Differential Effects of Oleuropein and Hydroxytyrosol on Aggregation and Stability of CFTR NBD1- ∆ F508 Domain

: Cystic Fibrosis (CF) is caused by loss of function mutations in the Cystic Fibrosis transmembrane conductance regulator (CFTR). The folding and assembly of CFTR is inefﬁcient. Deletion of F508 in the ﬁrst nucleotide binding domain (NBD1- ∆ F508) further disrupts protein stability leading to endoplasmic reticulum retention and proteasomal degradation. Stabilization and prevention of NBD1- ∆ F508 aggregation is critical to rescuing the folding and function of the entire CFTR channel. We report that the phenolic compounds Oleuropein and Hydroxytryosol reduce aggregation of NBD1- ∆ F508. The NBD1- ∆ F508 aggregate size was smaller in the presence of Hydroxytryosol as determined by dynamic light scattering. Neither phenolic compound increased the thermal stability of NBD1- ∆ F508 as measured by differential scanning ﬂuorimetry. Interestingly, Hydroxytyrosol inhibited the stabilizing effect of the indole compound BIA, a known stabilizer, on NBD1- ∆ F508. Molecular docking studies predicted that Oleuropein preferred to bind in the F1-type core ATP-binding subdomain in NBD1. In contrast, Hydroxytyrosol preferred to bind in the α 4/ α 5/ α 6 helical bundle of the ABC α subdomain of NBD1 next to the putative binding site for BIA. This result suggests that Hydroxytyrosol interferes with BIA binding, thus providing an explanation for the antagonistic effect on NBD1 stability upon incubation with both compounds. To our knowledge, these studies are the ﬁrst to explore the effects of these two phenolic compounds on the aggregation and stability of NBD1- ∆ F508 domain of CFTR. F508del-CFTR VX-809


Introduction
The Cystic Fibrosis transmembrane conductance regulator (CFTR) gene encodes an ATP-binding cassette (ABC) transporter expressed in the apical membrane of epithelial cells that regulates chloride, bicarbonate, and water secretion. CFTR is composed of two nucleotide binding domains (NBD1, NBD2), two membrane spanning domains (MSD1, MSD2), and a unique regulatory domain (R). ATP-dependent "head-to-tail" heterodimerization of the NBDs and Phosphorylation of the R domain controls channel gating [1,2]. This complex multi-domain membrane protein requires an extensive proteo-stasis network for proper biogenesis and folding [3]. Mutations in the CFTR gene causes the autosomal recessive genetic disorder Cystic Fibrosis (CF) that affects >70,000 people worldwide [4][5][6]. Clinical manifestations of CF include pancreatic insufficiency, cardio-pulmonary impairment, digestive disfunction, liver disease, CF-related diabetes, and male infertility [7].
The native fold of NBD1 is in equilibrium with a low-energy "molten-globule" conformation that has native-like secondary structure [22,23]. The ∆F508 mutation destabilizes the native conformation shifting the equilibrium to favor the molten-globule conformation that leads to robust aggregation in vitro. Binding of ATP to NBD1 can help to stabilize the native fold and prevent aggregation. In addition, multiple second-site mutations and chemical correctors have been identified that suppress formation of the molten-globule conformation caused by ∆F508 mutation and rescue the plasma membrane trafficking defect of CFTR [9][10][11][12][13][22][23][24][25][26]. This shift in the folding pathway is based on extensive thermodynamic analyses and is thought to occur through a "mass action" effect [15,23,24]. Therefore, extensive screens have been conducted over the years to identify molecules that can prevent aggregation and stabilize NBD1-∆F508 [5].
Olive oil phenolic compounds possess anti-inflammatory, cardioprotective, neuroprotective, antioxidant, and anti-aggregatory activities [27]. Two compounds in olive oil that are important to these activities are Oleuropein and Hydroxytyrosol [27]. Oleuropein has been shown to inhibit amyloid (Aβ 1-42) aggregation in vitro and in vivo [28,29]. Oleuropein and Hydroxytyrosol both inhibit the aggregation of Tau [30]. In this study, we sought to determine if these two phenolic compounds could prevent the aggregation, and thermally stabilize the NBD1-∆F508 domain of CFTR.
In this paper, we demonstrate that Oleuropein and Hydroxytyrosol suppress the aggregation of NBD1-∆F508 in vitro. Furthermore, aggregate size is reduced in the presence of Hydroxytyrosol as measured by dynamic light scattering. The thermal stability of NBD1-∆F508 is unaltered in the presence of Oleuropein and Hydroxytyrosol as determined by differential scanning fluorimetry. Interestingly, Hydroxytyrosol antagonizes the stabilizing effect of Bromoindole-3-acetic acid (BIA). Molecular docking studies predict that Hydroxytyrosol binds to similar surfaces as BIA and this may explain the antagonistic effect. To our knowledge, these studies are the first to explore the effects of these two phenolic compounds on the aggregation and stability of NBD1-∆F508 domain of CFTR.

Compounds
Oleuropein and Hydroxytyrosol were purchased from Sigma (St. Louis, MO, USA) and were >98% pure. Bromoindole-3-acetic acid (BIA) was purchased from Sigma and was 97% pure. All compounds were dissolved in dried DMSO and stored at −20 • C.

mNBD1-∆F508 Expression and Purification
Mouse NBD1-∆F508 (residues G404-L644) was expressed and purified as described [31]. Briefly, BL21 (DE3) E. coli transformed with pET28a-mNBD1-∆F508 plasmid were inoculated into 4 mL LB media containing 30 µg/mL kanamycin and grown overnight at 37 • C with vigorous shaking. The 4 mL saturated culture was diluted into 200 mL media and grown until OD 600~0 .60 then expression was induced with 1 mM isopropyl-1-thio-β-Dgalacto-pyranoside for 3 h at 37 • C. Cells were harvested by centrifugation at 3200× g for 20 min at 4 • C and the cell pellet was resuspended in 40 mL buffer A (20 mM Tris-HCl pH 7.9, 500 mM NACl, 5 mM imidazole) plus protease inhibitors (cOmplete TM EDTAfree cocktail, Roche). Cells were lysed by sonication (3 × 30 s, 50 duty cycle) on ice and centrifuged for 30 min at 14,993× g. The post-sonicated pellet was resuspended in 5 mL buffer A containing 6 M Guanidine hydrochloride (GdnHCl) and incubated on ice for 1 h. Resuspended pellet was centrifuged at 39,156× g for 20 min at 4 • C. Clarified lysate was incubated with 1-2 mL Ni-NTA resin (Qiagen) for 1 h on ice. Resin was washed sequentially with 10 mL buffer A with 6 M GdnHCl and 6 mL buffer A with 6 M GdnHCl and 20 mM imidazole. Protein was eluted with buffer A containing 6 M GdnHCl and 400 mM imidazole. Eluted protein was dialyzed against 2-4 L of 100 mM Tris-HCl pH 7.4 and 2 mM EDTA overnight at 4 • C. Precipitated protein was collected, dried and stored at −80 • C. The mNBD1-∆F508 protein was~90% pure based on densitometry of Coomassie stained gel.

Aggregation Assay
Purified mNBD1-∆F508 (G404-L644) was resuspended in 100 µL of 6 M GdnHCl, 20 mM HEPES, pH 7.5 that yielded 28-37 µM protein based on 280 nm absorbance measurements. The resuspended protein was diluted to a final concentration of 2.5 µM in refolding buffer (100 mM Tris-HCl, pH 8, 385 mM L-arginine, 200 mM KCl, 20 mM MgCl 2 , 1 mM DTT) and aggregation was monitored by turbidity (light scattering) at a wavelength of 400 nm at 35 • C in a Spectrovis-Plus Spectrophotometer (Vernier) in the absence or presence of the indicated concentrations of Oleuropein or Hydroxytyrosol. An equivalent amount of DMSO was used in the control experiments. Absorbance values were normalized to one using the control at the 40-min (Oleuropein) or 30-min (Hydroxytyrosol) timepoints. Experiments were performed with purified protein from at least two different preparations and two to five experiments per condition: Oleuropein control (n = 2), Oleuropein treatment (n = 4), Hydroxytyrosol control (n = 5), Hydroxytyrosol treatment (n = 4). Kinetic data was fitted to a four-parameter logistic equation to determine the inflection timepoint using Quest Graph™ Four Parameter Logistic (4PL) Curve Calculator.

Dynamic Light Scattering
Particle size was determined using a Malvern Zetasizer Nano ZS equipped with a 633 nm laser. Multiple ten-second scans were performed at room temperature and averaged for each run with dust filter algorithm enabled to remove noise. At the end of the aggregation assay (30-min timepoint), mNBD1-∆F508 protein sample was transferred from the Spectrovis-Plus Spectrophotometer to the Zetasizer and average particle diameter and polydispersity index were measured. Three to four experiments with multiple replicates were performed: control (n = 4, 12 replicates), Hydroxytyrosol treatment (n = 3, 9 replicates).

Differential Scanning Fluorimetry
The Protein Thermal Shift Dye Kit (Applied Biosystems) was used according to manufacturer's protocol. All reactions were assembled either on ice or in a cold room at 4 • C. Either 2 µM or 5 µM of mNBD1-∆F508 (residues T389-G673) was used in assays. The well plate was sealed with MicroAmp Optical Adhesive Film and spun at 1000 rpm for 1 min at 4 • C. A Melt curve experiment was performed with ROX reporter, no quencher, and continuous ramp mode on an ABI 7500 real-time PCR system. Settings were Step 1 at 25 • C and a ramp rate of 100% (1.6 • C/s) for 2 min followed by Step 2 at 99 • C with 1% ramp rate (0.05 • C/s). Results were analyzed by the dman2.6 tool to extract melting point temperatures [32]. Experiments were performed three times and samples were run in triplicate.

Molecular Docking Simulations
Missing residues 414 QSNGDRKHSSDENNV 428 residing in the regulatory insertion (RI) were added to the mNBD1-∆F508 (PDB:3SI7) structure using MODELLER [33]. Five structures with modeled regions were generated and the model chosen for docking simulations had an −1.67 zDOPE score and estimated 1.190 Å RMSD (0.957 estimated overlap) with the original 3SI7 structure. The refined structure was uploaded to SwissDock for docking simulations [30]. Oleuropein and BIA molecule files were created in MOL2 file format by the Dr. Lott Research group (WCU). Docking simulations of mouse NBD1-∆F508 were performed with each individual ligand using Swissdock default settings. Swissdock uses CHARMM format EADock DSS software variant that loads 5000-15,000 binding modes and calculate the most favorable energies using the FACTS implicit solvent model [34]. Ligand coordinates were removed from consideration if exhibited any of the following: orientation of ligand away from NBD1, positive full fitness, or positive ∆G scores. Molecular visualization and graphics were performed using Chimera software.

Statistical Analysis
Data is expressed as mean ± SEM. Statistical differences between control and Oleuropein/Hydroxytyrosol treated samples were compared by unpaired t-test or ANOVA. Results were considered statistically significant if p < 0.05.

Oleuropein and Hydroxytyrosol Suppress the Aggregation of mNBD1-∆F508
We expressed and purified mouse NBD1-∆F508 (mNBD1-∆F508) from bacteria using immobilized metal affinity chromatography. This mNBD1 construct includes residues G404 to L644 and contains the F1-type core ATP-binding subdomain allowing the measurement of early folding intermediates ( Figure 1A) [15,31,35]. Protein purity was~90% based on gel densitometry ( Figure 1B). Mouse NBD1-∆F508 was chosen over human for experiments because to obtain experimentally useful quantities of human NBD1-∆F508 requires multiple solubilization mutations and these mutations attenuate the ∆F508 defect and cause a twofold prolonged gating cycle duration compared to wildtype [17,26,36,37]. Furthermore, deletion of the regulatory insertion overcomes solubility issues but also suppresses gating and trafficking defects of ∆F508 [10]. Importantly, purified mouse NBD1-∆F508 has been used to probe Hsc70 interactions and successfully screen for pharmacological chaperones for human CFTR [13,38].
Purified mNBD1-∆F508 (2.5 µM) was incubated with 100 µM Oleuropein (OLE) or Hydroxytyrosol (HT) and aggregation was monitored by light scattering using a spectrophotometer (see materials and methods). OLE and HT suppressed mNBD1-∆F508 aggregation on average by 41 ± 6.9% and 34 ± 5.8%, at the terminal timepoint, respectively ( Figure 1C,D). The calculated p-values for the observed differences in aggregation were 0.0348 (OLE) and 0.0017 (HT) compared to control. Incubation with ten-fold lower concentrations of OLE or HT (10 µM) inhibited aggregation to a similar extent (~30-40%) but results were inconsistent, with inhibition being observed only in half the experiments. In addition, aggregation was inhibited to the same extent with 200 µM OLE with a slight decrease in the inflection timepoint compared to 100 µM concentration (~11 vs. 14 min). Therefore, 100 µM of OLE and HT was chosen for the remainder of the experiments.
HT is naturally present in the brain as a by-product of dopamine and tyramine metabolism and is the major metabolite of OLE [27]. Therefore, we chose to further investigate the effects of 100 µM HT on mNBD1-∆F508 aggregate/particle size using dynamic light scattering. The mNBD1-∆F508 particle diameter was 1758 ± 235 and 1408 ± 207 nm for control and HT treated samples, respectively (Supplementary Figure S1A). The calculated p-value for the observed difference in diameters was 0.032. Both control and HT treated samples were significantly larger in average diameter compared to the 170 nm reference beads that were 176.9 ± 10.7 nm (p = 7.17 × 10 −8 and 3.21 × 10 −6 ). There were larger aggregate species detected (~5×) for both control and HT treated samples but these represented a minority of the aggregate population (<10%) and there was no difference seen for these aggregates. The average polydispersity index for the 170 nm beads, control and HT treated aggregates were~0.3 indicating the particles were homogeneous (Supplementary Figure S1B). There was a trend of increased polydispersity for the HT treated aggregates compared to control but this was not statistically significant. HT is naturally present in the brain as a by-product of dopamine and tyramine metabolism and is the major metabolite of OLE [27]. Therefore, we chose to further investigate the effects of 100 μM HT on mNBD1-ΔF508 aggregate/particle size using dynamic light scattering. The mNBD1-ΔF508 particle diameter was 1758 ± 235 and 1408 ± 207 nm for control and HT treated samples, respectively (Supplementary Figure S1A). The calculated p-value for the observed difference in diameters was 0.032. Both control and HT treated samples were significantly larger in average diameter compared to the 170 nm reference beads that were 176.9 ± 10.7 nm (p = 7.17 × 10 −8 and 3.21 × 10 −6 ). There were larger aggregate species detected (~5×) for both control and HT treated samples but these represented a minority of the aggregate population (<10%) and there was no difference seen for these aggregates. The average polydispersity index for the 170 nm beads, control and HT treated aggregates were ~0.3 indicating the particles were homogeneous (Supplementary

Oleuropein and Hydroxytyrosol Do Not Thermally Stabilize mNBD1-∆F508
Since OLE and HT prevented aggregation of mNBD1-∆F508, we employed differential scanning fluorimetry (DSF) to determine if these polyphenolic compounds could also thermally stabilize the protein (see materials and methods). It is known that ATP and 5-Bromoindole-3-acetic acid (BIA) molecules bind and thermally stabilize NBD1 in vitro [12,18,39]. Therefore, both molecules were used as positive controls in our DSF assays. For DSF experiments, we employed a NBD1-∆F508 construct spanning residues T389-G673 in order to allow a more direct comparison with previous studies that investigated mNBD1 thermal stability [13]. In addition, this construct contains the regulatory extension that can obstruct head-tail homodimer formation and thus would minimize potential dimerization contribution to stability measurements [40]. The T m of mNBD1-∆F508 was 29.97 ± 0.32 • C and incubation with 1-or 5-mM ATP increased the average Tm by +3.73 ± 0.64 • C and +5.93 ± 0.05 • C similar to previous reports for mouse and human NBD1-∆F508 [12,13,18,39] (Figure 2A and Table 1). The calculated p-values (p 1 mM ATP = 0.00275, p 5 mM ATP = 0.00006) indicated the difference in melting temperature in the presence of the two ATP concentrations were statistically significant. In addition, 1 mM BIA increased NBD1 s thermal stability by an average of +2.13 ± 0.16 • C (p = 0.030) similar to a previous report ( Figure 2C) [18]. thermally stabilize the protein (see materials and methods). It is known that ATP and 5-Bromoindole-3-acetic acid (BIA) molecules bind and thermally stabilize NBD1 in vitro [12,18,39]. Therefore, both molecules were used as positive controls in our DSF assays. For DSF experiments, we employed a NBD1-ΔF508 construct spanning residues T389-G673 in order to allow a more direct comparison with previous studies that investigated mNBD1 thermal stability [13]. In addition, this construct contains the regulatory extension that can obstruct head-tail homodimer formation and thus would minimize potential dimerization contribution to stability measurements [40]. The Tm of mNBD1-ΔF508 was 29.97 ± 0.32 °C and incubation with 1-or 5-mM ATP increased the average Tm by +3.73 ± 0.64 °C and +5.93 ± 0.05 °C similar to previous reports for mouse and human NBD1-ΔF508 [12,13,18,39] (Figure 2A and Table 1). The calculated p-values (p1 mM ATP = 0.00275, p5 mM ATP = 0.00006) indicated the difference in melting temperature in the presence of the two ATP concentrations were statistically significant. In addition, 1 mM BIA increased NBD1′s thermal stability by an average of +2.13 ± 0.16 °C (p = 0.030) similar to a previous report ( Figure  2C) [18]. NBD1-∆F508 was incubated with both ATP (1.5 mM) and BIA (1 mM) to determine if their effects are synergistic, or not. The multiple ligands increased thermal stability by 5.67 ± 0.21 • C and was statistically significant (p = 0.00015). This result suggests that multiple ligands combined can provide a synergistic effect on thermal stability of NBD1-∆F508. This result is similar to the additive effects of MCG1516A, RDR1, VX809 in the presence of ATP that were reported to stabilize human NBD1-∆F508 [21]. OLE (100 µM) increased thermal stability of NBD1-∆F508 by +0.26 • C, but this was not deemed statistically significant (p = 0.61671) ( Figure 2D). HT tested at 100 µM also did not affect NBD1-∆F508 thermal stability. We tested 5× and 10× higher concentrations of HT, 500 µM to 1 mM, and confirmed there was no significant change in the protein T m (p = 0.799 and 0.593) ( Figure 2E). Interestingly, incubation of 1 mM HT combined with 1 mM BIA increased thermal stability by +1.56 ± 0.21 • C but the results were decreased compared to 1 mM BIA alone, suggesting an antagonist effect exerted by HT (p = 0.02076) ( Figure 2F).

Predicted Molecular Docking of BIA, Oleuropein, and Hydroxytryosol to mNBD1-∆F508
We performed molecular docking studies to gain insight into the apparent antagonist effect of HT on BIA induced thermal stabilization of mNBD1-∆F508. BIA, OLE, and HT were computationally bound to mNBD1-∆F508 using Swissdock (Table S1) [34]. Multiple binding modes were identified for BIA and one predicted binding site was in a cleft between the ABCα and F1-type core ATP-binding subdomains near the apex of the α4/α5/α6 helical bundle ( Figure 3A and Figure S2A). This predicted site is very close to the binding site determined from co-crystallization experiments performed by SGX pharmaceuticals and is termed the α-site [18]. In contrast to BIA, OLE only had two predicted binding modes, or surface next to the site of ATP binding in the F1-type core ATP-binding subdomain ( Figure 3B and Figure S2B). There were many predicted binding modes for HT but the majority of the predicted clusters were in the α4/α5/α6 helical bundle of the ABCα subdomain ( Figure 3C and Figure S2C). Interestingly, many of the predicted binding modes for HT overlapped or were in close opposition (2-12 Å) to the predicted BIA modes, including sites in the ABCα-, ABCβ-subdomain (also known as α/β-subdomain) and the regulatory extension region ( Figure S2A,C). Overall, these docking results suggest that HT binds to a region near the α-site and this could interfere with BIA binding ( Figure 3C).

Discussion
The antioxidants OLE and HT suppressed the aggregation but did not alter the thermal stability of mNBD1-ΔF508. Our kinetic data was sigmoidal shaped suggesting that aggregation occurred through a nucleation-growth process [41]. We fitted the aggregation data with a four-parameter logistic regression to determine the maximum aggregation rate for control versus treated conditions (i.e., inflection parameter). There was no significant difference between the average inflection values for control (10.94 ± 0.54 min) compared to HT (11.67 ± 0.68 min), and control (16.75 ± 0.90 min) compared to OLE (14.38 ± 0.77 min) treatments, respectively ( Figure S3). The plateau phase of aggregation but not the lag phase and initial growth were affected by OLE and HT. Interestingly, there was a significant difference between the inflection values for OLE and HT treatments (p = 0.019479).
Together with our DLS measurements ( Figure 2 and Table S2), these results suggest that growth but not initial formation of aggregates is suppressed, resulting in a modest reduction of overall aggregate size. This alteration of aggregate size could be due to prevention of monomer addition or disruption of aggregates. One possibility is that OLE and HT are shifting the equilibrium to disfavor the "molten-globule" aggregation prone conformer. Our data is consistent with the model of Omar et al. who investigated Aβ fibril formation and proposed that olive oil polyphenols inhibit protein aggregation through breakdown of pre-formed aggregates [42]. Our data is also consistent with Daccache et

Discussion
The antioxidants OLE and HT suppressed the aggregation but did not alter the thermal stability of mNBD1-∆F508. Our kinetic data was sigmoidal shaped suggesting that aggregation occurred through a nucleation-growth process [41]. We fitted the aggregation data with a four-parameter logistic regression to determine the maximum aggregation rate for control versus treated conditions (i.e., inflection parameter). There was no significant difference between the average inflection values for control (10.94 ± 0.54 min) compared to HT (11.67 ± 0.68 min), and control (16.75 ± 0.90 min) compared to OLE (14.38 ± 0.77 min) treatments, respectively ( Figure S3). The plateau phase of aggregation but not the lag phase and initial growth were affected by OLE and HT. Interestingly, there was a significant difference between the inflection values for OLE and HT treatments (p = 0.019479).
Together with our DLS measurements ( Figure 2 and Table S2), these results suggest that growth but not initial formation of aggregates is suppressed, resulting in a modest reduction of overall aggregate size. This alteration of aggregate size could be due to prevention of monomer addition or disruption of aggregates. One possibility is that OLE and HT are shifting the equilibrium to disfavor the "molten-globule" aggregation prone conformer. Our data is consistent with the model of Omar et al. who investigated Aβ fibril formation and proposed that olive oil polyphenols inhibit protein aggregation through breakdown of pre-formed aggregates [42]. Our data is also consistent with Daccache et al., who observed less total aggregated material when Tau or Tau-P301L were incubated with OLE and HT [30].
The indole-based compound BIA increased mNBD1-∆F508 stability by an average of 2.13 • C and this was similar to He et al.'s reported value of 1.4 • C [18]. Interestingly, He and colleagues incubated their NBD1 constructs with 4× higher BIA concentration (4 mM vs. 1 mM) to observe a similar stabilization compared to our study. The mNBD1-∆F508 construct, used in our studies, contains the naturally disordered RI region (residues 405-436) compared to He [18]. A second difference is that He used human NBD1 in their studies compared to mouse in our experiments. However, the average T m of our mNBD1-∆F508 (residues 389-673) was~30 • C compared to 30-31 • C for human NBD1-∆F508 (residues 389-678) as previously reported [12,19]. The concentration differences between our and He et al.'s results are most likely due to the inclusion/omission of the RI region.
These phenolic compounds suppress Tau, α-synuclein, and Aβ fibrillation with low micromolar IC 50 concentrations (2-36 µM) and close to 1:1 molar ratio [30,43,44]. In contrast, a 1:40 molar ratio (100 µM) of OLE and HT was required to suppress mNBD1-∆F508 aggregation. In our hands, a lower concentration of 10 µM (1:4 molar ratio) gave inconsistent results in suppressing aggregation. Biochemical studies suggest both covalent and non-covalent interactions are believed to contribute to the HT mediated prevention of aggregation [43]. The concentration differences we observed could imply that the covalent binding of OLE and HT to NBD1-∆F508 is very weak, or that non-covalent interactions predominate (see below).
OLE and HT have been reported to suppress the aggregation of Tau and Aβ aggregation in vitro and in vivo [29,30]. Recent biophysical studies suggest that the mechanism of aggregation prevention for OLE and HT are fundamentally different [43,44]. For example, Oleuropein aglycone (de-glycosylated Oleuropein) appears to block Aβ 1-42 fibril growth through binding to the N-terminus of the peptide [44]. In contrast, HT appears to increase the rate of harmless off-pathway amorphous aggregates [44]. The HT molecule may exert this effect through stabilization of specific regions of the protein [43]. Our molecular docking results predicted that OLE binds preferentially in the ATP binding pocket and HT preferentially binds to the ABCα-subdomain. Interestingly, HT partially inhibited the stabilizing effect of BIA and this predicted interference could be caused by HT stabilization of an NBD1-∆F508 conformation that disrupts BIA binding or competes for a similar binding site as predicted by our docking studies.
Recently, several nanobodies were discovered to bind to the top of the Walker A motif in a small cavity in the β-subdomain that overlaps with the regulatory insertion packing of NBD1 ( Figure 1A) [45]. The predicted binding modes for OLE are also in the β-subdomain but these sites are between~15-25 angstroms away from the nanobody binding surface and thus would not likely interfere. Interestingly, several of the predicted binding modes for HT in the β-subdomain are~6.5 angstroms from G550 and~7.20 angstroms from S605. Both of these residues participate in polar interactions with the nanobodies to facilitate binding [45]. Therefore, the predicted HT binding sites could interfere with the binding of the nanobodies to NBD1 and potentially abrogate the thermal stabilization observed by Sigoillet et al. [45].
Several nucleotide analogs, including dTTP, are known to bind with higher affinity to the same site as ATP in NBD1 and restore ∆F508-CFTR electrophysiological function [22]. The predicted binding sites for OLE are near the ATP pocket. In contrast, the HT and BIA predicted binding sites are not near the ATP pocket. Interestingly, lower scoring binding modes for OLE, HT and BIA located in between the two subdomains could interfere with residues at the NBD1-NBD2 dimerization interface ( Figure S2) [46]. Correctors known to restore the intracellular loop 4 (ICL4) interaction with NBD1 by either direct binding to the ∆F508 loop or through indirect allosteric effects are not predicted to be affected by OLE or HT binding based on our docking studies [47][48][49]. Importantly, binding to remote sites (removed from F508) in NBD1 can lead to allosteric changes and stabilization. The predicted neutral or negative impacts of OLE or HT, described above, would need to be verified experimentally through a technique such as hydrogen/deuterium exchange mass spectrometry [49]. Overall, OLE and HT would be predicted to have limited potential as a corrector or potentiator due to the high concentrations needed for their effect, inability to stabilize NBD1-∆F508, and the predicted interference with NBD1-NBD2 dimerization, and correctors (such as BIA and nanobodies). Additional biophysical studies will be needed to fully understand the anti-aggregatory activities of these phenolic compounds towards mNBD1 and to test these predictions. Finally, future lab efforts will be directed towards characterization of the aggregate geometry and morphology in the presence of OLE and HT using electron microscopy techniques (TEM/SEM/AFM).