The crystal structure of the CmABCB1 G132V mutant, which favors the outward‐facing state, reveals the mechanism of the pivotal joint between TM1 and TM3

Abstract

CmABCB1 is a homologue of human P‐glycoprotein, which extrudes various substrates by iterative cycles of conformational changes between the inward‐ and outward‐facing states. Comparison of the inward‐ and outward‐facing structures of CmABCB1 suggested that pivotal joints in the transmembrane domain regulate the tilt of transmembrane helices. Transmembrane helix 1 (TM1) forms a tight helix–helix contact with TM3 at the TM1–3 joint. Mutation of Gly132 to valine at the TM1–3 joint, G132V, caused a 10‐fold increase in ATPase activity, but the mechanism underlying this change remains unclear. Here, we report a crystal structure of the outward‐facing state of the CmABCB1 G132V mutant at a 2.15 Å resolution. We observed structural displacements between the outward‐facing states of G132V and the previous one at the region around the TM1–3 joint, and a significant expansion at the extracellular gate. We hypothesize that steric hindrance caused by the Val substitution shifted the conformational equilibrium toward the outward‐facing state, favoring the dimeric state of the nucleotide‐binding domains and thereby increasing the ATPase activity of the G132V mutant.

Short abstract

PDB Code(s): 7DQV;

1. INTRODUCTION

P‐glycoprotein (P‐gp; MDR1; ABCB1) is a member of the ATP‐binding cassette (ABC) transporter family that transports substrates from the cell in a reaction coupled to ATP hydrolysis. ¹ , ² By pumping various chemical compounds out of the cell, P‐gp protects the body against xenobiotics (drugs) and contributes to the acquisition of multidrug resistance in cancer cells. ³ , ⁴

Three dimensional structures of P‐gps from several eukaryotes have been solved using X‐ray crystallography and cryo‐EM single‐particle analysis. ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ These structures support the alternating access model, ²⁰ in which the solvent‐accessible opening switches alternately to either side of the phospholipid membrane in a process driven by conformational changes of the transporter. ²¹ , ²² CmABCB1, the P‐gp derived from Cyanidioschyzon merolae, is highly thermostable, and its crystal structures have been successfully solved in both the inward‐facing (IF) and outward‐facing (OF) states. ⁹ , ¹⁶ A mutant of CmABCB1 (referred to as the QTA mutant) harboring the Q147A and T381A point mutations yielded high‐resolution crystal structures of both the IF and OF states ¹⁶ because the Q147A/T381A mutation abolished a hydrogen bonding network observed in the wild‐type CmABCB1 IF structure. ⁹ Comparison of the CmABCB1 IF and OF structures revealed the interaction between transmembrane helix (TM) 1 and TM3 (referred to as the TM1–3 joint) and the interaction between TM3 and TM6 (the TM3–6 joint) as key components that ensure the correct conformational change between the IF and OF states (Figure 1(b), (c)). The interactions of TM joints involve residues with small side chains that are highly conserved among eukaryotic P‐gps ⁵ , ⁶ , ⁹ but not prokaryotic ABCB family members such as Sav1866 ²³ and MsbA. ²⁴ In a previous study, we identified an interesting mutation at the TM1–3 joint, G132V, that caused highly elevated ATPase activity. ¹⁶ To understand the molecular mechanism underlying this change, we performed structural analysis of the mutant using X‐ray crystallography. Here, we present a high‐resolution crystal structure of the CmABCB1 G132V mutant and discuss the structural changes caused by the mutation in the OF state structure. The G132V mutation seems to affect the equilibrium of the conformational change between the IF and OF states, and the dynamic structure is likely to be more favorable in the OF state, leading to an increase in ATPase activity.

FIGURE 1.

Outward‐facing structure of the G132V mutant. (a) ATPase activities of CmABCB1 WT (black circle) and G132V mutant (red circle). (Left) ATP‐dependent ATPase activities. Inset: Enlarged curve of WT. Error bars indicate standard deviation (n = 3). (Right) Rhodamine 6G‐dependent ATPase activities (n = 3). (b) CmABCB1 structures solved in the inward‐ and outward‐facing states. Each structure is shown in two colors to represent the two symmetrical subunits. (c) (Left) Overall outward‐facing structure of the G132V mutant. (Right) Schematic illustration of the G132V structure. The NBD and the TMD from one subunit and the NBD from the other subunit, marked with an asterisk, are described. The side chains of Gln147 and Thr381 are shown as black sticks. Side chains of Val132 are shown as bold lines. (d) (Left) One subunit is represented as a tube with a color gradient, and the other molecule is represented as a cartoon colored in yellow. The color gradient and the tube radius represent Cα displacement from the QTA structure in the three‐dimensional structural alignment; residues that moved greater distances are shown as thick tubes colored in red, whereas those that moved less are shown as thin tubes colored in blue. (Right) Zoomed‐in view of the region in the dashed box, showing superposition of the G132V structure in reddish colors and the QTA structure in gray. The side chain of Val132 is shown as spheres. Helices except for transmembrane helix 1 (TM1), TM3, and TM6, and the corresponding helices from the other molecule were removed to allow a clear view

2. RESULTS

2.1. Overall architecture of the outward‐facing G132V structure

The G132V mutant had 20‐fold higher ATPase activity (k _cat and k _basal) and lower K _m value for ATP than the wild‐type protein (Figure 1(a), left). Kinetic curves (Figure 1(a), right) with the substrate rhodamine 6G revealed that while the basal activity of the wild‐type transporter increased with the rhodamine 6G concentration but was strongly suppressed at higher substrate concentrations (>0.1 mM), the G132V mutant abolished the stimulation of basal activity and exhibited substrate inhibition at a much lower substrate concentration (the apparent K _i of 14 μM). This is consistent with the previous finding that the G132V mutant had significantly lower transporter activity, as determined by rhodamine 6G susceptibility assay in Saccharomyces cerevisiae AD1‐8u⁻ cells. ¹⁶

We crystallized the G132V mutant in the presence of AMP‐PNP and magnesium and solved its OF structure at a resolution of 2.15 Å (Figure 1(b), (c)); data collection and refinement statistics are presented in Table 1. The overall architecture was similar to that of the previously reported OF structure of the QTA mutant ¹⁶ with a root‐mean‐square deviation (RMSD) of 0.18 Å. A comparison between the G132V structure and the OF QTA structure revealed a significant tilt of three transmembrane (TM) helices on the extracellular side (Figure 1(d)). The extracellular part of TM1 from the mutation site of the G132V structure tilted away from TM6 by 8.1° relative to the QTA structure, whereas the extracellular side of TM3 from Ala246 and that of TM6 from Gly389 of G132V tilted away from TM1 by 2.5° and 6.8°, respectively. Thus, the extracellular ends of TM1 and TM6 in the G132V structure were shifted by 3.1 and 3.2 Å, respectively, toward a further opening of the extracellular gate consisting of TM1, TM6, TM1*, and TM6* (in this notation, * indicates helices belonging to the other subunit). The intracellular regions of these structures showed little difference; in particular, the nucleotide‐binding domains (NBDs) of G132V and QTA are almost identical to each other (RMSD = 0.14 Å). For example, the residues that participate in the magnesium and nucleotide binding and catalytic residues, including Glu610 in the Walker B motif ²⁵ and His643 in the H‐loop, ²⁶ are well superimposed.

TABLE 1.

Data collection and refinement statics

Data collection
Space group	P4132
Cell dimensions
a, b, c (Å)	175.7, 175.7, 175.7
α, β, γ (°)	90, 90, 90
Wavelength	1.0000
Resolution (Å)	48.72–2.15 (2.28–2.15)
R merge	0.070 (0.766)
R meas	0.072 (0.816)
Total reflections	1,686,152 (128,754)
Unique reflections	95,087 (15,216)
I/σI	27.38 (2.25)
Completeness (%)	99.7 (98.4)
Redundancy	17.7 (8.46)
CC1/2 (%)	100.0 (78.9)
Refinement
Resolution (Å)	48.72–2.15 (2.17–2.15)
No. reflections	95,079
R work/R free	0.1850/0.2099 (0.2936/0.3476)
No. atoms
Protein	4,497
AMP‐PNP	31
Mg2+	1
Detergent	66
NO3 −	4
Water	354
B‐factors (Å2)
Protein	76.5
AMP‐PNP	32.5
Mg2+	32.1
Detergent	73.3
NO3 −	38.7
Water	50.9
R.m.s. deviations
Bond length (Å)	0.007
Bond angles (°)	0.838
Ramachandran plot (%)
Favored region	96.76
Allowed region	3.07
Outlier region	0.17

Note: Values for the highest‐resolution shell are shown in parentheses.

2.2. The TM1–3 interaction and helical tilts caused by Val132 (G132V mutation)

The G132V mutation affected van der Waals contacts at the TM1–3 interface as well as the tilts of TM1 and TM3. In the G132V structure, the side chain of Val132 formed van der Waals contacts with Ala246 and Gln247. The Cα–Cα distance between Val132 and Ala246 widened to 6.6 Å, and the distance between Val132 and Gln247 widened to 5.6 Å (Figure 2(a)). In the previously solved OF structure of the QTA mutant, interactions among small residues (Ser128, Gly132, Gly239, Ala246) at the TM1–3 joint provide tight contacts between TM1 and TM3 (Figure 2(b)). Gly132 interacts with Ala246 and Gln247 via its Cα atom with Cα–Cα distances of 5.0 and 4.6 Å, respectively, keeping those helices closely (Figure 2(b)). The other TM1–3 contact between Ser128 and Gly239 was only slightly altered by the mutation. The Cα–Cα distance of Ser128 and Gly239 is 5.7 Å in the G132V structure (Figure 2(a)), whereas in the QTA structure it is 5.9 Å (Figure 2(b)).

FIGURE 2.

Comparison of transmembrane helix 1 (TM1), TM3, and TM6 conformation between the G132V structure and the QTA structure. (a, b) TM1–3 interactions of the G132V structure (a) and the QTA structure (b). The regions around Val132 and Gly132, indicated by the dashed boxes, are enlarged. Side chains of the residues participating in van der Waals contacts are shown as spheres. Stick model represents main chains of the residues participating in the hydrogen‐bonding network in the TM1 backbone. (c) TM3–6 interactions of the G132V structure (left) and QTA structure (right). (d) Alternating‐access model for the transport mechanism of CmABCB1 of WT (upper) and G132V mutant (bottom) between the inward‐facing state (left) and outward‐facing state (right). The NBD and TMD from one subunit and the NBD from the other subunit, marked with an asterisk, are described. The TM1–3 joint is represented as a ball. The side chains of Gln147 and Thr381 are shown as black sticks. The side chain of Val132 is shown as bold lines

The G132V mutation also caused reorganization of the hydrogen bonding network in the TM1 backbone around the site of the mutation. In the G132V structure, hydrogen bonds are formed between the carbonyl oxygen of Ile129 and the amide group of Val132, between the carbonyl oxygen of Leu130 and the amide group of Ala133, and between the carbonyl oxygen of Glu131 and the amide group of Thr134 (Figure 2(a)). The 3.9 Å distance between the carbonyl oxygen of Ser128 and the amide group of the Val132 and the 3.9 Å distance between the carbonyl oxygen of Ile129 and the amide group of Ala133 are unfavorable for the formation of hydrogen bonds. In the QTA structure, the amide group of Gly132 forms hydrogen bonds both with the carbonyl oxygen of Ser128 and the carbonyl oxygen of Ile129 (Figure 2(b)). In the QTA structure, Ile129 also forms a hydrogen bond with the amide group of Ala133.

2.3. Tilt and rotation of TM6, along with that of TM3, through a TM3–6 contact mediated by Thr381

The G132V structure revealed an interhelix interaction between TM3 and TM6 on the extracellular side that was not observed in the previous OF structure of the QTA mutant. Comparison of the two OF structures revealed rewinding of TM6 around Thr381 and Ser385, causing their side chains to face TM3, supporting the TM3–6 helix interaction via hydrogen bonds between the helices (Figure 2(c)). In the G132V structure, Thr381 forms hydrogen bonds with the carbonyl oxygen of Phe258 and the carbonyl oxygen of Asn378, whereas those hydrogen bonds are not present in the QTA structure due to the absence of a polar side chain caused by the T381A mutation. The side‐chain oxygen atom of Ser385 forms a hydrogen bond with the carbonyl oxygen of Gly251 in the G132V structure, but not in the QTA structure, in which the distance (5.9 Å) is unfavorable for hydrogen bonding.

3. DISCUSSION

The previous crystallization of an OF structure of CmABCB1 was achieved using the QTA mutation, in which the hydrogen bonding network via Gln147 and Thr381 is abolished and the IF conformation is destabilized. The G132V mutant was crystallized in the OF conformation despite the absence of the QTA mutation, suggesting that replacement of a hydrogen with an isopropyl group alone could destabilize the IF state and shift the conformational equilibrium toward the OF state.

Structural comparison of the OF structures of the G132V and QTA mutants revealed a widened outward opening in the G132V structure, especially on the extracellular sides of TM1, TM3, and TM6. The distance between TM1 and TM3 at the site of the mutation is larger due to the steric enlargement of the side chain of the introduced Val. The TM1 backbone around Val132 has undergone reorganization and seems to be pushed away from TM3 along with the upper side of the rest of TM1. However, this does not explain the tilt and rewinding of TM6, which does not contact TM1. We speculate that the conformational changes of TM1 and TM3 were transmitted to TM6, changing the conformation of TM6 between residues Thr381 and Gly389. In the G132V structure, we observed a TM3–6 interaction mediated by the Phe258‐Thr381 and Gly251‐Ser385 hydrogen bonds (Figure 2(c)). By acting as a “glue,” these hydrogen bonds are responsible for the tilt and rewinding of TM6. These hydrogen bonds were not observed in the previous QTA OF structure because the side chain of Thr381 was lost as a result of the T381A mutation.

We reported previously that the TM joints keep the TM1–3 and TM3–6 helix pairs closely juxtaposed in both the IF and OF conformations and work as hinge joints during conformational changes. ¹⁶ The G132V mutation at the TM1–3 joint affected the conformation of TM6 around the extracellular gate via the TM3–6 interaction. Opening the extracellular gate composed of TM1 and TM6 is accomplished by interactions among the aromatic and hydrophobic side‐chains of residues, including Phe138, Phe142, Phe383, Phe384, Ile387, and Leu388. ⁹ , ¹⁶ The association and dissociation of the interactions are driven by the tilt of TM1 and the tilt and rotation of TM6 during the overall conformational change between the IF and OF states. This implies that TM joints play important roles not only in ensuring the gate opening, but also in defining the side chain movements of specific residues that play a key role in the gating mechanism.

The compact TM1–3 joint could be a necessary structural motif for the tight closing of TM1–6 at the extracellular gate in the IF structure, and might regulate transporter activity (Figure 2(d)). The extracellular gate of the wild‐type IF structure is stabilized by the hydrogen bonding network via Gln147 and Thr381, which suppresses the ATPase activity. Closing of the extracellular gate may be supported by the tight TM1–3 contact, and the G132V mutation would open the extracellular gate by widening the TM1–3 distance and increasing the tilt angle of TM6 along with that of TM1, destabilizing the IF state. Thus, the conformational equilibrium would shift toward an OF state capable of hydrolyzing ATP, presumably causing the G132V mutant to have higher basal ATPase activity and a greater probability of ATP binding between dimeric NBDs in the OF state. The shift of the conformational equilibrium toward the OF state by the G132V mutation also could shrink and/or deform the dynamic structure of the inner chamber; this contraction makes it competent for substrate uptake. ¹⁶ We reason that substrate inhibition in CmABCB1 G132V is due to nonproductive binding of the substrate at an allosteric site situated in the vicinity of the extracellular entrance in the deformed chamber.

4. MATERIALS AND METHODS

4.1. Protein expression and purification

The G132V mutant of CmABCB1 was expressed in S. cerevisiae AD1‐8u⁻ cells. The cells were cultured in 10 ml of yeast extract‐peptone‐dextrose (YPD) medium overnight at 30°C with shaking at 220 rpm (BioShaker BR–23FP, TAITEC). A 10 ml culture was used to inoculate 100 ml of YPD medium at a starting OD₆₀₀ of one, and the cells were cultured at 30°C with shaking at 220 rpm until the OD₆₀₀ reached four. The cells were then diluted in 1 L of YPD medium to an OD₆₀₀ of 0.1, and then grown in 2.5 L baffled flasks for 24 h at 25°C with shaking at 220 rpm. Cells were harvested by centrifugation (3000 g for 15 min) and stored at − 80°C until use.

Harvested cells were resuspended with 20 mM Tris–HCl (pH 7.5) and 150 mM NaCl and were disrupted using an EmulsiFlex‐C3 (Avestin) at 25,000 psi. Crude lysates were centrifuged at 1500 g for 20 min to remove the unbroken cells and debris, and then membranes were collected by ultracentrifugation (100,000 g for 1 h). The membranes were mechanically homogenized in buffer (20 mM Tris–HCl pH 7.5, 300 mM NaCl, and 20 mM imidazole), and solubilized for 1 h by adding 1% (wt/vol) polyoxyethylene(9)dodecyl ether (C₁₂E₉) (Wako) per 5 mg ml⁻¹ protein concentration. Detergent‐solubilized material was ultracentrifuged at 100,000 g for 30 min. The supernatant was incubated for 3 h with Ni‐charged immobilized metal affinity chromatography resin (Bio‐Rad) pre‐equilibrated with wash buffer (20 mM Tris–HCl [pH 7.5], 300 mM NaCl, 20 mM imidazole, and 0.05% [wt/vol] C₁₂E₉). Nonspecifically bound proteins were removed by washing the resin, and bound proteins were eluted with wash buffer supplemented with 300 mM imidazole. A 6× His tag and the N‐terminal 92 residues were removed by trypsin treatment, and CmABCB1 was isolated by gel filtration using a Superdex200 column (GE Healthcare) with buffer (20 mM Tris–HCl [pH 7.5], 150 mM NaCl and 0.2% [wt/vol] n‐decyl‐β‐D‐maltopyranoside (DM) (Anatrace)). Peak fractions were pooled and concentrated to approximately 10 mg ml⁻¹ using a 50 kDa molecular weight cutoff Amicon Ultra‐4 concentrator (Millipore). For the ATPase assay, gel filtration was performed with 0.05% (wt/vol) n‐dodecyl‐β‐D‐maltopyranoside (DDM) (Anatrace) instead of DM. Purified proteins were flash‐frozen and stored at − 80°C until further use. All procedures were carried out at 4°C.

4.2. Crystallization

Crystallization was performed by sitting drop vapor diffusion method after 9.9 mg ml⁻¹ purified protein was incubated with 12.5 mM of AMP‐PNP and 12.5 mM MgCl₂ at 20°C for 16 h. The preincubated sample was mixed with the reservoir solution, composed of 19% (wt/vol) PEG 2000 MME, 100 mM Mg(NO₃)₂, and 100 mM KCl, in a ratio of 1 to 1 μl. A cubic crystal appeared after incubation at 20°C for 3 weeks. The crystal was flash‐cooled in liquid nitrogen after cryoprotection with a solution containing 30% (wt/vol) PEG 2000 MME and 5% (vol/vol) 1,4‐butanediol.

4.3. X‐ray data collection, processing, and structure determination

An X‐ray diffraction data set were collected at 1.0000 Å wavelength at beamline BL41XU of SPring‐8 (Hyogo, Japan) using a PILATUS 6 M detector (Dectris). The data set were processed using XDS. ²⁷ The data were initially phased by rigid body refinement using the outward‐facing QTA structure (PDB entry 6a6m) as a search model. Model building and refinement were carried out using COOT ²⁸ and PHENIX, ²⁹ respectively. Model validation was performed with MolProbity. ³⁰ All molecular images were rendered in PyMOL. ³¹

The coordinate and structure factor files have been deposited in the Protein Data Bank under the accession code 7DQV.

4.4. ATPase assay

ATPase measurements were performed as described previously. ¹⁶ Briefly, for ATPase measurements with various concentrations of ATP, purified protein was mixed with 50 mM Tris–HCl (pH 7.5) at 37°C, 150 mM NaCl, 0.05% (wt/vol) DDM, 10 mM MgCl₂, and 0–5 mM ATP. Reaction samples were incubated at 37°C and were mixed with equal amount of 12% (wt/vol) SDS to stop the reaction. Released inorganic phosphate was quantified using the phosphomolybdate method. ³² For ATPase measurements with a substrate, reaction samples including 5 mM ATP and various concentrations of Rhodamine 6G were used. The initial hydrolysis rate was calculated, and kinetic parameters were estimated by using Michaelis–Menten equation (Equation 1) for fitting of the ATP concentration‐response curve, Equation 2 for fitting of the drug concentration–response stimulation and inhibition curve for WT, and Equation 3 for fitting of the drug concentration–response inhibition curve for the G132V mutant.

$$v=\frac{k_{\text{basal}}\left[e\right]\left[s\right]}{K_{\mathrm{m}}^{\mathrm{ATP}}+\left[s\right]},$$ (1)

$$v=\left[e\right]\left(k_{\text{basal}}+\frac{\left(k_{\mathrm{sub}}-k_{\text{basal}}\right)\left[s\right]}{K_{\mathrm{m}}^{\text{Drug}}+\left[s\right]}\right)\left(1-\frac{\left[s\right]}{K_{\mathrm{i}}^{\text{Drug}}+\left[s\right]}\right),$$ (2)

$$v=\left[e\right]\left(\frac{\left(k_{\text{basal}}-k_{\mathrm{sub}}\right)}{1+\frac{\left[s\right]}{K_{\mathrm{i}}^{\text{Drug}}}}+k_{\mathrm{sub}}\right),$$ (3)

here, v is the initial ATP hydrolysis rate; [e] is the concentration of CmABCB1; [s] is the concentration of ATP or Rhodamine 6G; k _cat is the catalytic rate constant; k _basal and k _sub are the catalytic rate constants in the absence and presence of substrate, respectively; K _m ^ATP is the apparent Michaelis constant for ATP; and K _m ^Drug and K _i ^Drug are the apparent Michaelis constant for substrate activation and inhibition, respectively. Fittings were performed by GraFit (Erithacus Software).

AUTHOR CONTRIBUTIONS

Keita Matsuoka: Conceptualization; investigation; writing‐original draft; writing‐review & editing. Toru Nakatsu: Investigation; writing‐review & editing. Hiroaki Kato: Conceptualization; funding acquisition; project administration; writing‐review & editing.

CONFLICT OF INTEREST

The authors declare no competing interests.

Matsuoka K, Nakatsu T, Kato H. The crystal structure of the CmABCB1 G132V mutant, which favors the outward‐facing state, reveals the mechanism of the pivotal joint between TM1 and TM3. Protein Science. 2021;30:1064–1071. 10.1002/pro.4058