Vanderbilt Institute of Chemical Biology



Discovery at the VICB







Energy Transduction in a Major Drug-Resistance Transporter


By: Carol A. Rouzer, VICB Communications
Published:  March 29, 2017


Double electron-electron resonance and molecular dynamics studies of the P-glycoprotein reveal how energy from ATP hydrolysis drives substrate transport.


The ABC (ATP-binding cassette) transporters comprise a class of single- or multi-subunit proteins that include a transmembrane domain and a membrane-associated ATPase. These proteins use the energy from ATP hydrolysis to power conformational changes in the transmembrane domain that result in transport of substrates across the cell membrane against a concentration gradient. The P-glycoprotien (P-gp, also known as multidrug resistance protein 1) is one of the most thoroughly studied ABC transporters. Expressed in bacteria, fungi, and animals, P-gp transports more than 200 structurally diverse substrates, including many xenobiotic compounds. Expressed in the brain, liver, gastrointestinal tract, and kidney in mammals, P-gp is important for the ability of the blood-brain barrier to restrict entry of many xenobiotics to the brain, and it is responsible for the active excretion of a wide range of compounds via the urine or feces. Consequently, P-gp plays an important role in the pharmacokinetics of numerous drugs and toxicants, and it is also a key player in the ability of many cancer cells to resist the effects of anti-tumor therapeutic agents. Despite intensive study for over 40 years, the mechanism by which ATP hydrolysis leads to substrate transport in P-gp, as well as most of the ABC transporters, remains unclear. This led Vanderbilt Institute of Chemical Biology member Hassane Mchaourab, his collaborators Robert Nakamoto (University of Virginia) and Emad Tajkhorshid (University of Illinois at Urbana-Champaign), and their laboratories to apply double electron-electron resonance (DEER) and molecular dynamics approaches to develop a better understanding of energy transduction between P-gp's ATPase activity and substrate transport. [B. Verhalen, R Dastvan, et al., (2017) Nature, published online March 16, 2017, DOI:10.1038/nature21414 . Copyright 2017].



P-gp is a monomeric protein comprising two nucleotide binding domains (NBDs), each located at the C-terminus of its two halves and a transmembrane domain (TMD) comprising twelve alpha-helices. These twelve transmembrane helices (TMs) are gathered into two bundles, one containing TMs 1-3, 6, 10, and 11, and the other containing TMs 4, 5, 7-9, and 12. In the absence of substrate or ATP (apo state), the TMD adopts a V-shape with the vertex lying at the extracellular leaflet of the plasma membrane. At the intracellular leaflet of the membrane, the two helical bundles are spread apart forming an opening through which substrate can enter the transport channel. Attached to the intracellular ends of helix 6 and 12 are the two NBDs, which extend into the cytoplasm (Figure 1a). This is the "inward-facing" conformation of the protein.



FIGURE 1. (a) Diagrammatic representation of the crystal structure of apo P-gp. The N-terminal half of the protein is shown in yellow, and the C-terminal half in cyan. The N- and C-terminal NBDs are each attached to α-helixes 6 and 12, respectively. Each of these helices is part of a 6-helix bundle that forms the transport channel. Purple spheres denote sites of spin-labels used for DEER studies. (b) Distance distributions between two spin-labels, one in the N-terminal NBD, and the other in the C-terminal NBD. Distributions are provided for the protein incubated under various conditions as indicated by the color coding. Note that conditions that mimic the transition-state (red line) result in a marked shortening in the distance between the two labels. (c) Same as (b) except in this case the labels are on opposite arms of the TMD at the cytosolic leaf of the membrane. (d) View of the protein looking down from the extracellular environment and distance distributions for labels placed on opposite arms of the TMD in this region. Note here that conditions that mimic the transition state lead to increased rather than decreased distances and that the distributions are broad and often include multiple peaks. Image reproduced by permission from Macmillan Publishers Ltd, from B. Verhalen, R Dastvan, et al., (2017) Nature, published online March 16, 2017, DOI:10.1038/nature21414 . Copyright 2017.


A complete DEER analysis requires that researchers first construct a mutant protein in which all cysteine residues have been converted to alanine or serine. They then used this cysteine-less protein to design and express a series of mutant proteins, each containing two cysteine residues located at carefully selected sites that are predicted to move with respect to each other during a conformational transition. Once this is accomplished, the expressed mutant proteins are labeled with a sulfhydryl-targeted spin-label. The Mchaourab lab used MTSSL [(1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl) methanethiosulfonate] for this purpose. After careful verification that the structural changes do not significantly affect protein function, the spin-labeled proteins are ready for DEER measurements. These enable the researchers to monitor changes in the distance between the two spin-labeled sites in the protein. For their P-gp experiments, the investigators designed mutant proteins carrying pairs of labels at opposing positions on the two NBDs, or on the two TMD bundles at either the cytoplasmic or extracellular ends of the domain (Figures 1a and 1d).


To attempt to capture the full range of protein conformations, the investigators generated DEER data under five sets of conditions. They incubated the protein in the apo state, in the presence of the substrate verapamil (Ver) alone, in the presence of Ver and ATP (conditions that should result in ATP hydrolysis and Ver transport), in the presence of Ver and AMP-PNP (a nonhydrolyzable ATP adduct), and in the presence of Ver, ATP, and vanadate (Vi, which traps the protein in a conformation resembling the transition state). They used DEER to measure the distance between different parts of the protein under each condition.


Because proteins are not rigid, static structures, DEER data are actually distributions that provide the probability that the two labels will be at a particular distance from each other in a given state. A broad distribution suggests conformational flexibility between the two labeled sites, while a narrow distribution suggests rigidity. Remarkably, the DEER data showed very similar distributions – suggesting very similar conformations – for all conditions studied except for one. Incubation with Ver, ATP, and Vi to mimic the transition state led to a sharp reduction in distance between the two NBDs and between the TMD bundles at the intracellular-facing side of the membrane. In all cases, a relatively sharp and uniform distance distribution suggested that these regions of the protein had adopted a fairly uniform conformation under these conditions. In contrast, distances between the TMD bundles at the extracellular-facing side of the membrane increased, and the distributions were generally broad and often showed multiple peaks, suggesting the presence of multiple conformations in this region of the protein in the transition state.


These initial DEER data were consistent with prior models of P-gp function. In the apo state, the protein adopts the inward facing conformation to provide access into the transport channel from the cytosol. After substrate enters the channel, ATP binds to the NBDs, causing them to dimerize, and the inward-facing channel opening closes. Upon ATP hydrolysis, the channel adopts an "outward facing" conformation that provides an egress route for the substrate into the extracellular environment. Thus, it appeared that the DEER data captured the dimerization of the NBDs and the closing of the intracellular access route into the transport channel. They also suggested opening of the extracellular egress site, though protein flexibility in this region appears to remain high.


Prior work had indicated that dimerization of the NBDs leads to the formation of two nucleotide binding sites where ATP hydrolysis subsequently occurs. The researchers were interested in learning if these two binding sites could be distinguished structurally. So, they created a P-gp protein bearing two spin labels, one in each of the NBDs in the proximity of the first nucleotide binding site (NBS1). They then created a second P-gp protein bearing labels in the corresponding locations in the second nucleotide binding site (NBS2). Construction of several such pairs of proteins enabled them to evaluate the effects of conformational changes on various regions of the two nucleotide binding sites. They found that, in most cases, the conformational changes observed in the two sites were symmetrical. There was one exception, however. Adoption of the transition state conformation induced a motion in the region of residue 1043 (in NBS2) that was not observed in the comparable residue (400) in NBS1. These two residues lie next to the "A-loop", a structure that interacts with ATP's adenosine moiety, and the results suggested a difference in the motion of the NBSs in this region.


To further explore nucleotide binding site asymmetry, the researchers created P-gp proteins bearing a site-directed mutation at one or both of two conserved glutamate residues – E522 in NBS1 and E1197 in NBS2 – to glutamine. These glutamates help to polarize a water molecule and position it for attack on the γ-phosphate of ATP during catalysis. Mutation of either glutamate reduced the observed asymmetry in nucleotide binding site motion, and the E1197Q mutation actually reversed it. The double E522Q/E1197Q mutation resulted in nearly symmetrical motion between the two sites. In addition, these mutations resulted in a stabilization of a transition state-like conformation in the presence of ATP and Ver, presumably due to the impairment of the ability of the enzyme to hydrolyze ATP in the absence of the catalytic glutamate carboxyl group.


Currently, no crystal structure data are available that reflect the outward open conformation of P-gp. Thus, the investigators used a molecular dynamics approach to construct a model of that conformation (Figures 2a-c). This independently built model is compatible with their DEER data. As expected, the model predicts that the two NBDs have come together to form a dimer containing two nucleotide binding sites. In addition, the portions of the TMD bundles that face the cytosol have come tightly together, closing access to the transport channel on this side of the membrane. However, an opening now exists between the TMD bundles at the extracellular side of the membrane, providing egress for the substrate.


FIGURE 2. (a) Diagrammatic representation of a side view of the outward facing conformation of P-gp as proposed by molecular dynamics simulation. (b) View of the outward-facing conformation as seen from the extracellular environment. (c) View of the outward-facing conformation as seen from the cytosol. (d) Proposed model of energy transduction in P-gp. (d1) Conformational sampling of the apo-protein leads to "loose" ATP binding at the nucleotide binding sites in the NBDs. (d2) After binding of substrate, the NBDs dimerize, leading to tight binding (occlusion) of ATP at NBS1. (d3) ATP hydrolysis at NBS1 fuels a conformational change to the doubly occluded form, trapping the substrate in the channel, and leading to tight binding of ATP at NBS2. (d4) Hydrolysis of ATP at NBS2 propels the conformational change to the outward-facing state, allowing release of substrate and the products of ATP hydrolysis, so the cycle can be repeated. Image reproduced by permission from Macmillan Publishers Ltd, from B. Verhalen, R Dastvan, et al., (2017) Nature, published online March 16, 2017, DOI:10.1038/nature21414 . Copyright 2017.



Assembling all of their data, plus what was already known about the function of P-gp, the investigators proposed a new model for energy transduction in the protein (Figure 2d). Their model suggests that P-gp spends most of its time in the inward-facing conformation. The apo protein samples multiple conformations that enable the binding, but not hydrolysis of ATP at the nucleotide binding sites. Upon binding of substrate, the NBDs dimerize, and ATP becomes tightly bound (occluded) at one of the sites, likely NBS1. The energy harnessed from the subsequent hydrolysis of ATP at NBS1 induces additional conformational changes that close the channel, leading to a doubly occluded conformation. This conformational change also results in tight binding and hydrolysis of ATP at NBS2, and the resultant energy propels yet another change, this time producing the outward-facing conformation. Finally, release of the substrate, along with the products of ATP hydrolysis, returns the protein to the apo state, enabling another cycle. Note that the new model includes asymmetry of structure and function at the two nucleotide binding sites, proposing that separate ATP hydrolysis events propel distinct conformational changes. The model also suggests that the outward-facing conformation is short-lived, a mechanism that maximizes the availability of the inward-facing conformation to bind additional substrate.


These new insights into the function of P-gp are important to our understanding of the mechanism by which this protein impacts the uptake and elimination of a large assortment of drugs and toxicants. Such knowledge is key to our ability to design modulators of P-gp function that may prevent the development of anti-tumor drug resistance or other P-gp-mediated alterations of drug pharmacokinetics.



View Nature article: Energy transduction and alternating access of the mammalian ABC transporter P-glycoprotein







The Vanderbilt Institute of Chemical Biology, 896 Preston Building, Nashville, TN 37232-6304, phone 866.303 VICB (8422), fax 615 936 3884
Vanderbilt University is committed to principles of equal opportunity and affirmative action. Copyright © 2014 by Vanderbilt University Medical Center