There is a disagreement between the weak oligomerization of the MS1 peptide in mild detergents [17
] and its apparently very strong oligomerization under the denaturing conditions of SDS-PAGE [13
]. We thus synthesized the peptides shown in and labeled them with dye molecules to study the nature of peptide-peptide interactions in SDS solution and in SDS gels using Forster Resonance Energy Transfer (FRET). MS1 was originally engineered from the GCN4 transcription factor coiled-coil dimerization domain to be a Membrane Spanning α-helix with a 20 residue hydrophobic segment containing a leucine zipper-like dimerization motif and a native asparagine residue at the 12th
position in the 20 residue segment, or 14th
position overall. A control peptide with leucine at the 14th
position is monomeric [13
]. As shown in , we use a naming scheme for all three sets of peptides that is based on the original MS1 sequence in which the 14th
residue overall is either asparagine (N14) or leucine (L14). The three sets of peptides all contain the MS1 sequence, and two of them have additional basic residues added to the N-terminus to promote ease of handling and purification. As shown below, the three sets of Asn/Leu peptide pairs behave similarly in all experiments.
Polyacrylamide Gel Electrophoresis
Peptide mobility on polyacrylamide gel electrophoresis is frequently used to determine oligomer states of TM helical peptides [6
]. We assayed N14 and L14 peptides for their apparent molecular weights using several types of peptide gels; 10–20% gradient Tris/Tricine SDS gels (BioRad) and 16.5% Tris/Tricine SDS gels (BioRad) developed in BioRad SDS running buffer, and NuPAGE® Novex 4–12% Bis-Tris Gel developed in NuPAGE® MES SDS Running Buffer (Invitrogen). For precise molecular weight standards in the appropriate range we performed a partial crosslinking of a peptide fragment of the colicin E3 protein binding domain [31
], which is 3800 Da, to produce bands of 3800, 7600 and 11,400 Da. These markers migrated at their appropriate molecular weights when compared to commercially available marker sets. Unlabelled L14 migrates at its molecular weight of about 3100 Da. The addition of fluorescent dyes, or the addition of the K4
segment to the L14 sequence shown in reduced mobility of the TM peptides on the gel slightly, as expected from the increased molecular weight. Most importantly, replacement of the 14th
position leucine by asparagine, which changes molecular weight by a single mass unit, decreased peptide mobility substantially. In we show a coomassie-stained gel in which pairs of lanes have a 40 fold difference in the amount of N14 or L14 peptide. The left two sample lanes contain 0.5 μg (5 μM in the well) of either L14 or N14 peptide, while the right two sample lanes contain 20 μg (200 μM in the well) of either N14 or L14 peptides. Bands containing 20 μg of peptide ran slightly slower. However in both cases, the N14 peptide runs at an apparent MW that was about twice the apparent MW of L14. In we show a gel with various N14 and L14 peptide pairs. We found an apparent MW ratio ranging from 1.8–2.4 (N14/L14) in roughly 25 side-by-side comparisons such as the one shown in . The median apparent MW ratio was 2.0. Unlike other published examples of TM helix dimerization [32
], no evidence of a concentration effect and no evidence of monomer-oligomer coexistence was observed in any gel electrophoresis experiment we performed. N14 peptides appeared to be 100% dimeric under all conditions, suggesting a very strong interaction.
Anatomy of an SDS-PAGE experiment
A peptide in an SDS-PAGE experiment experiences a range of possible environments during the course of the electrophoresis. With respect to SDS concentrations samples are initially dissolved in a “loading” buffer with at least 2% (70 mM) SDS. The gels themselves are made in a “casting” buffer which typically contains 0.2% (7 mM) SDS while the “running” buffer that bathes the gel and conducts current contains 0.1 % (3.5 mM) SDS.
Peptide-detergent complexes form in the loading buffer. For the gel in , about 30 μl of sample containing about 600 μg (70 mM) of detergent and 5 μg (50 μM) peptide in buffer was loaded into a well and an electric field was applied, causing the detergent and peptide to move into the top layer, or “stacking” layer of the gel. The stacking layer is a low crosslink density polyacrylamide gel matrix which allows for less restricted migration of charged molecules. Peptide and detergent are compacted into a very thin band at the intersection of the stacking gel and the more highly crosslinked separating gel matrix. We estimate this compacted band to be as much as 20 fold more concentrated than the original sample which means that the detergent concentrations are on the order of 1 molar and the peptide concentrations are on the order of 1 millimolar as the sample enters the separating gel.
During the separation phase, the smaller peptide-free micelles will presumably electrophorese into the gel faster than the larger peptide/detergent complexes, creating an environment in the gel with an intermediate detergent concentration. Strong peptide-detergent interactions in the sample will cause the peptide to co-migrate with detergent bound to it. Because we do not know the exact SDS concentration sampled by a peptide during the course of an SDS-PAGE experiment, in this work we explore the self association of the peptides in in buffers that encompass the possible SDS concentration range: in loading buffer with 2% (70 mM) SDS and in gel running buffer with 0.1 % (3.5 mM) SDS. We also measure interactions in the polyacrylamide gel itself.
Peptide Secondary Structure in SDS Solution
Hydrophobic peptides such as TM helices are frequently helical in SDS solutions, however this is not always the case. Some methods of sample preparation give peptides with some β-sheet character, which can lead to insolubility and unstable solutions. In part, this is because dried peptides are typically β-sheets and will remain so unless completely dissolved. We have found that dissolving the peptide in hexafluoroisopropanol (HFIP) or HFIP/water promotes helical structure and once dried from HFIP, the peptides remain helical and dissolve readily in detergent solutions [34
]. In we show CD spectra for various N14 and L14 peptides in SDS solution, either 2% (70 mM) or 0.1 % (3.5 mM) dissolved using the methods described above. All variants, whether labeled or unlabelled, are apparently fully helical and have the same secondary structure. Throughout the course of the experiments described below we routinely used CD spectroscopy to verify helical secondary structure in detergent solutions.
FRET in SDS Gel Loading Buffer
In this work, we set out to characterize the association of N14 peptides in SDS using Forster resonance energy transfer (FRET). The peptides in were labeled on their amino-termini using succinimidyl esters of the dyes Fluorescein (FL) (λexc= 490 nm, λem=540 nm), TAMRA (λexc=540 nm, λem=580 nm), CY3 (λexc=550 nm, λem= 565 nm), CY5 (λexc=580, λem=675), bodipy fluorescein (BFL) (λexc=500 nm, λem= 515 nm) and Rhodamine-X (ROX) (λexc=580 nm, λem= 610 nm). These six dyes give five Acceptor-Donor pairs for FRET: Fluorescein-TAMRA, Fluorescein-CY3, CY3-CY5, BodipyFL-ROX and TAMRA-ROX. R0 values for these pairs are in the range 45-55 Å. Because the peptides are helical in SDS and there are either 14 or 18 residues (21 or 27 Å) between the amino terminus and the asparagine residue, parallel dimers will have dye moieties less than 25 Å apart while antiparallel dimers are expected to have dye moieties less than 40–50 Å apart. In either case, dimerization will give rise to FRET which should be easily measurable using these dye pairs.
In a typical FRET experiment, we prepared three parallel samples and collected six spectra. One sample had donor
only (D), one had acceptor
only (A) and one had donor
(D+A). Three FRET spectra were collected with excitation at the donor
excitation wavelength and three were collected with excitation at the acceptor
excitation wavelength. The latter spectra (not shown) were used to verify the concentration and solubility of acceptor
peptides in each sample. Comparison of the emission intensity of the donor
in the donor-only sample with the donor
intensity in the donor
sample reports on acceptor
dependent quenching that might occur even in the absence of classical FRET. The acceptor
only sample allows for the measurement of acceptor
emission that comes from “direct excitation”; the emission of the acceptor
that occurs when it is excited directly at the donor
excitation wavelength. Comparison of the acceptor
emission intensity in the acceptor
only sample with the donor
sample reports on classical FRET. In particular, any enhancement of acceptor
emission over direct excitation is due to FRET. Thus, If classical FRET is occurring, the D + A sample will show a loss of donor
fluorescence and a concomitant gain in acceptor
]. There are reports that suggest fluorescein and TAMRA do not always undergo classical FRET [36
], but instead can show proximity by donor
quenching without enhanced acceptor
emission. Furthermore, comparison of donor
intensity between N12 and L12 samples also reports on donor
quenching in N12 samples. Thus, by comparing the spectra collected for N14 and L14 for each experiment we can detect dimerization whether it is causes only donor-quenching or classical FRET.
Examples of FRET spectra collected for 1:1 ratios of TA-N14+FL-N14 and for TA-L14+FL-L14 in gel loading buffer are shown in . In the mixed sample, the fluorescein emission of the donor was very similar to the donor alone and the minor amount of emission of the acceptor TAMRA at 580 nm was accounted for by the small direct excitation peak. Furthermore the N14 donor emission intensity was very similar to the L14 donor emission intensity. In all samples, we conclude that the FRET efficiency for N14 peptides in SDS loading buffer was less than 5%. The lack of detectible FRET and the observation that L14 and N14 were identical shows that there was less than 5% dimerization of N14 in this experiment, which was performed under loading conditions of an SDS-PAGE gel 2% (7-0 mM) SDS. The three spectra collected while the acceptor was excited are not shown, but in all cases these measurements verified that the acceptor was present at the expected concentration in all samples.
Figure 4 Fluorescence of peptide mixtures in gel loading SDS buffer containing 2% (70 SDS. Samples containing a 1:1 mixtures of Fluorescein and TAMRA labeled peptides were dried from HFIP stock solutions and dissolved at 1 μM each in 2% (70 mM) SDS in (more ...)
To verify the observation that FL-TAMRA labeled N14 peptides do not give FRET in 2% SDS, and to eliminate dye-specific effects as possible factors, we repeated FRET experiments in gel loading buffer with N14 peptides labeled with two other dye pairs: Bodipy Fluorescein(D) + Rhodamine-X (A) and TAMRA(D) + Rhodamine-X (A). The results, in show no detectable FRET for these other dye pairs. In all cases the D+A sample spectrum was essentially the sum of the donor-only and acceptor-only spectra.
Figure 5 FRET experiments with other dye pairs in gel loading buffer containing 2% (70 mM) SDS. Panel A: Mixtures of N14 peptides labeled with TAMRA(D) + Rhodamine-X (A) dissolved in gel loading buffer. Panel B: Mixtures of N14 peptides labeled with or with bodipy-fluorescein(D) (more ...)
If the N14 peptides are oligomeric in SDS, addition of unlabelled peptide will compete for labeled peptide in the oligomers and decrease quenching or FRET [16
]. In , we show that the addition of unlabelled peptide has no effect on fluorescence indication that the N14 peptides were monomeric. This conclusion is corroborated by the observation that the intensity of donor
fluorescence in the donor-only N14 samples was always the same as the intensity of the donor-only L14 samples (). Thus no significant donor-donor self-quenching is taking place. Given the propensity of fluorescein to self-quench [38
] this observations supports our conclusion that N14 peptides are not dimerizing in SDS.
FRET in SDS at high peptide concentration
We hypothesized that the apparent dimerization of N14 peptides in gel electrophoresis experiments may be due to the relatively high peptide concentration in the gel (5–50 μM in the starting sample) compared to FRET experiments in solution where 1 μM peptide was typical. To test this hypothesis we conducted FRET experiments in gel loading buffer with 50 μM total peptide instead of 1 μM. The results are shown in . To eliminate inner filter artifacts [40
] due to the high optical density of the samples, we measured fluorescence in a glass capillary with a path length less than 0.5 mm. A significant amount of FRET was observed in this experiment; however the FRET was identical for N14 and L14 peptides, indicating that it was not due to a specific interaction between N14 peptides. Instead, the observed FRET is likely due to random proximity effects [28
] as described below, which are expected to be significant at these peptide/detergent ratios (1:1000). We note here that peptide:detergent ratio in this experiment was lower than the range at which N14 peptides were shown to interact in other detergents by other techniques [13
] but higher than the typical ratio in an SDS gel experiment. Importantly, this experiment also provides a positive control for the signature of FRET between fluorescein and TAMRA under our experimental conditions. For FRET pairs with large R0
> 45 Å) FRET will occur due to random co-localization of the freely diffusing acceptor
] when peptide to detergent ratio exceeds about 1:1000. We and others have developed algorithms to simulate or calculate FRET from random colocalization in two dimensional systems, which have been used successfully to correct FRET measurements in membranes [41
]. If we consider an SDS solution to be a continuous (i.e. rapidly exchanging) two-dimensional system, then we can estimate FRET using these published methods. A peptide:SDS ratio of about 0.0008 (50 μM peptide 60 mM micellar SDS) leads to a predicted FRET of about 20% from random colocalization [28
]. This is similar to what we observed. The fact that N14 and L14 peptides gave identical FRET shows that random colocalization was responsible for the observed FRET, rather than a sequence-specific interaction between N14 peptides.
Figure 6 FRET experiments with high peptide concentration in gel loading buffer containing 2% (70 mM) SDS. FRET experiment with TAMRA and fluorescein-labeled peptides at 1:1 Acceptor:Donor and 50 μM total peptide concentration in gel loading buffer. Measurements (more ...)
FRET in Gel Running Buffer
We hypothesized that the specific components of the gel running buffer (0.1%, (3.5 mM), SDS, 100 mM Tris, pH 8.5) were driving N14 peptide dimerization in the gels
, but not in the loading buffer which has much higher SDS concentrations. This idea is consistent with the observation that these peptides associate measurably only at high peptide:detergent ratios in other detergents [17
]. Therefore, we performed FRET experiments in Biorad gel running buffer using the Fluorescein-TAMRA donor-acceptor pair. By circular dichroism, we found that all peptides studied () were fully alpha-helical and stable for at least a day despite the fact that the SDS concentration in the running buffer (3.5 mM) is near the lower end of the reported CMC values for SDS. Presumably the peptides can bind enough SDS to be soluble and helical under these conditions. We performed these experiments using all three versions of N14 and L14 () labeled with fluorescein (donor) and TAMRA (acceptor). To increase the stringency of the experiment for detecting dimerization, these experiments were conducted at acceptor:donor ratios of 10:1. In this case, roughly 90% of the donors will have an acceptor
as partner if dimerization occurred and was random. For these experiments the contribution of direct excitation of the acceptors was larger as shown in . Although there were subtle differences in the shape and position of the peaks depending on the N-terminal sequence of the peptides, comparison of the equivalent N14 and L14 peptides showed no acceptor
emission greater than the direct excitation and no quenching of donor
fluorescence in any of the samples. Furthermore, the N14 and L14 peptides give almost identical intensities corroborating the conclusion that there is no donor
quenching in N14 peptides and thus there is no dimerization.
Figure 7 FRET experiments in SDS gel running buffer containing 0.1 % (3.5 mM) SDS. Pairs of fluorescein and TAMRA labeled peptides () were characterized at 1 μM donor and 10 μM acceptor peptide. We show spectra for donor only (D), acceptor (more ...)
Although peptides were always premixed before dissolution in SDS, we also considered the possibility that equilibration is slow to occur. This is consistent with the observation that other dimerizing peptide systems show monomer and dimer bands in the same gel [10
]. To assess the equilibration rate, we measured FRET and ran PAGE gels before and after sample boiling, which should accelerate equilibration, but observed no changes in any of the experiments. We also monitored FRET in such samples for time periods ranging from minutes to days after sample preparation. No time-dependent changes were ever observed.
We show in that the lack of FRET or donor quenching did not depend on the particular dye pairs used. In addition to the fluorescein-TAMRA pair discussed above, we also made measurements in running buffer (3.5 mM), using CY3(D)+CY5(A) () and fluorescein(D) and CY3(A) (not shown). No evidence of dimerization was observed in gel running buffer with any other dye pair.
Figure 8 FRET experiments in SDS running buffer containing 0.1 % (3.5 mM) SDS. These experiments were done with CY3(D) and CY5(A) labeled peptides. Measurements were made on mixtures of CY3 and CY5 labeled N14 peptides at 1:1 D:A ratio (Panel A) and 1:10 D:A ( (more ...)
In situ FRET in polyacrylamide
Based on the disagreement between the gel electrophoresis experiments, which indicated that N14 peptides were completely dimeric in SDS gels under all conditions, and the FRET experiments which failed to detect any dimerization in SDS in any experiment, we hypothesized that some unknown factor in the electrophoresis experiment itself was driving dimerization of asparagine-containing TM helices in the gels. To test this idea, we electrophoresed 1:1 mixtures of fluorescein- and TAMRA-labeled peptides on SDS gels and then performed FRET measurements in situ on the bands from the gel by cutting them out and immediately placing the intact gel slabs in a cuvette containing gel running buffer for spectroscopy. Peptides were not extracted from the gel, but rather were characterized immediately in situ using fluorescence spectroscopy on the intact band in the gel. In control experiments, we found that physical factors such as slab size and shape, angle to excitation beam and physical imperfections in the gel changed the fluorescence intensity from the sample, presumably through light scattering and sample illumination effects, but did not change the band shape, spectral characteristics, or the apparent amount of FRET. We tested the stability of the in situ band in the gel by cutting out control bands and soaking them in gel running buffer while monitoring the peptide remaining in the gel. Peptide diffusion out of the gel into the solution was negligible for at least several hours after the band was cut. Fluorescence measurements were always made within 15 minutes of cutting the band from the gel. Therefore, in this experiment we were measuring FRET between peptides exactly as they ran in the gel.
In a typical experiment, we loaded of peptide solutions of 5–50 μM in the well. We estimate that the final concentration of peptide in the band in the gel was similar to the starting concentration of peptide within a factor of two. Typical in situ FRET measurements are shown in for samples prepared with 10, 20 and 40 μM peptide. In the 10 μM experiment there was little or no FRET in excess in the N14 band over the direct excitation peak. The N14 and L14 spectra were very similar, indicating that no dimerization was taking place, even though the N14 band migrates at the molecular weight of a dimer. In the high concentration (40 μM) experiment some FRET observed for both L14 and N14 samples. This is shown by the TAMRA emission at 580 nm that was in excess of the direct excitation peak. However the amount of FRET was always similar for the mixtures of FL-L14 + TA-L14, which ran as monomers in the gel, and mixtures of FL-N14 and TA-N14 which ran as apparent dimers. FRET from random colocalization is expected to occur at these peptide concentrations (see above) and this accounts for the observed FRET. We note that N14 migrates at exactly the same dimer molecular weight at every peptide concentration studied, and shows essentially no in-situ FRET, except at the highest concentrations where proximity FRET takes place. Most importantly, in all cases, the amount of in situ FRET from N14 “dimer” bands was always similar to the amount of FRET from L14 “monomer” bands. Thus, we conclude that the N14 peptides in the gel, which always migrate dramatically slower than L14 and have apparent molecular weights about two times larger than L14, are actually monomeric.
Figure 9 In situ FRET experiments in SDS-PAGE gels. Mixtures of 1:1 Fluorescein and TAMRA labeled N14 or L14 peptide were dissolved in gel loading buffer and electrophoresed into a Biorad 16.5% SDS/Tris/Tricine gel at 80 V for 120 minutes. Panels A and B show (more ...)
We also tested for interactions between N14 peptides using a high throughput screen we developed for helix-helix interactions. In this screen, which we have previously described [30
], polystyrene/polyethylene glycol polymer beads with tethered peptide sequences are incubated with dye-labeled target peptides in detergent solution. Interaction between tethered and soluble sequences cause the dye labeled peptide in solution to accumulate on the bead, which is visible by its fluorescence. We have previously shown that this is a sensitive assay for helix-helix interactions [30
], however no interaction was detected between dye-labeled N14 peptides in detergent solution and N14 sequences on beads. Experiments were performed in various SDS concentrations including loading and running buffers. We never observed evidence that any dye-labeled N14 peptide in detergent solution interacted specifically with any bead-tethered N14 peptide.
Interactions between peptides and SDS micelles
Finally, we hypothesized that the apparent changes in N14 molecular weight in SDS gels, compared to L14, were due to differences in how the peptides interact with the detergent rather than how the peptides interact with each other. To test this idea we performed a dynamic light scattering experiment on mixtures of 20 mM SDS and L14 and N14 peptide concentrations up to 400 μM peptide. About half of the 20 mM SDS will be micellar. Given that the aggregate number of SDS is ~100, this experiment encompasses nominal peptide:micelle ratios ranging up to about 4:1. In the absence of peptide, the SDS solution contained particles that were roughly 3.5 nm in diameter, consistent with the expected radius of SDS micelle (). As peptide concentrations were increased, the average particle size increased. When there was about 1 peptide per micelle the particle size was twice as large for both peptides. Above this concentration range, the particles size in increases dramatically, with a 50-fold increase observed for L14/SDS between 1 peptide per micelle and 4 peptides per micelle. The individual micelles were probably not becoming 50 fold larger, but instead were probably aggregated together into super aggregates, perhaps bridged together by the hydrophobic L14 peptide. In the presence of N14 peptides the increase in micelle size was much less dramatic. At 4 peptides per micelle the N14/SDS particles were only 1/20th as large as the L14/SDS particles. Although we do not know exactly how the peptides affect the SDS micelles, why the particles sizes were so different, or why this difference affects PAGE migration the way it does, this experiment clearly demonstrates that there is a significant difference between the interactions of N14 with SDS compared to the interaction of L14 with SDS.
Figure 10 Dynamic light scattering measurements of SDS/peptide solutions. A solution of 20 mM SDS in Tris buffer, pH 8.4 was mixed with an identical solution also containing 400 μM N14 or L 12 peptide to create solutions with increasing peptide/detergent (more ...)