|Home | About | Journals | Submit | Contact Us | Français|
Caveolin-1 is a membrane protein that possesses an unusual topology where both N- and C- termini are cytoplasmic as a result of a membrane-embedded turn. In particular, proline 110 has been postulated to be the linchpin of this unusual motif. Using a caveolin-1 construct (residues 62–178) reconstituted into dodecylphosphocholine micelles with and without a cholesterol mimic, the changes that occurred upon P110A mutation were probed. Using far UV circular dichroism spectroscopy it was shown that cholesterol attenuated the helicity of caveolin-1, and that mutation of P110 to alanine caused a significant increase in the α-helicity of the protein. Near UV circular dichroism spectroscopy showed significant changes in structure and/or environment upon mutation that again were modulated by the presence of cholesterol. Stern-Volmer quenching and λmax analysis of tryptophan residues showed that the proline mutation caused W85 to become more exposed, W98 and W115 to become less exposed, and W128 showed no change. This finding provided evidence that regions proximal and far away from the proline are buried differentially upon its mutation and therefore this residue is strongly tied to maintaining the hydrophobic coverage along the caveolin-1 sequence. In the presence of cholesterol, the accessibilities of the two tryptophan residues that proceeded position 110 were altered much more significantly upon P110A mutation than the two tryptophans aft P110. Overall this work provides strong evidence that proline 110 is critical for maintaining both the structure and hydrophobic coverage of caveolin-1 and that cholesterol also plays a significant role in modulating these parameters.
Caveolae are 50–100 nm invaginations present in the plasma membrane of many cell types, and are involved in signal transduction, endocytosis, and mechanoprotection . The integral membrane protein caveolin is essential to caveolae formation and mediates many of its functions . Misregulation and mutation of caveolin has been linked to disease states such as muscular dystrophy, heart disease, Alzheimer’s disease, and cancers . Although three isoforms of caveolin exist (−1,−2, and −3), caveolin-1 is the most ubiquitously expressed across cell types. The relationship between caveolin and cholesterol appears to be intimate; some have professed that it is a cholesterol binding protein . Caveolin-1 contains a 33 amino acid long intramembrane domain (residues 102–134) that has been postulated to form a highly unusual intramembrane loop, resulting in a cytoplasmic location for both the N- and C-termini [5–8].
A number of studies have opined as to where the putative intramembrane loop may be located sequence-wise. The first study was based solely on primary sequence analysis, and suggested that four residues (G108, I109, P110, and M111) were responsible . This assertion was based on the fact that both proline and glycine have high turn forming propensities, and are located near the center of the intramembrane domain. With the implication of proline 110 as being part of the turn, there was increased interest in examining the effect that the mutation of this residue to alanine has on the topology and structure of caveolin. An in vivo study on an N-terminally FLAG-tagged caveolin-1 construct showed that upon P110A mutation, the N-terminus that is normally cytoplasmic became extracellular . From this result, coupled with molecular dynamics simulations, the authors postulated that the P110A mutation caused caveolin-1 to adopt a transmembrane as opposed to an intramembrane loop configuration. However, this postulation was not supported by an in vitro glycosylation study which concluded that P110A had a topology identical to that of the wild-type protein . In contrast, studies using a short, solubility-enhanced construct encompassing residues 103–122, showed significant changes in bilayer depth and α-helicity when the P110A mutation was made .
Secondary structure analysis using nuclear magnetic resonance spectroscopy (NMR) showed that the intramembrane domain was largely helical with a break at residues 108–110, which corroborated the initial primary sequence analysis about the location of the turn, but revealed that M111 may not be part of the putative turn forming motif [12–14]. Furthermore, NMR experiments showed that the mutation of the proline at position 110 to alanine appeared to significantly alter the behavior of the native protein .
In this report, we employed a caveolin-1 construct encompassing residues 62–178 (Cav162–178) reconstituted into dodecylphosphocholine micelles (DPC) with and without a cholesterol mimic. Using near and far UV circular dichroism spectroscopy (CD) coupled with fluorescence spectroscopy of individual tryptophan residues, we were able to probe the structural and solvent accessibility changes that occur when proline 110 is mutated to alanine.
DNA for Cav162–178, was synthesized by Genscript Corporation (Piscataway, NJ). The Cav162–178 gene was cloned, over-expressed, and purified according to previously reported protocols . After purification using high performance liquid chromatography, the identity of the protein was confirmed using matrix assisted laser desorption-ionization time of flight spectrometry. Next, purified Cav162–178 was aliquoted and lyophilized using protocols described by Rieth et al . Mutant constructs were prepared using the Agilent quik change mutagenesis kit (Santa Clara, CA) for a total of nine additional constructs. Cav162–178 contains four tryptophan residues: W85, W98, W115, and W128. For single tryptophan mutants, one of the four native tryptophan residues was retained and the other three were mutated to phenylalanine to generate the following constructs: W85 Cav162–178, W98 Cav162–178, W115 Cav162–178, and W128 Cav162–178. Additionally proline 110 of Cav162–178 was mutated to alanine to generate the following constructs: Cav162–178 P110A, W85 Cav162–178 P110A, W98 Cav162–178 P110A, W115 Cav162–178 P110A, and W128 Cav162–178 P110A.
To 1.2 mg of lyophilized Cav162–178, 3 mL of ice-cold buffer comprised of 20 mM phosphate pH 7.0, 100 mM NaCl, and 50 mM DPC (Anatrace, Maumee, OH) was added to reconstitute the protein. Samples had final protein concentrations of 30 µM except for near UV CD experiments where the protein concentration was 150 µM. After vortex mixing until clarification, each sample was filtered using a 0.2 µm filter to remove particulates. All mutants were reconstituted in an identical manner.
For samples containing cholesterol-PEG600 (Sigma Aldrich, St. Louis, MO), an alternative reconstitution procedure was utilized. Cav162–178 constructs were first dissolved into 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) at a concentration of 2 mg/mL. A 1.2 mg quantity of protein was added to a 3 mL solution of 60 mM phosphate pH 7.0, 300 mM NaCl, 90 mM DPC, and 60 mM cholesterol-PEG600. The HFIP concentration was adjusted to 50% (v/v) and the solution was lyophilized over-night. The solution was then rehydrated with 9 mL of water and lyophilized for an additional 15 hours to remove all traces of HFIP. The powder was taken up into 9 mL of H2O to yield a solution which had a final buffer composition of 20 mM phosphate pH 7.0, 100 mM NaCl, 30 mM DPC, and 20 mM CPEG. Samples had final protein concentrations of 30 µM except for near UV CD experiments where the protein concentration was 150 µM. This solution was then reconstituted by incubation on ice for 1–2 hours with occasional vortex mixing. The solutions were filtered through a 0.2 µm filter before use.
Both far and near UV circular dichroism spectroscopy experiments were performed using a JASCO circular dichroism spectrophotometer (Easton, MD). The experiments were carried out at 298 K using a quartz cuvette with a 0.1 mm path length and a 1 mm path length for far UV and near UV experiments respectively.
For far UV experiments, spectra were obtained from 240 to 190 nm with a 1 nm data point interval, 2s digital integration time, and a 1 nm bandwidth in step mode accumulating 16 scans. For near UV experiments, spectra were obtained from 320 to 260 nm with a 0.5 nm data point interval, 1s digital integration time, and a 0.5 nm bandwidth in step mode accumulating 16 scans. In each case, a background spectrum employing buffer without reconstituted Cav162–178 was subtracted from the protein containing spectra. Circular dichroism spectroscopy experiments were performed on Cav162–178 and Cav162–178 P110A and were repeated two times for each micelle system.
Steady-state fluorescence emission spectra were acquired at 298 K with an Eclipse fluorometer (Agilent, Santa Clara, CA) using a 1 × 1 cm quartz cuvette. The excitation wavelength was 295 nm which selectively excites tryptophan residues . Both excitation and emission slits were set to 5 nm. The emission spectrum was measured from 315–500 nm with a scan speed of 1 nm/s and 0.5 nm data point increments, averaging four scans. The λmax and Γ values were determined by fitting the data to a log-normal distribution using Igor Pro 6.22A (WaveMetrics, Inc., Lake Oswego, OR) . ΔΓ values were calculated to assess conformation heterogeneity differences between single tryptophan mutants used in the study. First, each determined λmax value was utilized to generate Γideal by using the line equation based on the λmax-Γ relationship for N-acetyl-tryptophan-amide dissolved into solvents of varied polarity derived by Ladokhin et al . This value was then subtracted from the actual Γ value obtained from the log-normal fit to obtain ΔΓ.
Fluorescence quenching experiments were performed directly after λmax determinations utilizing a kinetic mode of data acquisition. First, freshly prepared dithiothreitol was added to a final concentration of 1 mM. The sample was then continuously excited at the 295 nm wavelength and the fluorescence intensity at the λmax of emission was monitored at a rate of 1 measurement/s. The fluorescence intensity was monitored for 3 minutes to ensure the sample was at equilibrium and to determine the unquenched intensity. After this time period, an 18 µL aliquot of 20 % (w/w) potassium iodide in 1 mM dithiothreitol was added and the intensity was monitored for one minute before the next addition of quencher. Addition of the quencher was repeated to obtain quenching data in the 0–100 mM range. Intensity averages at each quencher concentration were fit to the modified Stern-Volmer equation to determine the fractional accessibility of tryptophan residues :
where I0 and I are the unquenched and quenched fluorescence intensity respectively, Ka is the Stern-Volmer quenching constant for the accessible fraction, Q is the quencher concentration, and fa is the fraction of the initial fluorescence that can be quenched at an infinite quencher concentration. All fluorescence experiments were repeated three times for the following constructs, W85 Cav162–178, W98 Cav162–178, W115 Cav162–178, W128 Cav162–178, W85 Cav162–178 P110A, W98 Cav162–178 P110A, W115 Cav162–178 P110A, and W128 Cav162–178 P110A in both micelle systems.
To probe the structural and accessibility changes of caveolin-1 that occur upon the mutation of proline 110 to alanine, a construct comprised of residues 62–178 (Cav162–178) was utilized, as previous studies in our laboratory have shown that Cav162–178 is monomeric in DPC micelles . The monomeric behavior of Cav162–178 was important, because we did not want the wild-type construct to be biased by spectroscopic changes arising from oligomerization, which could cloud comparisons to the P110A mutant and/or effect of the cholesterol mimic. It should also be noted that studies have shown that residues 62–178 of caveolin-1 behave similarly to full-length caveolin-1 in a cellular context, which supports the relevancy of employing the 62–178 construct . Caveolin-1 contains three cysteines that are palmitoylated (residues C133, C143, and C156). However, Dietzen et al have shown that mutation of these residues to serine was not deleterious to caveolin-1 function . Additionally, synthetic palmitoylation of caveolin-3, a close homologue of caveolin-1, at the analogous three sites, resulted in little to no change in the behavior of the protein . Therefore, in our construct we mutated each of the cysteines to serines to avoid non-biologically relevant disulfide bonding. A methionine at position 111, which is not strictly conserved (it is a leucine residue in both caveolin-2 and caveolin-3 which are very close homologues of caveolin-1), was mutated to leucine to facilitate protein purification (see Materials and Methods for details). Figure 1 shows the sequence of the construct utilized.
In order to fully characterize the accessibility changes imparted by the P110A mutation along the sequence of the protein, single tryptophan mutants of the four native tryptophan residues in caveolin-1 were generated for both Cav162–178 and the P110A mutant yielding a total of 8 constructs. Importantly, the tryptophan residues (W85, W98, W115, and W128) are reasonably spaced around proline 110. In each of these constructs, one tryptophan was maintained while the others were mutated to phenylalanine. Phenylalanine was chosen because of its similarity to tryptophan in terms of hydrophobicity, size, and aromaticity. Far UV circular dichroism of the single tryptophan mutants showed that variations in helicity between the mutants were less than 10% when compared to the wild-type (data not shown). In addition, for data analysis, only differences within a particular single tryptophan mutant are analyzed; therefore, the slight variation in helicity between the single tryptophan mutants is neutralized.
For these studies, dodecylphosphocholine (DPC) micelles were chosen because they mimic the high percentage of phosphatidylcholine headgroup containing lipids in caveolae, and have been used extensively in biophysical studies of membrane proteins [22, 23]. Micelles were chosen in preference to vesicles, to avoid orientational heterogeneity (Cav162–178 facing inside or outside of the vesicle) which would preclude clear biophysical interpretations, especially in the case of the quenching experiments where the population of inward facing caveolin would not be accessible to the quencher.
Cholesterol is estimated to range from 35 to 45% of the lipid population in caveolae [22, 24]. Because it is difficult to incorporate high levels of cholesterol in DPC micelles (maximum solubility ~5%), cholesterol-PEG600 (CPEG, 40%, average between 35% and 45%) was used because of its increased solubility. CPEG is a very close mimic of actual cholesterol because it has an identical ring structure; the only alteration to the molecule is a polyethylene glycol (PEG) moiety attached to the C3 hydroxyl group. CPEG has been widely used as a cholesterol mimic. For example, a fluorescent analogue of CPEG (fluorescein moiety incorporated at the distal end of the PEG linker) has been utilized to image cell surface localized lipid rafts, and to examine the distribution of intracellular cholesterol-rich domains . When added to live cells, CPEG co-localizes with the known lipid raft markers filipin, CD59 and GM1. Additionally, when CPEG was reconstituted into liposomes of low cholesterol content, it could be scavenged by cholesterol-rich liposomes. Importantly, when CPEG is added to cells, it localizes to caveolae, and was found to associate nearly exclusively with caveolin-1 using density gradient centrifugation experiments . These lines of evidence strongly suggest that CPEG has a behavior that will alter the micellar environment tending it towards a raft-like domain, akin to that of caveolae.
Proline 110 is postulated to be a key residue of the intramembrane turn of caveolin-1 [8, 9, 11–13]. Therefore, proline 110 could be essential for the proper fold of Cav162–178, and its mutation to alanine could cause significant changes in the secondary structure of Cav162–178. Far UV circular dichroism spectroscopy is a powerful technique which gives information about the secondary structure present in a polypeptide. Figure 2A shows the far UV circular dichroism spectra of Cav162–178 and its P110A mutant. As one can see, both have clearly visible α-helical signatures denoted by minima at 208 nm and 222 nm and a maximum at ~193 nm . However, the P110A mutant has a more intense maximum at ~193 nm, and slightly lower ellipticities at 208 nm and 222 nm indicating an increase in helicity. Using the method of Barrow et al for the 208 nm wavelength, the helicity of Cav162–178 and its P110A mutant can be roughly estimated . The difference, Δhelicity (Cav162–178 P110A − Cav162–178), was 21 ± 6% (average of two independent trials for both wild-type and mutant). This result shows that there is a significant increase in α-helicity when the P110A mutation is made. Likely, proline 110 is behaving characteristically as a helix-breaking residue in the native protein, and when the mutation to alanine is made, the protein becomes more helical.
We wanted to ascertain if the presence of CPEG would have an impact on the secondary structure of Cav162–178, as cholesterol is thought to have an intimate relationship with caveolin-1 . Figure 2B shows the spectra for Cav162–178 and Cav162–178 P110A, this time in the presence of CPEG at a biologically relevant concentration (40%). Again as seen in the case without cholesterol, a more intense maximum at ~193 nm is observed along with slightly lower ellipticities at 208 nm and 222 nm when the P110A mutation is made, which indicates an increase in helicity. However, in contrast, a Δhelicity of only 5 ± 4% (average of two independent trials for both wild-type and mutant) is observed revealing the P110A is not increasing the helicity nearly as dramatically as when CPEG is not present. In addition, CPEG lowers the overall helicity of Cav162–178 by 11 ± 4% (average of two trials). These results clearly show that cholesterol may have a strong modulating effect towards caveolin-1 by decreasing its helicity and thus making the difference between the wild-type and P110A mutant less significant. This supports the strong relationship that has been postulated between cholesterol and caveolin-1; however, further analysis will be needed to deduce the exact nature of this interaction.
Near UV circular dichroism spectroscopy is a technique that can be used to yield information about the specific state of a protein (e.g. tertiary fold); the aromatic amino acids phenylalanine, tyrosine, and tryptophan absorb in the 260–320 nm region, and can give rise to distinct elliptical signatures when they are held in an asymmetric environment . If the P110A mutation causes major changes in the environment around the aromatic residues in Cav162–178, the near UV spectrum will have an altered appearance. Fortunately, the sequence of Cav162–178 is conducive to near UV CD analysis as the aromatic amino acids are distributed fairly evenly across the protein (Figure 1). Figure 3A compares the near UV circular dichroism spectra of Cav162–178 and Cav162–178 P110A in DPC micelles, and Figure 3B shows the same comparison in the presence of CPEG. In both cases, the two spectra have distinct non-overlapping morphologies which are indicative of a significant change in the asymmetry of the aromatic sidechains, although the changes appear to be less dramatic when CPEG is present. Overall the differences between Cav162–178 and the P110A mutant could be due either to changes in the global tertiary fold of Cav162–178, alterations in the interaction between the micelle and Cav162–178, or differences in oligomeric state. Clearly, the near UV data indicates that the P110A mutation causes significant changes in the behavior of the protein and that CPEG modulates this behavior.
The comparison of Cav162–178 in different micelle environments (with and without CPEG) is shown in Figure 3C. There were noticeable changes in the morphology of the spectra as a result of altering the micelle environment. The fact that the spectral differences appear to be dependent on the composition of the membrane mimic (i.e. the presence of CPEG) suggests that changes in the protein-micelle environment are a very important component in the differences observed between the spectra. However, more experiments will be needed to conclusively flesh out the origins of the spectral differences.
The tryptophan residues native to Cav162–178 can act as reporters of solvent exposure by analyzing their fluorescence emission maximum (λmax, Supplementary Figure 1); they are located at positions 85, 98, 115, and 128 . W85 and W98 are located “N-side” to P110 whereas W115 and W128 are located “C-side”. The location of the tryptophan residues is adventitious; they are proximal to P110, in particular W98 and W115. Analysis of the change in λmax (Δλmax) can be used to compare the degree of aqueous exposure between Cav162–178 and Cav162–178 P110A constructs. Generally, a blue shift in λmax (Δλmax<0) is indicative of greater membrane burial or shielding from the aqueous environment, while a red shift indicates an increase in aqueous accessibility (Δλmax>0) [29–32]. Table 1 shows the λmax data.
The first tryptophan in Cav162–178 is at position 85, 25 residues away from P110. The Δλmax is 6.3 ± 0.8 nm, showing that this region of the protein becomes more solvent accessible when P110 is mutated to alanine. The next tryptophan in Cav162–178 is at position 98 and has a Δλmax of −6.7 ± 1.0 nm. This indicates that the proline replacement results in greater micelle coverage of this region. The next tryptophan in Cav162–178 is W115, and this position is closest to the putative turn region as it is located 5 residues away from position 110. Analysis of the spectra shows a Δλmax of −0.5 ± 1.6 nm, which indicates a relatively miniscule difference between Cav162–178 and the P110A mutant environments around W115. The last tryptophan is W128 which yielded a small Δλmax value of −3.1 ± 1.4 nm, indicating that W128 was only slightly more buried in the case of the mutant. Taken together, it appears that the P110A mutation affects the N-side region (W85 and W98) of the protein much more significantly than the C-side region (W115 and W128). However, it is interesting that W85 and W98 shift in opposite directions with W85 becoming more exposed while W98 becomes less exposed. This shows that there is a significant amount of “long-distance” communication in the protein that is facilitated by P110. It should also be noted that the aggregate λmax for Cav162–178 (all 4 tryptophan residues present) is consistent with the average of the λmax of the 4 single tryptophan constructs, and supports (but does not prove) that the mutations needed to generate the single tryptophan constructs have not significantly altered the nature of the protein (data not shown).
Identical experiments were carried out, this time in the presence of CPEG. W85 had a Δλmax value of 7.5 ± 0.2 nm again showing that the N-terminal region of the protein becomes more solvent accessible. W98 yielded a Δλmax of −6.0 ± 0.1 nm, and points to a deeper burial for this position upon mutation of proline 110 to alanine. For W115 the Δλmax of 0.1 ± 0.1 nm, was in agreement with the changes observed in pure DPC micelles suggesting that the hydrophobic coverage of this region of the protein is not strongly affected by the P110A mutation. The final tryptophan at position 128 had a Δλmax value of −0.1 ± 0.2 nm showing a negligible change in solvent exposure. Again this shows that P110 plays an important role in modulating the solvent accessibility of the protein, and these changes are localized to the N-side tryptophan residues. In addition, the magnitude of changes in the Δλmax values were very similar whether CPEG was present or not, suggesting that the P110A mutation modulates the exposure of Cav162–178 in the same fashion.
The width of a tryptophan emission spectrum (Γ) can be used to evaluate the conformational heterogeneity (i.e. dynamics) of membrane proteins; this method is known as position-width analysis . Since the width of a peak is also a function of λmax, in order to truly assess conformational heterogeneity in a comparative manner, Γ must be corrected (see Materials and Methods for more information). Supplementary Table 1 shows corrected Γ values (ΔΓ) for all single tryptophan mutants (W85, W98, W115, and W128) with and without the presence of CPEG. Next, to compare ΔΓ with and without the P110A mutation, the differences between the corrected widths is taken (ΔΔΓ, Supplementary Table 1). The ΔΔΓ values for each single tryptophan mutant did not vary significantly between Cav162–178 and its P110A mutant (no ΔΔΓ values were greater than 1.5 nm). This result indicated that the P110A mutation was not causing drastic alterations in the conformational flexibility of Cav162–178 in regions flanking P110.
It should also be noted that in general, CPEG causes a red-shift or more solvent exposure for each tryptophan (speaking here of λmax not Δλmax). This supports the notion that the CPEG changes the way that the protein associates with or orients in the micelle as a whole, perhaps indicating a more superficial interaction. One caution in interpreting Δλmax in the context of solvent exposure, is that significant H-bonding of the indole NH group with partners other than water can red-shift the λmax values and can cloud the direct interpretation of the data .
In order to bolster the conclusions from the λmax shifts and avoid misinterpretation due to non-water hydrogen bonding to the indole NH group, fluorescence quenching was carried out to verify the aqueous exposure of each tryptophan (Modified Stern-Volmer plots are shown in Supplementary Figure 2) . Quenching of tryptophan fluorescence by the soluble quencher iodide is greater for residues that are exposed to the aqueous environment (large fractional accessibility, fa) than for those buried within a membrane (small fa). In these studies a Δfa >0, means that the tryptophan has become more exposed, whereas Δfa<0 means that the tryptophan has become less exposed upon P110A mutation.
Table 2 shows the fa data for Cav162–178 single tryptophan mutants reconstituted into DPC micelles. The first tryptophan, W85, shows a Δfa of 6 ± 1% upon P110A mutation. This indicates that a significant increase in solvent accessibility occurred, and is in accordance with the changes observed in λmax. The next tryptophan residue in the sequence, W98, yielded a Δfa of −5 ± 1%, indicating that the accessibility had decreased. This agrees with the change in λmax between the constructs discussed in the previous section. W115 yielded a Δfa of −7 ± 1%, and shows that this sequence position is less solvent accessible. Finally, W128 gave a Δfa of 2 ± 1%, a very minimal increase in accessibility. This was also reflected in the small change in the emission maximum of W128 between Cav162–178 and the P110A mutant. Therefore, it appears that the two N-side tryptophan residues, W85 and W98, become more and less exposed respectively, similar to what was observed before with Δλmax. Of the two C-side residues only W115 is affected, where it has deeper coverage by the micelle. This is very interesting because this lessened solvent accessibility was not reflected in Δλmax, which showed no change, and is likely indicative of increased non-water molecule hydrogen bonding to W115 in the case of the mutant. A physically plausible explanation for this result is that, in the case of the P110 mutant, W115 is in a less deeply buried position within the micelle where it is able to hydrogen bond to the DPC headgroups. This finding indicates that P110 may be critical for the proper positioning of the juxta-turn residues within the hydrophobic matrix.
Next, the changes in fa were assessed in the CPEG rich micelles (Table 2). Here the first tryptophan, W85, experienced a massive shift in its aqueous accessibility upon the P110A mutation yielding a Δfa of 19 ± 1%. W98 gave a Δfa of −15 ± 2%, which indicates that this position has slid deeper into the hydrophobic core of the micelle. W115 gave a Δfa of −7 ± 1%, in agreement with the change observed in pure DPC micelles in terms of increased burial and the magnitude of the accessibility change. The final tryptophan, W128, gave a Δfa of 1 ± 2%, showing that the environment had not been altered upon P110 mutation to alanine. Therefore, it seems that the sterol environment causes the same tryptophan accessibility changes that are observed in pure DPC micelles. However, the magnitude was greater for the N-side portion of Cav162–178 where both W85 and W98 have become significantly more and less exposed respectively. The degree of the change is similar between micelle types for C-side tryptophans, W115 and W128. Therefore the inclusion of CPEG in the micelle changes the exposure of the N-side region (W85 and W98) of the Cav162–178 P110A mutant more so than the C-side region. This finding also supports there being a significant amount of “play” between CPEG and the protein.
Interestingly, the tryptophan residues in Cav162–178 were always less accessible to iodide when reconstituted into micelles containing CPEG than they were in micelles composed purely of DPC. This indicates that the micelle coverage of the protein was much deeper in the presence of CPEG, in contrast with λmax data which showed that Cav162–178 constructs were always more red-shifted. A plausible explanation would be that the PEG moiety was substantially increasing the ability of the indole NH to find non-water molecule hydrogen bonding partners, which would red-shift the spectra. This result highlights the importance of examining both quenching and changes in λmax to obtain the maximum amount of information from a system. It should also be noted that the increase in burial that was observed for Cav162–178 constructs reconstituted into the DPC micelles containing CPEG is incongruous with studies performed by Aoki and co-workers who saw that cholesterol limited the penetration of caveolin-1 (residues 103–122 with added flanking lysine residues for solubility) into vesicles . However, differences in construct length and design, and membrane mimic make a clear scientific reconciliation between the data difficult.
Recent biophysical studies have indicated that the proline residue at position 110 may be critical for the formation of the putative intramembrane turn of caveolin-1 [8, 9, 11–13]. Using circular dichroism spectroscopy in the far UV region, we have shown that there is a modest but significant increase in helicity when the P110A mutation is present for Cav162–178. Most likely this increase is localized to the region of the proline 110, as it is known to be helix breaking. In addition, this increase in helicity appears to be dependent on cholesterol, as cholesterol seems to modulate the overall helicity of the protein decreasing it significantly. Near UV circular dichroism measurements provide evidence that changes in the asymmetric environment of the aromatic sidechains (near UV region) occurred, and suggest that P110 is important for modulating the overall disposition of the protein. The replacement of the proline 110 with alanine clearly modified the solvent accessibility of Cav162–178. Interestingly, it appeared that the accessibility of caveolin-1 in micelles containing a cholesterol mimic was altered to a much greater extent, and suggests that cholesterol is very important for the proper depth of the protein within a hydrophobic matrix. It is likely that cholesterol could impart a synergic effect on caveolin-1 induced caveolae formation in vivo. Our data supports a model by which cholesterol drives caveolin-1 penetration deeper into a hydrophobic matrix. This would have a strong impact on the structure of the bilayer as the omega-shaped intramembrane domain is driven deeper in the bilayer, disrupting lipid packing to a much greater extent than a shallowly buried protein. This postulation is in line with studies that have shown caveolae flattening caused by cholesterol depletion, and with studies that have suggested a wedge-like mechanism of caveolin-1 induced membrane curvature [34, 35].
By probing each tryptophan individually, we were able to obtain a high degree of resolution which allowed the individual characterization of the regions flanking P110. Clearly, proline is important in controlling the depth of the N-side region of the protein, as evidenced by the large changes in aqueous accessibility of W85 and W98. On the other hand, the C-side tryptophan residues were much less altered by the mutation of proline to alanine. The observed changes in aqueous accessibility indicate that proline 110 is important for governing the hydrophobic coverage of the N-side region of Cav162–178. This result is remarkable considering that a single amino acid would be able to modulate the solvent exposure of a significant portion of the protein without major changes in side chain hydrophobicity (i.e. both proline and alanine are hydrophobic amino acids), and this highlights the biological conservation of proline at the 110 position. Clearly a better understanding of the role of P110 has led to new insights into its function in governing caveolin-1 behavior.
This work was supported by NIH R01 GM093258-01A1 awarded to K.J.G. We would like to thank Carolyn Sivco for her laboratory assistance. We would like to thank Dr. S.M. Kelly of the University of Glasgow for helpful advice regarding circular dichroism spectroscopy. We would like to thank Sarah Plucinsky for critical reading of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Kyle T. Root, Department of Chemistry, Lehigh University, 6 E. Packer Ave, Bethlehem, Pennsylvania 18015, USA.
Kerney Jebrell Glover, Department of Chemistry, Lehigh University, 6 E. Packer Ave, Bethlehem, Pennsylvania 18015, USA. Tel.: 1 (610) 758-5081, Fax: 1 (610) 758-6536, Email: ude.hgihel@602gjk.