|Home | About | Journals | Submit | Contact Us | Français|
The peroxisome proliferator-activated receptor is a member of the nuclear receptor superfamily of transcriptional regulators. Regulation of the nuclear receptors occurs through changes to the structure and dynamics of the ligand-binding domain. Therefore, the need has arisen for a rapid method capable of detecting changes in the dynamics of nuclear receptors following ligand binding. We recently described how solution-phase amide hydrogen/deuterium exchange (HDX) provides a biophysical technique for probing changes in protein dynamics induced by ligand interaction. Building from this platform, we have optimized the robustness of the differential HDX experiment by minimizing systematic errors, and have increased the efficiency of the chromatographic separation through the use of high-pressure liquid chromatography. Using knowledge gained previously from comprehensive HDX experiments of PPARγ, a modest throughput method to probe changes in the dynamics of key regions of the receptor was developed. A collection of ten synthetic and endogenous PPARγ ligands were characterized with this new method requiring approximately 24 h of analysis. This is a dramatic improvement over the 10 d of analysis that would have been required with our previous approach for comprehensive differential HDX analysis. In addition to demonstrating the utility of this approach, the study presented here is the first to measure changes to the dynamics of PPARγ upon the binding of putative endogenous ligands.
Nuclear receptors (NRs) are transcriptional regulators implicated in development, metabolism, cardiovascular function, inflammation, and circadian rhythms.1–3 These multidomain proteins typically contain a conserved ligand-binding domain (LBD) and a highly conserved DNA-binding domain (DBD). It has been well documented that many receptors within this superfamily respond to small-molecule ligands. The ability to alter transcriptional programs through small-molecule ligand binding to NRs provides opportunities for therapeutic intervention of disease. Thus, NRs have been the focus of intense drug discovery efforts, which have led to novel treatments for diabetes, cancer, and cardiovascular disease, and new forms of hormone replacement therapy.2,4,5 For example, the small-molecule fibrates are approved drugs for the treatment of hypertriglyceridemia, and these molecules target the lipid-sensing nuclear receptor PPARα.4,5 The thiazolidindiones (TZDs or glitazones) are small-molecule ligands that bind and activate the gamma isoform of PPAR, and these drugs have been shown to be clinically useful in the treatment of insulin resistance. Unfortunately, the use of TZDs have been associated with dose-dependent increases in body weight, plasma volume expansion and edema, and possibly increased cardiovascular risk.6 Although these unwanted effects limit the utility of PPARγ agonists in chronic treatment of insulin resistance, several PPARγ agonists remain on the market and several more are in development. Dissecting the beneficial effects of PPARγ agonists from the undesired effects will likely require a better understanding of the mechanism of ligand-dependent activation of the receptor.
There are three major isoforms of PPAR—α, γ, and δ7—as well as many other sequence variants.4,5 Common to most NRs, PPARs are comprised of an N-terminal A/B domain containing the autonomous transcriptional activation function AF-1; a highly conserved zinc-finger DNA-binding C domain, which drives sequence-specific DNA recognition; a less conserved D domain hinge region; and the ligand-binding E domain (LBD). PPAR lacks the highly variable C-terminal F domain. PPAR forms herterodimers with the retinoid X receptor (RXR) and binds to promoter regions of target genes that contain specific PPAR response elements (PPREs). Upon ligand binding, the PPAR/RXR heterodimer complex dissociates transcriptional corepressor proteins such as N-CoR or SMRT, allowing interactions with coactivator proteins such as the p160 family of steroid receptor coactivators (e.g., SRC-1), CBP, and p300. Coactivators have intrinsic chromatin remodeling activity that facilitates recruitment of basal transcription machinery (e.g., RNA polymerase II) to initiate transcription.
The X-ray structure of the PPARγ LBD was described in 1998, and it is structurally related to other NR LBDs, consisting of 12 helical regions and 1 β-sheet region.8,9 The PPARγ ligand binding pocket surface is large (~1200 Å2), and it has been shown that potent full agonists of the receptor sit in the pocket in such a way as to facilitate direct contact with helix 11/12. Stabilization of H11/12 has been implicated in facilitating recruitment of coactivators such as SRC-1.
The term selective modulators of nuclear receptors refers to ligands that regulate transcriptional programs resulting in a desired phenotype. An example of this strategy was the development of selective estrogen receptor modulators (SERMs) such as the drug raloxifene,10,11 which has beneficial estrogen effects including maintenance of bone density while exhibiting anti-estrogen effects in other tissues (e.g., minimizing uterine proliferation). In the case of PPARγ, a selective PPARγ modulator (SPPARM) would be a potent insulin sensitizer that would not induce weight gain or increase adipocity and plasma volume. One such example is the PPARγ ligand MCC-555. This TZD PPARγ modulator was shown to be a potent antidiabetic compound with a tenfold lower binding affinity for PPARγ than the TZD full agonist rosiglitazone (BRL49653).12,13 The degree of PPARγ agonism induced through binding of MCC-555 was shown to be context specific; in certain cases, MCC-555 functioned as a full agonist, and in other cellular systems the ligand behaved as a partial agonist or antagonist. In coactivator recruitment assays, MCC-555 exhibited a 10–50-fold reduction in the potency for steroid receptor coactivator-1 (SRC-1) recruitment when compared to rosiglitazone. More recently, a number of non-TZD PPARγ partial agonists have been described. Fmoc-L-leucine (F-L-Leu) is a unique PPARγ ligand characterized by the fact that two molecules F-L-Leu bind to each PPARγ molecule.14 The compound was shown to improve insulin sensitivity in db/db mice but with lower adipogenic activity as compared to rosiglitazone. F-L-Leu demonstrated differences in coactivator recruitment as compared to rosiglitazone.
The recurring observation of coactivator recruitment and the selectivity of PPARγ modulators reinforces the theory that structural changes within the PPARγ LBD induced by ligand binding result in differential recruitment of coactivators directly leading to changes in gene expression patterns; this would account for the retention of insulin sensitization, but reduced adipogenesis of SPPARMs. In support of this theory, results of a protease protection assay indicated a change in stability of the LBD between PPARγ bound with F-L-Leu, and rosiglitazone.14
Berger and co-workers described a non-TZD PPARγ partial agonist, referred to as nTZDpa.12 The compound was a tight binder of PPARγ (Ki ~ 10 nM) and was a potent insulin sensitizer, yet demonstrated weak transactivating potential in the context of a specific cellular reporter assay. The compound showed selectivity for PPARγ over α and δ and was able to antagonize the activity of γ full agonists. Importantly, the work described the first direct measurement of a change in the dynamics of the receptor LBD between PPARγ bound to a full and a partial agonist. Solution-phase NMR spectroscopy demonstrated that rosiglitazone and nTZDpa both stabilize PPARγ-LBD with respect to Apo-LBD, with rosiglitazone providing the most stable conformation, or subset of conformers. More recently, the same group described two compounds of particular interest,15 regioisomeric benzoyl-2-methyl indoles that differ in structure by the position of a single lactic acid group (ortho vs. meta). This subtle structural difference of the ligand is enough to switch the functional activity of the molecule from full to partial agonism.
It is important to note that partial agonists of PPARγ exhibit the SPPARM phenotype. Thus, profiling PPARγ ligands for partial agonist behavior might prove useful in the identification of novel SPPARMs. Unfortunately, the classification of PPARγ ligands as full or partial agonists remains somewhat subjective. One common approach is to determine the efficiency of cellular transactivation potential of a PPARγ ligand using a reporter gene assay (e.g., luciferase), comparing the activation of the reporter gene as a function of ligand relative to the full agonist rosiglitazone and presenting a percent transactivation activity (%TA). Full agonists are generally characterized as having transactivation efficiencies greater than ~80%, while partial agonists activate to less than 40% of the values obtained for rosiglitazone. Intermediate agonists, such as the non-TZD ligand BVT.13,16 have %TA potential somewhere between these values (~60 %TA). Cell-based transactivation of NR genes provides a good indication of activity, but is relevant only to the specific cell line used; due to the importance of expression of the appropriate coactivators; as is obvious from the discussion above, classification based on %TA is subjective. Profiling the coactivator recruitment preference of a specific ligand might also provide useful information, but these assays could be biased due to the potential for posttranslational regulation of coactivators. High-resolution analysis of the binding mode of the ligand in the binding pocket of the receptor may provide predictive information, but the approach is extremely labor intensive and requires that the ligand-receptor complex crystallizes. An alternative approach would be direct measurement of the dynamics of PPARγ LBD upon ligand binding for functional classification of ligands.
As discussed above, ligand-dependent activation of NRs may involve changes to the global structure and dynamics of the receptor. NMR and X-ray crystallography provide the highest resolution structural information and are ideally placed to detect large changes in protein structure. Although these global changes in the structure of the receptor can be measured to within a few angstroms, subtle changes to the dynamics of the receptor and changes to the distribution of protein conformations may not be detected.
Solution-phase hydrogen/deuterium exchange (HDX) mass spectrometry (MS) can be used to study protein stability, dynamics, and protein ligand interactions.17–33 HDX data often complement X-ray or NMR structures, and can provide information in cases where no structure exists. The extent to which each amide hydrogen is involved in hydrogen bonding reveals much about its environment. Changes in dynamics induced by ligand binding perturb hydrogen-bonding networks that are reflected in the measured rate of amide hydrogen exchange. The technique offers flexibility in measuring exchange kinetics under a wide range of solution conditions and protein concentrations (>1 μM). Typically, HDX analysis involves exposing the protein to a deuterated environment (referred to as on-exchange) for predefined time periods followed by immediate addition of a quench buffer (non-deuterated, low-pH, and low-temperature buffer) containing denaturants to significantly slow further on-exchange of deuterium and to minimize D/H back-exchange with non-deuterated solvent. Deuterated protein is then digested with acid-stable proteases under quench or slow-exchange conditions. The kinetics of HDX in specific regions of the protein can then be determined by liquid chromatography electrospray ionization mass spectrometry (LC-ESI-MS) by measuring the increase in number-average m/z values of peptide ion isotopic distributions for each peptide. To probe protein-ligand interactions, these experiments are performed on both the free protein (apo protein) and the protein-ligand complex. The difference in area under curve (AUC) plots of deuterium uptake versus on-exchange time period for the two experiments (apo versus ligand bound) reveals regions in the protein that possess differential exchange kinetics as a result of ligand interaction.
Yan and colleagues have utilized HDX to probe the conformation of the retinoid X receptor alpha LBD (RXRα LBD) in the presence and absence of 9-cis-retinoic acid (9-cis-RA).31 In those studies, perturbation in HDX of RXRαLBD upon 9-cis-RA binding correlated well with changes in hydrogen bonding, amino acid residue depth, and solvent exposure predicted from the co-crystal structure.34 However, perturbation in exchange kinetics was also observed in regions not predicted from the X-ray structure. That work highlights the complementary relationship between HDX and x-ray crystallography and it demonstrates the utility of HDX for probing drug interactions with nuclear receptors.
Previously, we have described a fully automated HDX system to probe conformational changes induced by ligand binding to PPARγ LBD.35–37 This work demonstrated that we could detect changes in amide exchange rate upon binding of full and partial agonists of PPARγ. Subsequent work compared the HDX data for a set of full and partial agonists with the corresponding X-ray structures (Structure, in press). It was demonstrated that ligand binding altered the dynamics of the receptor without significant changes to the overall structure of PPARγ LBD as measured with X-ray crystallography. Although HDX data has been acquired for the binding of synthetic ligands to PPARγ, no data has been reported for endogenous ligands such as prostaglandin J2 and oxidized lipids. After our initial HDX studies examining ligand binding to PPARγ, a number of improvements to the HDX experiment have been reported in the literature, such as the demonstration of the utility of sub-2-μm stationary phases in HDX MS experiments.38
In the work described here, our initial focus was on the reduction of systematic errors in differential HDX experiments. Improvements to the efficiency of the chromatography step via the use of HPLC columns packed with sub-2-μm particles have been incorporated into our automated HDX system. Subsequent to these improvements, we developed a two-step HDX method that can be used to probe ligand-induced changes to the dynamics of the PPARγ ligand-binding pocket. Five comprehensive differential HDX experiments were interrogated to identify a single on-exchange time point that reflects the maximum difference between apo- and ligand-bound HDX rates. A higher throughput, single-time-point, HDX method was then developed. In this current configuration of the experiment, less than 3 h of instrument time is required to complete the entire differential HDX experiment, as compared to approximately 1 d with our previous approach. The analysis of six synthetic and four endogenous ligands was completed in approximately 24 h. To obtain the equivalent data from a comprehensive experiment would have required around 10 d of instrument time. In addition, this work describes the first HDX data of a nuclear receptor bound to endogenous ligands. Based on our preliminary data for the estrogen receptor (ER), liver X activated receptor (LXR), and protein kinases, this two-step approach appears to be applicable to the characterization of other protein-ligand interactions.
PPARγ-LBD was purified in house. The PPARγ ligands MRL-20, MRL-24, and BVT.13 were synthesized in house using previously described chemical strategies.12 GW1929 and 15-deoxy-prostaglandin-J2 were purchased from Sigma (St. Louis, MO). MCC-555, 9(S)-HODE, 13(S)-HODE, and 15(S)-HETE were purchased from Cayman Chemical (Ann Arbor, MI). Rosiglitazone was purchased from ChemPacific Corp., MD. Each ligand was prepared as a 10 mM stock in DMSO.
Protein dilution buffer (H2O or D2O) was 20 mM Tris-Cl (pH 7.9), containing 100 mM KCl, and 1 mM DTT. Quench solution was 3 M urea in 0.5% aqueous trifluoroacetic acid (TFA). All other chemicals and solvents were purchased from Sigma (St. Louis, MO). For protein:ligand experiments, protein stock solutions (10 μM) were incubated with ligand solution (200 μM) for 3 h at 25°C. Solution-phase HDX was performed with a fully automated system constructed around a LEAP Technologies Twin PAL HTS autosampler (LEAP Technologies, Carrboro, NC) interfaced with a linear ion trap mass spectrometer (Thermo Finnigan, San Jose, CA).36 Four microliters of a 10 μM protein solution was diluted to 20 μL with D2O dilution buffer and incubated at 25°C for the following periods of time; 1 sec, 30 sec, 60 sec, 900 sec, and 3600 sec. Following on-exchange, unwanted forward or back exchange was minimized by dilution to 50 μL with quench solution (held at 1.5°C). Sample was then passed across an immobilized pepsinogen column (prepared in house39) at 200 μL/min (0.1% TFA, 2°C), and the resulting peptides were trapped onto a C8 trap cartridge (Microm Bioresources, Auburn, CA). Peptides were then gradient eluted (4% CH3CN to 40% CH3CN, 0.3% formic acid over 20 min, 2°C) across a 50 mm × 2.1 mm analytical column (3 μm Hypersil Gold, ThermoFisher Scientific) directly into the mass spectrometer.
For screening experiments, the on-exchange time was 30 sec. Each experiment was the average of four replicates, acquired in parallel. Each ligand was acquired and compared to its own apo dataset. The trap column was from Agilent (Zorbax Eclipse XDB-C8, Agilent, Santa Clara, CA) and the analytical column was from ThermoFisher (50 mm × 2.1 mm, 1.9 μm Hypersil Gold). The gradient time was 10 min with a flow rate of 250 μL/min; the back pressure was around 320 Bar.
Raw data files were converted to MxZML format and imported into a LIMS system (ProteusLIMS, GeneoLogics, Victoria, BC, Canada). Data were processed with in-house software40 and Microsoft Excel, and visualized with PyMOL (DeLano Scientific, Palo Alto, CA; http://pymol.sourceforge.net/).
Previously we described a fully automated system for HDX data acquisition.35,36 The introduction of robotic liquid handling and computer control of all timing events increased the precision of the HDX experiment and reduced the probability of gross errors. The increased throughput, compared to manual sample preparation, allowed us to determine the average HDX value from four replicate experiments. We have also developed a fully integrated software solution that facilitates data archiving and calculation of HDX kinetics.40 With a fully automated HDX platform developed and integrated software in place, we are uniquely positioned for reasonable-throughput, high-precision HDX analysis of proteins and protein complexes.
Our previously described experimental scheme operated in a linear fashion. Data from sample A were acquired immediately prior to data from sample B. Although the results were very precise, it became apparent that we could not account for some of the systematic errors present in the experiment. HDX experiments (multiple time points in replicates of three or more; apo versus ligand-bound protein) often require in excess of 12 h for completion. The extended timescale allows for subtle changes to the environment (pH and temperature) to systematically impact the precision of the experiment with respect to time. Variations in pH may also occur when mobile phase and buffer are replaced. Figure 1 presents data from the analysis of apo PPARγLBD acquired on three different days (day one, day five, and plus 2 mo). Percent deuterium exchange versus Log time(s) plots are shown for six regions of the receptor. Six different on-exchange time points were acquired on each of the three days, and every data point represents the mean of four replicate measurements (error bars are the standard deviation of the measurement). As might be expected, systematic errors become significant at the 2-mo stage, but measurement variability was also observed over the 5-d period, as shown in the data for region 391–401. The error at the 2-mo time point can be categorized as systematic due its displacement in the y-axis from the day one and day five data, with no change in slope of the line, or in the standard deviation in the measurement. A positive displacement of the data may indicate a more efficient recovery of deuterium, an increase in the pH of the on-exchange buffer, or an increase of the temperature for the on-exchange step. This positive displacement was observed for regions 279–287 (+2), 453–477 (+3), and 453–469 (+3). A negative displacement of the slope would indicate a reduction in the recovery of deuterium, an increase in the pH of the on-exchange buffer, or a decrease in the temperature for the on-exchange step.
Based on these observations we re-designed our HDX experiment to acquire data from two samples in parallel, and the order in which the multiple on-exchange time periods are acquired was randomized. Together, these changes should mitigate any systematic errors that may occur over the timescale of the experiment. For differential experiments, every analysis now includes its own apo sample to serve as an internal control. Although the acquisition of the apo control sample requires additional instrument time, the automation minimizes any adverse effect on the throughput of the experiment. Importantly, the inclusion of this internal control allows us to compare with confidence datasets acquired days and even months apart, or data acquired between our two HDX systems, which are interfaced with different mass spectrometers. Not following this type of approach is a significant limitation of the many previously published HDX data.
To evaluate this new configuration, the HDX rate of apo PPARγLBD was compared to PPARγLBD incubated with the PPARγ partial agonist, nTZDpa. The experiment required 24 h to complete, and no evidence of systematic error was observed in any of the 31 regions of the receptor, 6 of which are shown in Figure 2. It can be seen that binding of nTZDpa reduced the exchange rate of regions spanning residues 212–222, 299–309, and 341–351.
We recently completed six differential HDX experiments to investigate the changes in HDX rate of PPARγLBD upon binding of ligand. A complete description of the results and significance of this study are to be presented elsewhere; however, one feature became apparent. Looking at the complete dataset of 186 log time plots (6 ligands × 31 peptides ions), it was shown that if any change in exchange rate were observed for any region of the protein, it would be reflected at the 30-sec and 60-sec time points. This is not the case for the 1-sec, 900-sec, and 3600-sec time points. This behavior is illustrated in Figure 3; changes in HDX rate were observed for nTZDpa, GW1929, MRL-20, and MRL-24. In all cases, the reduction in exchange rate was reflected in the 30-sec data. Therefore, it should be possible to characterize the HDX rate profile of a ligand with a single time point experiment. However, the comprehensive differential HDX experiments are required to establish the optimal on-exchange time point for the protein of interest.
Having established that a single time point HDX experiment is feasible for discriminating between the HDX profiles of PPARγ ligands, it should now be possible to realize a significant increase in the throughput of the experiment. The comprehensive experiment required approximately 24 h to acquire the data for a single ligand. The chromatography for this experiment was based around a 15-min gradient and a 50 mm × 2.1 mm chromatography column packed with 5-μm-diameter particles. It has been demonstrated that columns packed with sub-2-μm-diameter particles can provide significant increases in the resolution of the HPLC separation. Although the resolution improves, the reduction in particle size results in a concomitant increase in the backpressure of the system. For HDX experiments, the pressure increase is magnified by the requirement to perform the chromatography at temperatures close to 0°C. Upon a switch from 5-μm-diameter stationary phase reversed-phase C18 column (50 mm × 2.1 mm) to an equivalent column packed with 1.9-μm stationary phase, we observed an increase in pressure from 50 Bar (725 psi) to 350 Bar (5080 psi) at a flow rate of 200 μL/min. This pressure is well below the upper limit of our flow-switching valves (10,000 psi); however, it was above that which could be accommodated by our C8 trap column. It would be possible to run the experiment without a trap column; however, we believe the inclusion of the trap column is critical to maintain the robustness of the system. To address this problem, we switched to a C8 trap column that is rated to 400 Bar. The decrease in particle size allowed us to reduce the gradient time from 15 min to 10 min with no loss in the resolution of the separation. Figure 4 shows the total ion chromatogram (TIC) and three extracted ion chromatograms obtained from the separation of the pepsin digest of PPARγLBD. The peak width at ½ height was approximately 6 sec and was sufficient to ensure each of the 31 peptide ions of interest was not overlapped with any other ions during the mass analysis. The reduction in the gradient time allowed us to reduce the injection-to-injection cycle time and reduce the time required to complete the data acquisition to around 3 h. Within this 3-h timeframe, four replicate 30-sec on-exchange points for samples A and B were acquired, along with four replicates of a non-deuterated control.
Having demonstrated that we could characterize ligand-induced changes to the HDX rate of PPARγLBD with a single-time-point experiment, we proceeded to characterize six synthetic and four endogenous PPARγ ligands. These data were acquired with approximately 24 h of instrument time, only marginally longer than the time required to perform a single differential HDX experiment. Data were obtained for 28 regions of the receptor, and none of these peptide ions exhibited any spectral overlap with other ions. Figure 5 shows some illustrative data for two regions of the receptor: 279–287 (top row) and 470–477 (bottom row). Data for three ligands are shown: rosiglitazone (left), MRL-20 (center), and BVT.13 (right). For each region of the protein, a t-test was used to establish whether there were any significant differences between the apo and ligand-bound HDX data (P < 0.001). Table 1 shows the difference between the percent HDX rates for 28 peptide ions that span the LBD of the receptor.
As expected, the high-affinity (Kd in the low nM range) synthetic ligands rosiglitazone, MRL-20, MRL-24, GW1929, and BVT.13 induced a significant reduction in HDX rates for multiple regions of the receptor. However, very little change in HDX rate was observed upon binding of the low-affinity (IC50 ~ 8 μM) partial agonist MCC-555. All ligands except for MCC-555 reduced the rate of exchange of residues 279–287, which are located in H3 and form part of the ligand-binding pocket. The full agonists rosiglitazone, MRL-20, and GW1929 induced the characteristic reduction in HDX rate for H11 and H12. Critically, the trends observed in this single-time-point HDX data set for rosiglitazone and nTZDpa match our previously published observations from comprehensive HDX studies.37
Here we provide the first HDX data describing the effects of binding endogenous ligands to the PPARγ-LBD. The prostaglandin J2 metabolite 15-deoxy-δ12,14-PGJ2 (15PGJ2 )has been shown to selectively bind to PPARγ and promote adipocyte differentiation.41,42 Activation of the receptor was achieved at 100 nM concentration, and the EC50 was determined to be 2 μM.42 It should be noted that 15PGJ2 has been shown to covalently modify PPARγ at Cys285; therefore, only the Cys285-containing peptide conjugated to 15PGJ2 was detected. Significant reductions in HDX rates were observed for other regions of the receptor, as detailed in Table 1. Residues 453–469, spanning H11 and the first two residues of H12, were stabilized upon binding of the 15PGJ2; however, no stabilization of H12 residues 470–477 was detected. The lack of stabilization of H12 by 15PGJ2 is consistent with the reduced transactivation efficiency (~65%) of PPARγ relative to the full agonist rosiglitazone (Figure 1, ref. 42). The β-sheet region 341–351 exhibited the largest reduction in HDX rate upon ligand binding.
A number of oxidized lipids, including 9(S)-hydroxy-10E,12Z-octadecadienoic acid (9(S)-HODE), 13(S)-hydroxy-9Z, 11E-octadecadienoic acid (13(S)-HODE), and 15(S)-hydroxy-5Z,8Z,11Z,13E eicosatetraenoic acid (15(S)-HETE) have been shown to be endogenous ligands of PPARγ.43,44 The relative potency in a cell-based transactivation assay was determined to be rosiglitazone > 15PGJ2 > 9(S)-HODE >13(S)-HODE > 15(S)-HETE.43 In a competition binding assay, both 9(S)-HODE and 13(S)-HODE were able to displace rosiglitazone from PPARγ. We have demonstrated that all three oxidized lipids induce unique changes to the HDX rates of the PPARγLBD. 9(S)-HODE was shown to stabilize regions of H12; however, this was not the case for 13(S)-HODE and 15(S)-HETE. The stabilization of H12 by 9(S)-HODE may account for the increased activation of PPARγ in the transactivation assay compared to 13(S)-HODE and 15(S)-HETE. The least potent activator (15(S)-HETE) induced the most changes to the HDX rate of the receptor and induced a pattern that closely matched the pattern observed for 15PGJ2, but without stabilization of the H11–H12 region spanning residues 453–477. The lack of stabilization of H11–H12 may account for the reduced potency of 15(S)-HETE compared to 15PGJ2.
It is interesting that there is no overlap between the HDX profiles of 9(S)-HODE and 13(S)-HODE. This appears to indicate that the two ligands, both of which are metabolites of linoleic acid, bind to the receptor in different orientations. This is possibly accommodated due to the large size of the PPARγ ligand-binding pocket. These HDX data have been overlaid onto the X-ray structure of PPARγLBD (2PRG.pdb) and are shown in Figure 6.
We have demonstrated the utility of HDX to determine differences in dynamics in nuclear receptors upon interaction with various ligands. Here, we demonstrate an improved automated HDX platform that couples integrated software and methodological improvements to minimize systematic error. Improvements include the use of high-pressure chromatography to improve chromatographic resolution, helping to avoid peak envelop overlap and to reduce the injection-to-injection cycle time. Coupled with low- or high-resolution mass spectrometry, rapid analysis of protein-ligand interactions is now possible. To this end, we have developed a two-stage method for rapid determination of protein-ligand interactions. This approach focuses on monitoring a single on-exchange time point that was previously determined as useful for ligand classification. This approach facilitates the analysis of approximately ten ligands in a 24-h period. It is important to note that the HDX fingerprints offer important information about binding mode of ligand and they facilitate functional classification of the ligand. We applied this technology to the analysis of several synthetic PPARγ ligands, each of which was shown to induce unique changes in the HDX rate of the receptor LBD.
The newly developed methodology was also applied to the analysis of receptor dynamics upon binding of various endogenous ligands. Interestingly, there was no overlap between the HDX fingerprints obtained for 9(S)-HODE and 13(S)-HODE, two structurally similar fatty acids. The increased sensitivity and reduced sample requirements of the new approach will facilitate screening ligands that are not available in large quantities, difficult to synthesize, or that are very expensive, such as many of the putative endogenous ligands.
The authors thank Theodore M. Kamenecka and Yuanjun He for providing MRL-20, MRL24, and BVT.13. The Scripps Research Institute—Scripps Florida is supported by the State of Florida.