ1 at 20°C, and there is no pre-melting transition. At about 65°C there is an increase in the 1623-cm Ϫ1 band signaling the melting. The amide II N-H band is small even at low temperatures due to the protein in a more exposed environment initially allowing more N-H to N-D exchange. The unexchanged parts are exposed upon melting, and the complete N-H to N-D exchanges occur. b, FT-IR difference spectra of bR monomer taking 20°C as the background and subtracting from the higher temperatures. This allows the spectral changes to be seen more clearly. Note that the only features in the amide I region are a decrease at 1652 cm Ϫ1 and an increase at 1623 cm Ϫ1 . treated bR is raised from 20 to 91°C and allowed to cool. To avoid solvent effects, the heating was done in water and exchanged into D 2 O after cooling for the FT-IR measurements. For the visible spectra, a decrease in the intensity of the peak at 560 nm is observed, together with a broadening of the peak. The intensity of the peak recovers to its original intensity upon cooling (after 5-10 min at 90°C), suggesting that the retinal environmental change is reversible on this timescale. This is in agreement with results of the native bR (18). The baseline shift is typical for increased sample scattering, possibly due to aggregation and increased turbidity, and the fact that this base-line shift does not recover on cooling also suggests that the scattering (aggregation) is irreversible. The FT-IR spectra in Fig. 6b show the stability and reversibility of the pre-melting over time. The samples were held at 91°C for 2-60 min before being allowed to cool. It is shown that, within the first 10 min, there is a slight broadening of the amide I peak, but on the whole there is very little permanent damage. This agrees with our previous report on the transition effects on the CD, FT-IR, and photocycle kinetics (33). However, after about 30 min we see that the reversibility of the pre-melting transition reduces significantly, and the red shift of the ␣-helix peak no longer returns to the original position. Furthermore, we see an increase in the 1623-cm Ϫ1 band signifying denaturation of some of the ␣-helical structure into random coils, even though we did not go above the melting transition temperature. This suggests that the pre-melting state is a kinetic state on timescales longer than about 10 min. Similar results were observed for the native bR upon holding the temperature at 91 and 83°C (above the pre-melting for native bR but not for delipidated bR). Furthermore, holding the native bR at 78°C (just below the premelting transition) for up to 60 min caused no permanent shift in the amide I band (results not shown), suggesting that the shift and irreversibility is caused directly by the pre-melting transition and is not just a thermal effect. DISCUSSION Removal of the first 75% of the lipids has only a slight effect on the protein amide structure at physiological temperature. The efficiency of the photocycle decreases, and the lifetime of M increases with added CHAPS to native bR as reported previously (34). The hexagonal crystal structure of bR is maintained upon removal of these 75% lipids, although the unit cell has slightly different dimensions (25,40,55), suggesting that these lipids are only important for the overall photocycle efficiency and kinetics (25). However, upon the treatment of native bR with CHAPS, it has been shown that most of the Ca 2ϩ and Mg 2ϩ are removed (35), which is consistent with the removal of the low affinity sites. The effect of removing cations (by acidification of bR to form the blue membrane) on the structure and thermal stability was recently studied (22), and it was found that the secondary structure is significantly altered to a greater extent, as shown by the ϳ6-cm Ϫ1 red shift in the amide I region of the FT-IR. The thermal stability is also significantly lowered from 97°C in the native bR to 65°C in the acid blue membrane (22,36).
It is postulated that the color of the retinal is affected by the protonation state of Asp-85 (18,37). There are two proposals regarding binding of cations to bacteriorhodopsin affecting the protonation state of Asp-85. The first proposal regards the cations binding to the hydrophilic side groups within the protein (acids or alcohols) and perhaps trapped water, and there is an equilibrium between protonated and cation-bound states dependent on the pK a of the group(s) (18,38,39). Perhaps it is binding to one or more of the groups by the cation that controls the Asp-85 protonation state. The other proposal considers that the cations bind randomly to the lipids and charges on the membrane surface (25, 40 -42), and the Guoy-Chapman effect explains that the pH of the side groups is mediated from the surface pH, which in turn is mediated by random cation binding at the surface. Both of these proposals leave the Asp-85 group protonated in the blue membrane and, therefore, block the proton acceptor in the photocycle but by different means. Another study has shown that the titrations of the Asp-85 and the color-controlling site are not coupled (43). It has become a difficult task to assign whether there is a cation in the interior of the protein and, thus, to explain the exact role of the cation in the proton-pumping mechanism.
We recently studied the effect of pH on the protein structure and thermal transitions in an effort to understand the interaction of cation with the protein (22). We found that the native bR at neutral pH has the highest melting point and that upon decreasing pH the stability is reduced to a minimum in the acid blue membrane of 65°C. If the effect of cation is a random one from binding to the lipids to control the pH, then removal of the lipids should have a similar effect on the thermal transitions as pH. This is because, according to the Guoy-Chapman theory, removal of 75% of the surface charge should lower the surface affinity for cations; i.e. the same effect as lowing pH decreases the cation affinity by protons competing for the negatively charged sites. It is clear from our results that the effect of lowing pH is far more dramatic on the melting transition than removal of up to 75% of the lipids.
The amide I band is red-shifted to a greater extent upon monomerization than it is by removal of the cations at 20°C, suggesting that removal of the cations only partially alters the protein structure relative to bR solubilization into monomers and the micellar environment. The bR environment in the micelle is considerably different than in the trimeric lattice, and the large conformational changes are not surprising. However, a surprising result is that the effect on the thermal stability is the same as removal of the cations; i.e. there is no pre-melting transition and the melting temperature is 65°C. This indicates close stabilizing effects of both the cations and the lipids (and/or lattice structure). On the other hand, removal of the cations causes the proton-pumping function to cease as shown by the lack of formation of the important M intermediate, whereas removal of the lipids that keep the lattice and micelle solubilized does not cause the proton pumping to cease. The kinetics are affected in the early stages of M formation, but the photocycle is qualitatively the same as the native in these two very different environments.
This suggests one of two possibilities: (i) either the effects of some of the cations and the lipids are separate (binding of the FIG. 6. a, visible absorption of CHAPS-treated bR as a function of temperature to show that heating to the pre-melting transition and cooling causes no permanent damage to the retinal binding on short (Ͻ10 min) timescales. b, extent of reversibility of amide I secondary structure upon holding the sample at 91°C for 2-60 min. For timescales Ͻ10 min the transition is mostly reversible, whereas for longer timescales the transition becomes irreversible. There is also some denaturation even though we did not go above the melting transition. cation directly to the protein) or (ii) if removal of the lipids that hold the lattice causes removal of the cations, the cations are not directly important for the function, at least for that of the delipidated bR or bR monomer.
Regeneration of the deionized bR with a variety of divalent cations at 10:1 divalent cation:bR ratios increased the thermal stability, although not to the same extent as the native bR. The melting temperature and mechanism were dependent on the identity of the divalent cation indicating that the interaction of the bR with the cation depends on the chemical identity of the cation (23). This is best explained by direct protein-cation binding rather than random surface binding where only the charge density should affect the structure and stability.
Our results show that the first 75% of the lipids are mostly superfluous to the structure and irreversible melting transition. The pre-melting transition, however, shows a strong dependence on the lipids whereas it showed no dependence on the pH or the cation with which deionized bR is regenerated (22,23). The origin of this pre-melting transition is still ambiguous at the present time. It had been discovered early that native bR has an anomalously high amide I stretching frequency at 1665 cm Ϫ1 (44), blue-shifted by ϳ12 cm Ϫ1 from the usual 1652-cm Ϫ1 ␣-helical frequency in globular proteins. The reason for this frequency has not been positively identified, but calculations on polyalanine (45) have pointed to the possibility of an ␣ II helix, in which the dihedral angles between the backbone CϭO and N-H planes change, resulting in a lengthening of the intrahelical H-bond by ϳ0.14 Å. Other spectroscopic evidence such as lunear dichroism (46), CD (47), 13 C NMR (48 -50), and Raman (51) also supports the existence of ␣ II helices in bR, from the shapes of the dichroic spectra, the shifts in the NMR, and the Raman shifts, respectively. Each was found to be inconsistent with ␣ I alone. There has been evidence based on x-ray structure analysis that the pre-melting transition may be a breaking of the crystal packing within the membrane into a liquid-like phase (52). Evidence based on CD and FT-IR spectroscopy has pointed to the possibility of a helical transition, in which the bR helices change conformation from predominantly ␣ II to predominantly a I during this transition (30,31,53). It is possible that these two explanations are coupled and that the helix conformation is determined by the lattice arrangement or vice versa.
In any case, the 1652-cm Ϫ1 band can be used to monitor protein conformational changes during the pre-melting transition, and we find that removal of 75% of the lipids control this transition much more than the pH. This clearly shows that the effect of removing some of the lipids and the effect of decreasing pH of the sample are not the same for this transition. In fact, the temperature of the pre-melting transition increases upon removal of 75% of the lipids. This transition is not observed upon removal of the remaining lipids that hold the lattice structure together, probably due to the protein adopting the more usual ␣-helical conformation at physiological temperature as a result of the disappearance of the lattice structure upon solubilization prior to heating.
An interesting observation that we have made concerns the reversibility of this pre-melting transition. We found that if the sample is left at 91°C (above the pre-melting transition but below the melting transition) for either the native or the 75% delipidated bR, the transition becomes irreversible between 10 and 30 min, whereas it is reversible on shorter timescales. The previous work using x-ray analysis, where it was found that the hexagonal crystal structure does not restore after 60 min at 85°C (54), suggested the lattice structure origin of the premelting transition, which becomes irreversible over time, and our results agree with this. It is possible that contacts present in the lattice structure begin to break upon pre-melting but can be reformed on short timescales if diffusion does not take them out of range. Over time, however, these contacts could diffuse too far away to reform. It seems that these contacts are stronger in 75% delipidated bR, suggesting that they are specific protein-protein or possibly protein-cation contacts. This is supported by the lower unit cell dimensions of delipidated bR (25,40,55), suggesting that bR monomers are closer to each other in lipid-depleted bR, thus increasing protein-protein interactions. This is not to say that the ␣-helical transition does not occur. In fact, the shift in the amide I peak to 1652 cm Ϫ1 becoming irreversible on the same timescale supports our hypothesis that these two effects may be coupled. Furthermore, our previous suggestion of the cation binding affecting the amide I peak (22) could be explained in these terms also. Upon going through the pre-melting transition, binding to one or more of the cations may be lost, which then begins to slowly diffuse away from specific binding sites in the protein. When this diffusion takes the one or more cations outside the range of possible rebinding, the transition becomes irreversible and eventually leads to melting. Considering the likelihood of low mobility of the cation in the protein (especially the interior), these timescales of ϳ10 min could support this hypothesis. This leads to the pre-melting being a kinetic process, with the temperature of its onset depending on the lipid removal but not on the proton concentration at the surface (22).
In contrast to the pre-melting transition, the melting transition shows a much larger dependence on pH and cation than on the lipids. There is little change in the melting temperature upon removal of as much as 75% of the lipids (and negative surface charge), whereas there is a large reduction in the melting temperature with pH Ͻ 7. This also adds to the suggestion that pH and lipids are not completely coupled effects and, hence, suggests that the pH affects the protein other than through the lipids. This is understandable, because changing the pH changes the protonation state of ionizable side groups within the protein. One needs to go to a pH of ϳ1.2 to change the ionized state of the phospholipid groups of the lipids (21,41).
Upon monomerization of bR, we see similar structural and thermal (melting) characteristics to that of deionized bR. Actually, this is surprising considering the fact that the framework that holds the two-dimensional lattice and trimeric structure is completely lost upon monomerization and many proteinprotein interactions are broken, whereas it is still intact in the blue bR. This shows that cations and the lipids that hold the lattice framework together in native bR hold similar and considerable thermodynamic constraints onto the protein.
The implications for the environment (lipids and cations) of the protein and its structure may affect conclusions drawn when changing the membrane structure. During the preparation of the three-dimensional crystals (17) prior to x-ray scattering experiments, the membrane environment of bR is changed. First, the bR is solubilized into neutral detergent followed by its insertion into lipidic cubic phases. We have shown that the monomerization into micelles alters the protein structure considerably, more than the cations do. Because the cations cannot be seen in the x-ray experiments, one should not draw premature conclusions as to the native bR structure (e.g. lack of ␣ II helices, lack of cations, etc.) until it is proven that the bR does not change structure upon reinsertion into such different environments. There have been recent advances in this by finding crystallization methods that do not require solubilization (56), even though cations still haven't been found.
In summary, we have shown that the effect of cation removal and the effect of lipid removal and aggregation state have drastically different effects on the protein structure and its thermal transitions in bR. Furthermore, monomerization into micelles and removal of all the cations affect the protein thermal stability in the same way. However, their effects on the color and proton pump function are very different. We have shown that the stability to the pre-melting transition depends more on the lipids than on the cation or pH but that the stability to melting depends much more on the cation than the first 75% lipids, until the lattice structure is broken. This indicates that even though the lipid and cation effects are partially connected, possibly by low affinity binding sites on the surface, the effects are not identical. The H ϩ ions change the ionization state of the amino acids that are important in cation binding and thermal stability of the protein, but 75% delipidation has little effect on them. These sites might only be destroyed upon removal of the lipids that break the lattice and monomerize the bR into a different, micellar environment.