General Principles of Secondary Active Transporter Function
(cid:105)(cid:105) (cid:105) “ms” — 2019/12/16 — 1:47 — page 1 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
Chapter G ENERAL P RINCIPLES OF S ECONDARY A CTIVE T RANSPORTER F UNCTION
Oliver Beckstein ∗ Fiona Naughton
Department of Physics, Arizona State University, Tempe AZ 85287, USA
Abstract
Transport of ions and small molecules across the cell membrane against electro-chemical gradients is catalyzed by integral membrane proteins that use a source of freeenergy to drive the energetically uphill flux of the transported substrate. Secondary ac-tive transporters couple the spontaneous influx of a “driving” ion such as Na + or H + tothe flux of the substrate. The thermodynamics of such cyclical non-equilibrium sys-tems are well understood and recent work has focused on the molecular mechanismof secondary active transport. The fact that these transporters change their conforma-tion between an inward-facing and outward-facing conformation in a cyclical fashion,called the alternating access model, is broadly recognized as the molecular frameworkin which to describe transporter function. However, only with the advent of high res-olution crystal structures and detailed computer simulations has it become possibleto recognize common molecular-level principles between disparate transporter fami-lies. Inverted repeat symmetry in secondary active transporters has shed light on howprotein structures can encode a bi-stable two-state system. More detailed analysis(based on experimental structural data and detailed molecular dynamics simulations)indicates that transporters can be understood as gated pores with at least two coupledgates. These gates are not just a convenient cartoon element to illustrate a putativemechanism but map to distinct parts of the transporter protein. Enumerating all dis-tinct gate states naturally includes occluded states in the alternating access picture andalso suggests what kind of protein conformations might be observable. By connectingthe possible conformational states and ion/substrate bound states in a kinetic model,a unified picture emerges in which symporter, antiporter, and uniporter function areextremes in a continuum of functionality. ∗ Direct all correspondence to Oliver Beckstein, Department of Physics, Arizona State University, Tempe,AZ, USA. E-mail address: [email protected] a r X i v : . [ q - b i o . B M ] D ec (cid:105) “ms” — 2019/12/16 — 1:47 — page 2 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) Keywords: membrane protein, transporter, symmetry, molecular mechanisms.
Contents
1. Introduction 22. The alternating access model 43. Thermodynamics and cycles 6
4. Inverted repeat symmetry 10
5. Transporters as gated pores 14
6. Unified transport cycle model 237. Conclusion 26
1. Introduction
Active transporters are integral membrane proteins that move substrate through the mem-brane against an electrochemical gradient by using a source of free energy. They are broadlyclassified as primary and secondary active transporters, depending on the free energy source(Mitchell 1967).
Primary active transporters harness chemical reactions (e.g., phosphory-lation by ATP) or light. Some examples are the sodium-potassium pump (Na/K ATPase)(Morth et al. 2007), the rotary F F -ATPase and the light-driven proton pump bacteri-orhodopsin (Buch-Pedersen et al. 2009), complex I in the respiratory chain (Sazanov 2015),or ATP-driven ABC transporters such as p-glycoprotein (Lespine et al. 2009). Secondary transport is driven by an electrochemical gradient in a driving ion , namelysodium or protons. Examples are neurotransmitter transporters SERT (serotonin) andDAT (dopamine) (Gouaux 2009), sodium-proton exchangers (NHE) (Fuster and Alexan-der 2014), the calcium exchanger (Ottolia et al. 2007), and AE1, the anion exchanger in redblood cells also known as Band 3 (Västermark et al. 2014)). Secondary active transporterscan be divided into two classes based on their physiological behavior (Mitchell 1967).
Sym-porters move their substrate in parallel with the driving ion (Figure 1A). Both driving ionand substrate are bound at the same time and move in the same direction. One part of thetransport cycle consists of the movement of the substrate- and ion-free (apo) transporter. (cid:105) “ms” — 2019/12/16 — 1:47 — page 3 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 3 insideoutsideinsideoutside “driving ion” substrate
A B
Figure 1. Schematic transport cycle of A a symporter (transport of substrate and drivingion in the same direction) and B an antiporter (transport in opposite directions). The cen-tral cartoon summarizes the physiological function. The V-shaped triangle symbolizes amembrane-embedded transporter protein in the outward facing conformation in which itsbinding sites are accessible from the outside. The hat-shaped triangle indicates the inwardfacing conformation. The driving ion is drawn as a circle while the transported substrate isshown as a square. The predominant direction of reactions is shown by arrows, with hori-zontal arrows indicating binding/unbinding and vertical arrows conformational transitions.The order of binding and unbinding events and the stoichiometry of substrate to drivingions may differ from this cartoon.In the antiporter transport cycle (Figure 1B), the driving ion is bound during one leg ofthe cycle while the substrate is bound in the other leg and is transported in the oppositedirection.Variations of the above scheme are common, though. For instance, many symporterstransport another ion back instead of the apo leg of the transport cycle. Sometimes, thedriving ion is effectively part of the substrate as in the AdiC transporter (Fang et al. 2009),which exchanges L-arginine with its decarboxylated product agmatine to effectively exportprotons.A third class of related transporters consists of non-coupled transporters. These uni-porters facilitate diffusion through the membrane. Although we specifically focus on activetransporters, the discussion on transport cycles (Section 6.) will make clear that the uni-porters are closely related to active transporters and it is plausible that small changes in theprotein may convert between the two.In this chapter we focus on overarching principles that are common across almost allsecondary active transporters. We begin with the alternating access model which providesthe “standard model” for explaining transporter function in a structural context (Section 2.).Although evolution always finds ways to add a few exceptions to common rules (for in-stance, there are a few transporters that do not appear to follow the classical alternatingaccess model), the physical principles under which transporters operate are not negotiable.Transporters function out of equilibrium as “physical enzymes” that catalyze transport byfree energy transduction through cyclic processes (Section 3.). Ten years ago, a remarkable (cid:105) “ms” — 2019/12/16 — 1:47 — page 4 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
2. The alternating access model
The alternating access model in its basic form was described by Jardetzky (1966) as a poly-mer molecule that contains binding sites for substrate and is able to assume two differentconformations that alternatingly expose the binding sites to the interior and the exterior. Theidea of a cyclical process facilitated by a molecule that changes accessibility was expressedby Mitchell (1967) in his “circulating carrier” model. Together these models describe inabstract terms a basic framework or model to understand driven transport across the cellmembrane. The key insight was that coupling of two fluxes (substrate and driving ion)could be accomplished by binding to different conformations of the same molecule as dis-cussed in more detail in the next Section 3.. In particular, it is physically not possible tomove substrate against a gradient through a continuous pore, i.e., one that is simultaneouslyaccessible from both sides, regardless of any energy consuming mechanism to open or closethe pore (Tanford 1983). The consequence of this insight is that transporters cannot functionif continuous pores are formed through the membrane. The alternating access model withits two distinct states provides a conceptual framework that avoids pore formation. How-ever, it requires that a membrane protein is able to change between different conformationson the sub-millisecond time scale , a speed that is easily achievable for macromolecularconformational changes (Henzler-Wildman and Kern 2007; Schwartz and Schramm 2009).The alternating access model also does not give any insights into the actual molecular struc-ture of a transporter except the general requirement that substrate and ion binding sites mustswitch accessibility in different conformations. In order to obtain deeper mechanistic in-sights actual atomic-scale structures of transporters in multiple conformations are needed.The first secondary active transporter for which the major states in the transport cy-cle were resolved at atomic resolution was the sodium-coupled symporter Mhp1, a mem-ber of the nucleobase-cation-symporter 1 (NCS1) family (Weyand et al. 2011; Jackson et
1. Turnover numbers of transporters range from one transport event per millisecond (for sodium/protonexchangers) to one per second (some amino acid/cation transporters from thermophiles operating at room tem-perature) with a typical number on the order of one hundred events per second (e.g., lactose permease); see“transporter turnover rate” in the BioNumbers database https://bionumbers.hms.harvard.edu/ (Milo et al. 2010). This means that any step in the transport cycle, including the conformational transition,must be faster than 1 millisecond for the fastest transporters, and 10 milliseconds or 1 second for the slowerones. (cid:105) “ms” — 2019/12/16 — 1:47 — page 5 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 5
A B C
Figure 2. Conformations of the nucleobase/sodium-coupled symporter Mhp1 from X-ray crystallography. A Outward-facing open conformation (PDB ID 2JLN) (Weyand etal. 2008). B Outward-facing occluded conformation with bound substrate benzylhydantoin(PDB ID 4D1B (Simmons et al. 2014); this structure superseded the original 2JLO struc-ture (Weyand et al. 2008) but the structural differences are small). C Inward-facing openconformation (PDB ID 2X79; (Shimamura et al. 2010)). The approximate position in themembrane is indicated by the gray rectangle in the background. The two cartoons under A and C indicate the two states of the classical alternating access model as used in Figure 1.al. 2013). The structures of wild-type Mhp1 revealed a sodium binding and a substrate bind-ing site deep at the center of the transporter, roughly at the membrane midplane (Weyandet al. 2008). In one structure, these binding sites were accessible from the extracellularside, making this the outward facing (OF) conformation as shown in Figure 2. Shimamuraet al. (2010) managed to crystallize wild-type Mhp1 in an inward facing (IF) conformationin which the binding sites were exposed to the intracellular side. Together they representthe two key conformations required by the alternating access model. A third conformationwas also found: in this occluded conformation the binding sites were not accessible fromany compartment (Weyand et al. 2008; Simmons et al. 2014). The alternating access modeldoes not require such a conformation. As will be argued in Section 5., such occluded statesare a necessary consequence of a molecular architecture in which the alternating accessconformations are formed by gate domains.The hallmark of the alternating access mechanism are relatively large conformationalchanges in the protein conformation and these appear to exist in many secondary trans-porters for which the alternating access mechanism remains the standard structural frame-work in which to understand transporter function (Boudker and Verdon 2010; Law, Mal-oney, and Wang 2008; Forrest and Rudnick 2009; Gouaux 2009; Krishnamurthy, Piscitelli,and Gouaux 2009; Abramson and Wright 2009; Boudker and Verdon 2010; H. RonaldKaback et al. 2011; Forrest, Krämer, and Ziegler 2011; Schweikhard and Ziegler 2012;Henzler-Wildman 2012; Yan 2013; Shi 2013; Slotboom 2014; Diallinas 2014; Li etal. 2015; Drew and Boudker 2016; Bai, Moraes, and Reithmeier 2017; Kazmier, Claxton,and Mchaourab 2017; Henderson, Fendler, and Poolman 2019). (cid:105) “ms” — 2019/12/16 — 1:47 — page 6 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
3. Thermodynamics and cycles
Transport is driven by spontaneous influx of a driving ion. The free energy dissipation fromflowing down its electrochemical gradient is coupled to the vectorial transport of a substratemolecule or ion. Hill (1989) clearly explained the principle of free energy transduction intransporters (and enzymes) through a cyclic process that tightly couples driving ion flux andsubstrate flux. Following his treatment, we will first qualitatively explain how a cyclicalprocess that operates out of equilibrium transduces energy. We will then briefly revisit thethermodynamic driving forces of the process in order to motivate the idea that transportersare really enzymes that catalyze transport.
Consider, for instance, a hypothetical antiporter that uses one driving ion (red circle inFigure 3A) to move one substrate molecule (blue square). We initially imagine the systemto exist in equilibrium, i.e., the net fluxes between all states are zero, also known as thedetailed balance condition. The inside and outside populations i of particles are in Nernstequilibrium, i.e., when considering the concentrations on either side of the membrane andthe membrane potential, no net flux of particles would occur if a pore selective for species i were opened in the membrane. For example, the binding of a driving ion to the transporterin the outward facing conformation is the equilibrium reaction T + I (cid:10) T : I (1)and the isomerization between outward facing and inward facing conformation (the alter-nating access transition) is T : I (cid:10) T : I. (2)
2. It is also necessary that the transporter state populations are in equilibrium with each other and the ionand substrate concentrations. However, any imbalance in the transporter populations would soon move towardsthe equilibrium values, provided that the ion and substrate concentrations are at equilibrium. A transportcycle cannot be driven by the transporter; only the binding/dissociation of external ions and substrates cancontinuously draw on a source of free energy. (cid:105) “ms” — 2019/12/16 — 1:47 — page 7 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 7 insideoutsideinsideoutside : Transport cycle
A B
Figure 3. Transport by an antiporter is a cyclical out-of-equilibrium process. A Equilibrium—all concentrations are at equilibrium values and all reactions obey detailedbalance. B Out-of-equilibrium—the outside ion concentration is raised over its equilibriumvalue, which leads to moving all states out of equilibrium. The states are numbered so thatone can refer to, say, the outward facing apo state (neither ion nor substrate bound) of thetransporter as T or the inward facing, substrate-bound state as T : S where the presenceof the substrate is included for clarity even though the label “5” includes the presence of thesubstrate (as opposed to state 4, which does not include it).Because all individual fluxes are zero, no net transport takes place. On average, for everysubstrate molecule that is moved from inside to outside in a given unit of time, the samenumber of molecules are moved from the outside to the inside.We now perturb the system away from equilibrium by increasing the outside concen-tration of the driving ion, as indicated by the larger number of driving ions in Figure 3B.Following Le Chatelier’s principle, the equilibrium of the binding reaction Eq. 1 is movedas to increase the concentration of products (Dill and Bromberg 2003), i.e., the number ofion-loaded transporters T : I increases above its equilibrium value. Because the reactants(inputs) of the isomerization reaction Eq. 2 are provided by the products (output) of thebinding reaction Eq. 1, which have increased, Le Chatelier’s principle equally applies tothe isomerization and pushes this equilibrium towards the ion-loaded inward facing confor-mation, T : I . The same reasoning is applied to each subsequent reaction and in this way,net flux of substrate from the inside to the outside is induced in steps 5 →
6. Crucially, thereactions form a cycle so that after the steps 1 → → → → → → I out + S in → I in + S out . (3)
3. Different stoichiometries require different stoichiometric coefficients. For instance, a 2:1 antiporter wouldbe described with I out + S in → I in + S out . (cid:105) “ms” — 2019/12/16 — 1:47 — page 8 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) I out + S out → I in + S in . (4)The interested reader is referred to Hill (1989) who makes the above reasoning quan-titative by considering how the net fluxes between states, which are zero in equilibrium,become biased in one direction when a component is perturbed. Based on reaction kineticshe develops a theory of cycle fluxes that can be applied to arbitrarily complex cycles to com-pute steady state populations and fluxes. In particular, more realistic transporter schemescontain additional transitions such as the one between the two apo states (cid:10) , often re-ferred to as leaks or slippage. Such a transition would allow three cycles to become possible:The transport cycle that was just described and two leak cycles: cycle (cid:10) (cid:10) (cid:10) (cid:10) would dissipate the ionic gradient. Cells spend a substantial amount of their chemical en-ergy to establish the driving ion gradient. In mammals an estimated 19%–28% of ATPare used to power the Na + -K + -ATPase that establishes the transmembrane sodium gradient(Rolfe and Brown 1997). Therefore, dissipation of the sodium gradient is costly and reducesthe organism’s fitness. The other leak cycle (cid:10) (cid:10) (cid:10) (cid:10) would run in the oppositedirection and let substrate molecules enter the cell, counter to the physiological necessity ofthe transporter to remove them from the cell. Under physiological conditions, leak cyclesmust be suppressed by decreasing the rate for slippage transitions such as (cid:10) .The qualitative discussion makes clear that energy transduction, i.e., the use of the freeenergy stored in the driving ion gradient, requires a complete cycle that contains both ionand substrate translocation steps. If any part of the cycle is broken, no energy transductioncan take place. Thus, energy transduction is a property of complete cycles and not ofindividual states (Hill 1989). Therefore, there is no specific step in the cycle that could bedescribed as an “energized” state or a state where energy is “gained by a binding reaction”(Hill and Eisenberg 1981).In general, a protein that functions according to the alternating access mechanism can-not function if it presents a continuous, leaky pathway (Tanford 1983) as this preventsenergy coupling. Similarly, non-productive leak cycles also reduce the efficiency of a trans-porter. Although here we generally discuss ideal, fully efficient cycles to elucidate the basicprinciples, real transporters leak and therefore their transport stoichiometry is generally notthe ideal one (Hill 1989; Henderson, Fendler, and Poolman 2019). For example, instead ofan ideal 1:1 stoichiometry one might measure only 1:0.75, i.e., on average 1.33 driving ionsare needed to move one substrate because only 75% of the total flux comes from productivecycles (1:1 stoichiometry) and 25% comes from leak cycles (1:0).The ion and substrate binding or dissociation steps are necessary components of thecycle because without them the cycle cannot be driven in a specific direction: these stepsprovide the only external “handle” to control the process (Zuckerman 2019). Therefore,no cyclical process with a net flux in one direction exists in which only a protein changesthrough a repeated sequence of conformational states; coupling to an external source of freeenergy is always necessary.Finally, it is worth emphasizing that because the transporter protein moves cyclicallythrough different conformations, it is not altered in any permanent way. In the energeticdescription of the process (see Section 3.2. below), the transporter does not appear. Thus, (cid:105) “ms” — 2019/12/16 — 1:47 — page 9 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 9transporters act as enzymes for moving substrate, similar to how biochemical enzymes cat-alyze the formation and breaking of chemical bonds. In this sense, transporters are “phys-ical enzymes” or “molecular machines” in that they catalyze a physical process instead ofa chemical one. Other proteins of this kind are molecular motors, which turn chemicalenergy into movement of the protein itself, or rotary pumps such as the V-type and P-typeATPases, which turn chemical energy into rotary motion and movement of protons or ionsacross the cell membrane; the latter can run in reverse to turn rotary motion by ion flow intochemical bonds. Similarly, transporters run backwards if the concentrations are changedappropriately, which becomes obvious when analyzing the thermodynamic driving forces.
Quantitatively, the only thermodynamic driving forces X i are the ones originating in elec-trochemical potential ( µ (cid:48) = µ + kT ln c/c + q Ψ ) differences of ions and substrates acrossthe cell membrane (Dill and Bromberg 2003); free energy differences due to the differentstates of the protein cancel in the whole cycle and play no role (Hill 1989). The drivingforce for species i ∈ { I, S } is X i = µ (cid:48) i, in − µ (cid:48) i, out = kT ln c i, in c i, out + q i ∆Ψ (5)where c i is the concentration (or activity) on the indicated side of the membrane, q i thecharge, ∆Ψ = Ψ in − Ψ out is the transmembrane potential, T is the temperature and k isBoltzmann’s constant. The membrane potential is typically negative, ∆Ψ < .Thus, fortypical driving cations (Na + , H + with q = +1 e ) and ∆Φ ≈ − mV the membranepotential contributes at T = 298 K about q I ∆Ψ ≈ − . kT . Typical sodium concen-trations are on the order of mM on the outside and mM inside a cell and hence kT ln c I, in c I, out = − . kT . If the substrate is neutral (the electrostatic component is zero for q S = 0 ) then a positive net charge is moved into the cell down an electrostatic potential anda sizable fraction of the available free energy will be provided from the membrane potentialcomponent. In general, any electrogenic transport (movement of a net charge) is affectedby the membrane potential.Denote by J i the flux at which particle i is transported across the membrane (in particlesper unit time), with the direction out → in counting as J i < and the reverse as J i > . Notethat in a simple cycle such as the one in Figure 1, exactly one ion is moved for each substratemolecule and hence the absolute values of these fluxes must be the same, | J I | = | J S | butthe signs will differ, depending on symport or antiport processes.When the driving force is negative, e.g., X I < , then spontaneous movement occurs,such as influx of the driving ion and hence J I < . The antiporter is supposed to movesubstrate against a gradient from the inside to the outside, i.e., against the opposing drivingforce X S < under which S particles would spontaneously move into the cell. The rate offree energy dissipation is Φ = J I X I + J S X S ≥ . (6) Φ = 0 holds in equilibrium but then no transport occurs (see Section 3.1.). The secondlaw of thermodynamics requires Φ > in non-equilibrium steady state, i.e., when con-centrations remain fixed at their non-equilibrium values and do not change (Hill 1989). In (cid:105) “ms” — 2019/12/16 — 1:47 — page 10 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
10 O. Beckstein and F. Naughtonsteady-state, the transporter moves ions and substrates at a constant flux. Under which con-ditions will the antiporter move S from inside to outside, i.e., given J S > (even though X S < ), what is required of I ? Rearranging Eq. 6 J I X I > − J S X S (7)and noting that the right hand side is positive, it follows that J I X I also has to be positive,i.e., the driving ion must flow down its electrochemical gradient from the outside to theinside ( J I < , X I < ). In other words, spontaneous fluxes always dissipate free energy,which can be coupled to the substrate flux. This free energy dissipation rate must be largerthan the rate of free energy required to move S against its driving force. For a simpleantiporter cycle without leakage, J S = − J I (for each I transported to the inside, one S istransported to the outside, in the same amount of time) and hence − J S X I > − J S X S andwith J S > , X I < X S (simple antiporter) (8)is required for transport. The amount of available free energy per driving ion translocationevent must be larger (more negative) than the substrate gradient against which S is movedbecause out of equilibrium not all free energy can be transformed into useful work and afraction always increases the entropy of the universe in the form of heat, as required bythe second law. The condition Eq. 7 can also be fulfilled with J I > , X I > , i.e., aspontaneous flux of ions from the inside to the outside. In this case the transporter wouldneed to be able operate as a symporter to move driving ion and substrate together (it cannothappen in separate cycles (Hill 1989)).For a simple symporter with X I < , X S > and J I = J S < , the conditionequivalent to Eq. 8 reads X I < − X S (simple symporter) . (9)We will come back to the question of the relationship between symporters and antiporters inSection 6. where we will see that one can write a universal kinetic scheme that encompassessymporters, antiporters, and uniporters.
4. Inverted repeat symmetry
The alternating access model together with the thermodynamic cycle analysis explains howtransporters function in principle, i.e., they describe the physical constraints under whichany transporter protein has to operate. However, understanding how these principles areembodied in an actual biomolecule requires structural atomic-resolution data, primarilyprovided by X-ray crystallography and electron microscopy. The most important require-ment of the alternating access model is the existence of two states that make binding sitesaccessible to the outside or the inside, generally referred to as an outward facing (OF) con-formation and an inward facing (IF) conformation . It turns out that a viable evolutionarypath to create a switchable two-state system in a single protein molecule can be based onan internal two-fold structural symmetry, so-called inverted repeats . This deep insight into (cid:105) “ms” — 2019/12/16 — 1:47 — page 11 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 11a fundamental principle of transporter function was only recently discovered by Forrestet al. (2008) and since then broadly recognized as nearly universal (Lucy R Forrest 2013,2015).
Structural symmetry is well represented among membrane proteins. These symmetries canarise due to oligomerization, often seen in cyclically symmetric channels and pores, ordue to the presence of internal repeats within the protein sequence (Lucy R. Forrest 2015).Notably, internal repeats occur more frequently in membrane protein super-families thanoverall (Myers-Turnbull et al. 2014); in the case of secondary active transporters, mostknown structures show an inverted repeat symmetry (Bai, Moraes, and Reithmeier 2017;Shi 2013)—that is, internal repeats which adopt similar folds but start on opposite sidesof the membrane, giving rise to C pseudosymmetry about an axis parallel to the mem-brane plane. Further common to most secondary active transporters is the presence of twobundles or domains, with the substrate binding sites located near the interface and ofteninvolving discontinuous helices (Shi 2013). The exact number of transmembrane helicesand distribution of the inverted repeats over the two domains differs, with several commonfolds observed (Figure 4): (3 + 3) + (3 + 3): Major Facilitator Superfamily (MFS). The core MFS fold contains12 transmembrane helices (TMs) (Law, Maloney, and Wang 2008; Yan 2013). An N-and C-domain are each formed from a pair of 3 TM inverted repeats, and are themselvestwo-fold pseudosymmetric. The MFS is one of the largest transporter families found acrossmultiple organisms; the first structures were reported for lactase permease LacY (Abramsonet al. 2003) and glycerol-3-phosphate transporter GlpT (Huang et al. 2003), with manydetermined since. (5 + 5): LeuT fold.
The LeuT fold consists of 5 TM inverted repeats, with the first twohelices from each repeat forming a core bundle (i.e. TMs 1, 2, 6, 7), while the next two(TMs 3, 4, 8, 9) form a scaffold/hash domain; TMs 5 and 10 may act as gates (Kazmier,Claxton, and Mchaourab 2017). First observed in the neurotransmitter/sodium symporterLeuT (Yamashita et al. 2005), several other transporters have been found to adopt this fold,including the sodium/hydantoin transporter Mhp1 (Weyand et al. 2008). (5 + 5): NhaA fold.
Also consisting of 5 TM repeats, the NhaA fold is observed insodium/proton antiporters (e.g. NhaA (Hunte et al. 2005)) and the apical sodium dependentbile acid transporter (ASBT) family (e.g. ASBT NM (Hu et al. 2011)). The first two helicesof each repeat (TMs 1, 2, 6, 7) form a panel or dimer domain, with the remaining three(TMs 3-5, 8-10) forming a core domain (Fuster and Alexander 2014; Padan 2014). (7 + 7): 7-TM inverted repeat (7TMIR) fold. Relatively recently identified, transporterswith this fold include the proton/uracil symporter UraA and chloride/bicarbonate antiporterAE1; four helices from each repeat (TMs 1-4, 8-11) for a core domain, while the remainingthree (TMs 5-7, 12-14) form a gate domain (Chang and Geertsma 2017). (cid:105) “ms” — 2019/12/16 — 1:47 — page 12 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
12 O. Beckstein and F. Naughton A B C + ++ ++ Figure 4. Representative structures of common secondary active transporter folds, high-lighting the inverted repeats. In each case, the protein is first shown whole, then with therepeats separated translationally, and finally with the second repeat rotated and overlayedon the first. A : LacY (inward facing, MFS fold; PDB ID IPV6) B : Mhp1 (outward facing,LeuT fold; PDB ID 2JLN) C : ASBT NM (inward facing, NhaA fold; PDB ID 3ZUY).Internal repeats such as there have been speculated to arise from the duplication of anancestor gene and subsequent fusion event, in this case following a flip of one duplicaterelative to the membrane; possible candidates showing these initial “half” folds have beenidentified in the DedA (for the LeuT fold) and SWEET (for the MFS fold) families (Keller,Ziegler, and Schneider 2014). The EmrE multidrug transporter is proposed to come togetheras an antiparallel dimer and function through an exchange of asymmetrical structures sim-ilar to that described below (Korkhov and Tate 2009; Morrison et al. 2012), and representsa possible pre-fusion step in the proposed duplication-and-fusion evolutionary process ofinverted repeat symmetry.Distinct inward- and outward-facing conformations arise from asymmetry in the exactfolds of the repeats composing the two domains (discussed in Section 4.2.), which changesthe relative locations/orientations of these domains. Several mechanisms for this relativemotion have been proposed (Drew and Boudker 2016; Forrest, Krämer, and Ziegler 2011):the domains may rotate about the substrate binding site to alternatively expose it to each sideof the membrane, as in the rocker-switch (where the domains are structurally symmetric,proposed for e.g. for MFS transporters (Radestock and Forrest 2011)) and rocker-bundle (cid:105) “ms” — 2019/12/16 — 1:47 — page 13 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 13 Figure 5. Cartoon showing how symmetry-broken inverted repeats generate the two majorconformations in the alternating access mechanism. Each repeat (labeled A and B) maytake on one of two conformations (shown as 1 and 2 in the inset), giving rise to 2 x 2 = 4possible conformations, though two (occluded and “leak” states, shown inset) are not partof the basic alternating access mechanism.(where the domains are distinct, e.g. for LeuT fold transporters (Kazmier, Claxton, andMchaourab 2017)) mechanisms; or, as in the elevator mechanism , one domain predomi-nantly containing the binding site may move perpendicular to the membrane, moving thebinding site against to relatively fixed second domain to expose it to each side of the mem-brane in turn (proposed for e.g. for NhaA fold transporters (Padan 2014)).
While inverted repeats share an overall fold, they are found to take on different conforma-tions, giving rise to an asymmetry that allows the substrate binding site to be exposed to oneside of the membrane while blocked from the other. By exchanging conformations betweenthe two repeats—with the first repeat adopting the conformation of the second and viceversa—the protein is thus able to switch between an inward facing and an outward facingstate (Forrest, Krämer, and Ziegler 2011) (Figure 5).Exchanging conformations in this way means there is no or little net energetic changein the overall protein structure from the inward-to-outward or outward-to-inward transitions(Lucy R. Forrest 2015). The presence of two possible conformations for each of the tworepeats also brings up the question of whether the repeats can possess the same conforma-tion at a given time; such overall conformations might form occluded (closed at both side)or leaky (open on both sides) states of the transporter (Figure 5; inset). The presence ofoccluded and leak states in the transport cycle is discussed further in Section 5..The above described repeat swapping has been taken advantage of to generate homol-ogy models of transporters in different states, given a structure in only one state: the con-formation of each repeat is used as a template for the other, forcing the exchange of con-formations (Vergara-Jaque et al. 2015). This method was first applied to LeuT (Forrestet al. 2008), producing a structure that latter proved to be consistent with an experimentalstructure (Krishnamurthy and Gouaux 2012), and has since been used to generate struc-tures for a range of secondary active transporters, with subsequent experimental validationobtained in several cases; including the glutamate transporter GltPh (Crisman et al. 2009; (cid:105) “ms” — 2019/12/16 — 1:47 — page 14 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
14 O. Beckstein and F. NaughtonReyes, Ginter, and Boudker 2009), LacY (Radestock and Forrest 2011), CcdaA (Zhou andBushweller 2018), and NhaA (Schushan et al. 2012).
5. Transporters as gated pores
Läuger (1980) envisaged ion channels as pores with a single free energy barrier, which canbe identified with the gate of the channel that controls ion flow in response to external stim-uli (Hille 2001). He could model transporters as pores with two coupled barriers (Läuger1980); motivated by the suggestive original “gated pore model” (Klingenberg 1979) (whichwas more precisely renamed the “single binding center gated model” (SBGP) (Klingenberg2007)) one may also call these two barriers gates (Figure 6) (Abramson and Wright 2009;Krishnamurthy, Piscitelli, and Gouaux 2009; Forrest, Krämer, and Ziegler 2011). Such agate should be thought of as a switch or bi-stable element that can exist in two states thatare generally called “open” and “closed” although it might also carry the meaning “outwardfacing” or “inward facing”. The gate picture reduces the nuanced view of free energy bar-riers with variable barrier height to one in which either a very high barrier exists (“closed”)or the barrier is small compared to thermal fluctuations (“open”). This simplification allowsone to broadly enumerate and categorize states and make general (but necessarily approxi-mate) statements for whole classes of proteins. It also allows one to create simple cartoonsof transporter states that summarize transporter conformations succinctly. As will be shownbelow, the cartoon is a useful abstraction because gates correspond to physical moleculardomains in transporter proteins (i.e., they have a molecular identity) and thus states gener-ated from the gate picture directly correspond to observable protein conformations.In ion channels, a change in membrane potential or binding of a signaling moleculeopens the gate and ions spontaneously flow through the open pore down their electrochem-ical gradient (Hille 2001), as shown for a pore with N = 1 gate in Figure 6. In the“transporters as gated pores” picture, the coordinated movement of two (or more) gatescreates the conformations of the alternating access model (Section 2.). The outward fac-ing open state is formed when the outer gate opens while the inner gate remains closed.Conversely, the inward facing state consists of the outer gate closed, while the inner oneis open, as shown for a transporter with N = 2 gates in Figure 6. Simultaneous openingof both gates must be avoided—a “leak” state (see Figure 6)—to prevent leakage of thedriving ion and dissipation of the ionic gradient. The coordination of the gates is termed“coupling”. Furthermore, conformational changes (i.e., changes in the gates) must also becoupled to the binding/dissociation of driving ion(s) and substrate(s), a point that will berevisited below and explicitly included in Section 6.. For simplicity, we will focus on thedifferent conformational states of the protein while keeping in mind that the presence ofions and/or substrates will likely change the structure to some degree.It has been observed experimentally that under certain conditions transporters can alsofunction as channels (DeFelice and Goswami 2007), a view that fits naturally in the picture
4. Describing a channel with a single gate is an oversimplified picture because many ion channels contain anactivation gate and a separate inactivation gate, which switches the channel into a relatively long lasting inactivestate (Hille 2001). These gates are typically coupled to some degree, e.g., the inactivation gate might only closewhen the activation gate has closed but the coupling is presumably different from the coupling between gatesthat enables energy transduction in transporters. (cid:105) “ms” — 2019/12/16 — 1:47 — page 15 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 15 N open closed outsideinside = 2 occluded OF open IF open (leak) (leak) = 82 = 4 ion channelstransporters Figure 6. Ion channels and transporters as gated pores. In this simplified picture, ion chan-nels contain a single gate that is controlled by external stimuli. Transporters can implementalternating access by the coordinated movement of two or more gates. The number of gates N determines the total number of distinct states, N . Not all states might be physiologicallyobserved, and some, such as the leak states, will prevent energy transduction.of a transporter whose gates are not fully coordinated so that leak states may occur (see Fig-ure 6) although many questions remain in this somewhat under-explored area. Channel-likebehavior is characterized by spontaneous energetically downhill diffusion of ions (or sub-strates) in a non-stoichiometric and burst-like fashion, which differs from the leak cyclesthat can occur in the standard cycle due to loose coupling (Henderson, Fendler, and Pool-man 2019) in that the latter still only move individual particles. Even for the channel-likebehavior of transporters, it is not necessarily clear that the canonical transport pathway isused. It is possible that alternative pathways through the protein are responsible (Vanden-berg, Huang, and Ryan 2008).Nevertheless, the “transporters as gated pores” picture is more than just a convenientcartoon model because as we will discuss below, the gates generally represent a molecu-lar reality, i.e., secondary transporters contain distinguishable parts that function as gates.Therefore, conformational states that are predicted from the gated pore model generally cor-respond to conformations with distinct structural arrangements of the corresponding gates. Since X-ray crystallography has revealed the molecular structures of a range of transporters,various authors have identified domains of these proteins that regulate access to bindingsites with gates, as summarized by Forrest, Krämer, and Ziegler (2011). A particular termi- (cid:105) “ms” — 2019/12/16 — 1:47 — page 16 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
16 O. Beckstein and F. NaughtonFigure 7: The three gates inMhp1. A : Mhp1 in the membrane.The hash motif (helices 3, 4, 8,9) is shown in yellow, the bundle(helices 1, 2, 6, 7) in red, flex-ible (thin gate) helices 5 and 10in blue, and C-terminal helices 11and 12 in gray. The views on thegates (B–D) are indicated by bro-ken rectangles. B : extracellularthin gate (formed by TM10). C :thick gate, quantified by the dis-tance across the Na2 sodium bind-ing site. D : intracellular thin gate(TM5). A48 T361
T162 I230
A38 S312 T313 I41 A309
A BC DBDC nology of thin and thick gates originated in the structural analysis of LeuT-like transporters(Krishnamurthy, Piscitelli, and Gouaux 2009; Abramson and Wright 2009). Thin gates aregenerally considered to be parts of the protein whose movement can prevent the exchangeof ions or substrates with the intra- or extracellular solution. Perhaps somewhat confus-ingly, the conformational transition that is responsible for alternating access, or rather thesum of moving structural elements, is sometimes considered the thick gate. In other trans-porter families, such as the MFS transporters, no special distinction between thin and thickgates is commonly made.Although there is some ambiguity in how to define gates, they are nevertheless recogniz-able molecular entities. Diallinas (2014) concludes, based on work in the purine transporterUapA, that physiological transport properties are determined by intramolecular interactionsbetween binding sites and gating elements, similar to ones present in channels. LeVine etal. (2016) quantitatively analyzed the mechanism of the LeuT transporter with a particularemphasis on the allosteric coupling between ions, substrate, and the protein. Based on ex-perimental and simulation data, they concluded that LeuT is best described with a allostericgated pore alternating access mechanism in which gate movement is strongly coupled tobinding and the other gates.We will illustrate the physical reality of gates in transporters in an example, theMhp1 transporter (a member of the LeuT-like family of APC (amino acid-polyamine-organoCation) transporters (Västermark et al. 2014)), and in Section 5.2., where also mem-bers of the major facilitator superfamily (MFS) (Marger and Saier 1993) will be included.
Thin and thick gates in Mhp1
The hydantoin permease Mhp1 from
Microbacterium liquefaciens is a nucleobase-sodiumsymporter (Suzuki and Henderson 2006), a member of the NCS1 family. It co-transports (cid:105) “ms” — 2019/12/16 — 1:47 — page 17 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 17one sodium ion with one 5-substituted hydantoin. It shares a five-helix inverted repeatarchitecture with other members of the superfamily of LeuT-like transporters (Cameron,Beckstein, and Henderson 2013). X-ray crystallographic structures of Mhp1 in outwardfacing and inward facing conformations together with computer simulations revealed thestructural basis for the alternating access mechanism in this secondary transporter (Weyandet al. 2008; Shimamura et al. 2010; Simmons et al. 2014). The transporter can be understoodas a gated pore with two thin and one thick gate (Krishnamurthy, Piscitelli, and Gouaux2009), i.e., as a pore with three gates: The thick gate regulates the passage through thecenter of the membrane by means of the large conformational change that switches thetransporter from its outward facing to its inward facing conformation. In Mhp1 it consistsof the hash motif (formed by helices TM3, TM4 and their inverted-repeat counterparts TM8and TM9; see Figures 7A, C and 4B) that can rotate by about ◦ on an axis parallel to TM3relative to the four-helix “bundle” (TM1, TM2 and TM6, TM7) (Shimamura et al. 2010).Thin gates are formed by the N-termini of the pseudo-symmetry related helices TM5 andTM10 and the linker to each preceding helix (Shimamura et al. 2010). The extracellular(EC) thin gate (TM10; Figure 7B) governs access to the substrate binding site from theperiplasmic medium while the intracellular (IC; Figure 7D) gate fulfills the symmetricalrole of controlling the pathway to the cytosol.The sodium binding site is formed between bundle and hash, so opening of the thickgate, i.e., the alternating-access transition, opens up the sodium binding site and weakension binding to ensure rapid diffusion of the ion into the cytosolic compartment and openinga pathway for the substrate to follow (Shimamura et al. 2010).The role of the thin gates in Mhp1 appears to be more subtle (and might well differfrom the role of thin gates in related NSS-like transporters such as LeuT (Kazmier, Sharma,Islam, et al. 2014; Kazmier, Sharma, Quick, et al. 2014; Kazmier, Claxton, and Mchaourab2017). In Mhp1, the whole N-terminus of TM10 moves together with the linker betweenTM9 and TM10 (Weyand et al. 2008; Shimamura et al. 2010; Kazmier, Sharma, Islam,et al. 2014), thus forming a distinct gate structure that is mirrored in TM5 and the TM 4-5linker, which are related to TM9/10 through the inverted-repeat symmetry as described inSection 4. and shown in Figure 4B.As discussed in Section 2., a protein that functions according to the alternating accessmechanism cannot function if presents a continuous, leaky pathway (Tanford 1983) [or if itallows too many non-productive leak cycles to occur (Section 3.)]. The EC gate appears toprevent Mhp1 from leaking the driving ion, Na + , as demonstrated by modeling: Figure 8Bshows that in the inward facing conformation, with the EC gate closed, the solvent accessi-ble surface only extends from the IC side into the binding site at the center of the protein.However, when the EC gate is removed (the atoms were deleted from the structure as a sim-ple model of a hypothetical inward facing conformation with an open EC gate) a pathwayopens up through the membrane (Figure 8D). The calculated electrostatic solvation free en-ergy (Born energy) in the volume of the pathway shows that Na + ions could traverse themembrane because a low-energy path is visible (Figure 8D). On the contrary, in all otherstates, no low energy path can be found for a sodium ion (Figures 8A–C) because either thethick gate or the EC gate blocks the passage. Thus, the EC gate fulfills an important rolein preventing a sodium ion leak. It is required because the switch in the thick gate does notjust simply change the conformation from outward to inward facing but it really acts as a (cid:105) “ms” — 2019/12/16 — 1:47 — page 18 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
18 O. Beckstein and F. Naughton intracellularextracellular extracellularside kJ/mol intracellularextracellular extracellularside
IC gate EC gate
A BC D
Figure 8. The putative role of the extracellular (EC) gate of Mhp1 is to prevent a sodiumleak when the thick gate is open. The left four panels show a cut through the electro-static solvation free energy (Born energy) landscape of a Na + ion inside Mhp1 for differentconformations of the transporter, computed with the Poisson-Boltzmann equation. Red( − kJ · mol − ) to yellow ( +10 kJ · mol − ) regions can be considered accessible for sodiumions under typical conditions. (The Na + Born energy was calculated as described previ-ously (Stelzl et al. 2014).) The gate cartoons in A – C represent some of the states for thetriple-gated transporter in Figure 6; the cartoon in D symbolizes a leaky state that was ar-tificially modeled by removal of the EC-gate portion of TM10. A : Outward facing openconformation (EC gate open, thick gate closed). B : Inward facing open conformation (ECgate closed, thick gate and IC gate open). The solvent accessible surface (cyan) is shownfrom the side in the context of the protein helices (view on the surface from the top). Thecolor scheme for the helices is the same as in Figure 7, except that the N-terminal half ofthe EC gate (TM10) is shown in black. The closed EC gate prevents a continuous sodiumpathway. C : Outward facing occluded conformation (EC gate closed, thick gate closed). D :Simple model for a hypothetical inward facing open, leaky conformation with IC gate andthick gate open and EC gate removed. The solvent accessible surface representation and theelectrostatic free energy show a sodium pathway through the membrane-spanning portionof the transporter. (cid:105) “ms” — 2019/12/16 — 1:47 — page 19 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 19gate that is open for inward facing and closed for outward facing.Additionally, the EC gate is involved in substrate selectivity (Simmons et al. 2014).Mhp1 transports 5-substituted hydantoins where the substituent must be a bulky hydropho-bic moiety such as a benzyl or indolylmethyl group. However, if the 5-substituent istoo voluminous such as a naphtyl group, transport is inhibited even though the moleculebinds tightly. A crystal structure of outward-facing Mhp1 with 5-(2-naphthylmethyl)-L-hydantoin bound revealed that the EC gate was trapped in an open conformation due to asteric clash of the naphtyl ring with Leu363 (Simmons et al. 2014). A Leu363Ala mutantof Mhp1, which removes the clash, was competitive for transporting 5-(2-naphthylmethyl)- L -hydantoin. These results strongly suggest that closure of the EC gate is required for thealternating access transition to occur. In the language of the gated pore view, the thin ECgate is coupled to the thick gate. Structural comparison suggests that this coupling is dueto the geometrical architecture and the direct connection of the rotating hash motif to theEC gate through the 9-10 linker. The thick gate cannot move into the space occupied bythe open EC gate and therefore is prevented from closing. Conversely, the open thick gateappears to latch the EC gate in its closed position (Shimamura et al. 2010). Thinking of transporters as consisting of N gating elements that can individually switchbetween two states (such as open and closed as in Figure 6) suggests a simple count toenumerate the possible number of conformations of the transporter, n C = 2 N . (10)For a transporter with two gates, four states are possible, and eight states for N = 3 . Thesimple count ignores the fact that gate movement must be coordinated in some fashion.The type of coupling will depend on the individual molecule and may even depend on thesubstrate (Henderson, Fendler, and Poolman 2019) but for canonical transport one mightwant to assume that a leak state with all gates open plays no important role and so n C =2 N − . The primary advantage of such a simple enumeration is to provide a framework in whichto place experimentally or computationally observed conformations. Forrest, Krämer, andZiegler (2011) proposed a similar classification with eight states, consisting of differentconformations and with differing substrate occupancy. Their scheme makes use of thingates but places central importance on the major conformational switch between inwardand outward facing conformations. It has been successfully used to, for instance, categorizethe wealth of structural data for the APC transporter BetP, for which crystal structures havebeen obtained for most of the states (Ressl et al. 2009; Perez et al. 2012; Perez et al. 2014),and to analyze simulated transitions for four LeuT-fold transporters (Jeschke 2013).
5. We equate a state with a conformation of the transporter, assuming that each state is formed by an distin-guishable ensemble of conformers near the specific conformation. This typically implies that there is a kineticseparation between states. Although this is not necessarily always the case in practice, we will nevertheless usestate and conformation interchangeably to keep the discussion simple.6. One could write n C = 2 N − F where ≤ F < N is the number of “forbidden” conformations if oneknew through other means which conformations were not accessible. (cid:105) “ms” — 2019/12/16 — 1:47 — page 20 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
20 O. Beckstein and F. NaughtonAn almost trivial prediction of the gated pore model is the existence of occluded states .In an occluded state, the transporter obtains a conformation in which the binding sites arenot accessible from either compartment. In the doubly-gated pore, the occluded state natu-rally arises when the two gates are closed ( N = 2 in Figure 6). The alternating access modeland the associated kinetic and thermodynamic analysis do not require occluded states forenergy transduction and vectorial transport. Therefore, the existence of occluded states,which are not strictly necessary for function, could be interpreted as a consequence of thestructural constraints of the implementation of alternating access (via inverted repeat sym-metry) in proteins. Below we will show some structural evidence for occluded states. Butit is also noteworthy to point out that molecular dynamics (MD) computer simulations havebeen able to generate occluded states when started from crystallographic conformationscorresponding to inward or outward facing states: For example, Latorraca et al. (2017)simulated the LbSemiSWEET transporter with unbiased MD and observed full transitionsfrom outward to inward facing states that passed through an occluded state. The simulationspontaneously reached experimentally determined structures for inward open and occludedLbSemiSWEET. They found that the transitions were driven by favorable inter-helical in-teractions when either the extracellular or the intracellular gate closed and and unfavorablehelix configuration when both gates were closed. The two gates became tightly coupled,which prevented simultaneous gate opening, which would result in a leak state. Other sim-ulation examples are discussed below, which all point to the insight that molecular gates area simple way to generate alternating-access states. In the absence of specific coupling thatprevents two gates from closing at the same time, occluded states will occur. The MFS transporters all share a common fold with four inverted repeats (Figure 4A).It was originally believed that alternating access would proceed by a rigid body move-ment whereby the two halves of the protein would move relative to each other in a “rockerswitch” manner (Law, Maloney, and Wang 2008). The discussion concentrated on LacYfor which an alternative model described the protein as more flexible, with cytoplasmicand periplasmic openings governing access to the binding site, effectively describing gates(H. R. Kaback et al. 2007). The existence of an occluded state in LacY would corroboratethe gated pore model for its mechanism. Stelzl et al. (2014) hypothesized that LacY func-tioned as a pore with two coupled gates that could both close at the same time. With this as-sumption they could perform biased MD simulations to generate a model of occluded LacYwith both gates closed. The model broadly agreed with experimental electron-electron res-onance (DEER) spectroscopy data. More recently, experimental evidence for an occludedapo intermediate of LacY (based on sugar accessibility to cysteine cross-linked mutant)(Smirnova, Kasho, and Kaback 2018) corroborated the occluded model.The MFS transporter PepT So is a proton-coupled bacterial symporter for which onlyinward facing crystal structures are known (Newstead et al. 2011; Fowler et al. 2015).An open question has been the nature and molecular mechanism of the conformationaltransition between inward and outward conformation. Fowler et al. (2015) used an array ofexperimental and computational techniques, including X-ray crystallography, DEER, andMD, to elucidate the dynamics of PepT So . They found that PepT So is representative for a (cid:105) “ms” — 2019/12/16 — 1:47 — page 21 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 21 periplasmic (EC) min radius (Å) cy t op l a s m i c ( I C ) m i n r ad i u s ( Å ) transporter EmrDFucPGkPOTGlpTGLUT1LacY NarUNRT1.1PepT So PepT St PiPTXlyEYajR periplasmic (EC) gate H1,2,7,8 (Å) cy t op l a s m i c ( I C ) ga t e H , , , ( Å ) transporter EmrDFucPGkPOTGlpTLacYNarU PepT So PepT St PiPTNRT1.1XlyEGLUT1YajR
A B
Figure 9. Conformational state of MFS transporters from a survey of crystal structures. Thegray rectangles approximately indicate classification of the structures as outward open, oc-cluded, or inward open. A Minimal pore radii R computed with HOLE (Smart et al. 1996)near the periplasmic and cytoplasmic entrance. A conformation is classified as occluded ifboth R ≤ . Å. B Gate distances d , calculated as the minimum distance between the C α atoms of the relevant pairs of helix tips H1,2,7,8 (periplasmic gate) or H4,5,10,11 (cytoplas-mic gate) (Fowler et al. 2015). A conformation is classified as occluded for both d ≤ Å.Figure drawn after Fowler et al. (2015).large number of MFS transporters in that the ends of the first two helices in each of the fourinverted repeats (see Figure 4A) form gates at the EC and IC entrance. A wealth of structuraldata exists for the MFS transporters (Yan 2013) with many transporter structures havingbeen solved to atomic resolution in different conformations. Fowler et al. (2015) analyzed33 MFS structures in terms of the minimal pore radius near the periplasmic (extracellular,EC) and cytoplasmic (intracellular, IC) entrance (Figure 9). The structure could neatlybe categorized as outward facing, inward facing, or occluded, based either on geometricalconstriction radius (Figure 9A) or on the distances of the inverted-repeat gates (Figure 9B).Using multiple MD simulation with a Markov state model, Selvam, Mittal, and Shukla(2018) sampled conformational transitions from the inward facing conformation through anoccluded state to an outward facing conformation of PepT So . Their computational resultswere validated by comparison of simulated with experimental DEER spectroscopy data.The occluded state forms by closure of both ends of the protein. In a computed free energylandscape, the occluded state is a stable local minimum, as expected for a thermodynamicstate.Overall, the evidence suggests that MFS transporters can be described as transporterswith two gates. The gates are related to their internal repeat symmetry. Coupling allows oc-cluded states to occur, which were found to exist in crystal structures and MD simulations. (cid:105) “ms” — 2019/12/16 — 1:47 — page 22 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
22 O. Beckstein and F. Naughton outward facing open outward facing occluded inward facing open2JLN2JLO2X79 inward facing occluded open ) MDIC gate( closed ) A B CD
Figure 10. Molecular dynamics simulations sample an inward facing occluded state ofMhp1. A Gate distances (see Figure 7) from 100-ns MD trajectories of Mhp1 with a Na + ion in the Na2 site (simulations from Shimamura et al. (2010) and additional unpublisheddata). Contour lines are drawn at 20% increments of probability density. The shaded areaindicates the full extent of order parameters explored in the simulation. Simulations startedfrom crystal structures (circles) in three different conformational states (represented by thePDB ID), except for the magenta trajectories, which were started from a frame of the inwardfacing open simulation (gray) that showed an almost closed intracellular gate. Data forthree independent simulations each starting from the outward facing states (blue and red)are shown together with one from the inward facing open (gray) and two from the inwardfacing occluded (magenta) states. B, C
Putative inward-facing occluded conformation ofMhp1 (snapshot from the high-probability region of the inward facing occluded trajectory(magenta) in A). The cut through the transporter is shown in the same way as the crystalstructures in Figure 2. D Inward facing open crystal structure (PDB ID 2X79) with the openIC gate.
As an example of a transporter that can be described with three gates we look at Mhp1again. The crystallographic structures in Figure 2 show an occluded state (Weyand etal. 2008; Simmons et al. 2014) in addition to the outward (Weyand et al. 2008) and in-ward facing states (Shimamura et al. 2010) that are necessary for alternating access. Giventhe definition of the three gates (see Figure 7), the crystallographic occluded structure isin an outward-facing occluded conformation because the thick gate is close (outward fac-ing) and both thin gates are also closed. MD simulations had shown that the thin gatescan change conformations on the 100-ns time scale (Shimamura et al. 2010). A detailedanalysis of the simulations in terms of the gate distances (Figure 10A) showed that the ECgate was mobile when the thick gate was closed (outward facing conformation) and thesimulations sampled both outward open and outward occluded conformations. The IC thingate remained locked, though. Conversely, once the thick gate was open (inward facingconformation), the EC gate was locked and the IC gate could sample open and closed con-formations. The dynamic behavior of the thin gates reflects the two-fold symmetry that is (cid:105) “ms” — 2019/12/16 — 1:47 — page 23 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 23imposed by the inverted repeat symmetry (Figure 4B). Using site-directed spin labeling andDEER spectroscopy, Kazmier, Sharma, Islam, et al. (2014) showed that the IC gate formedby TM5 undergoes motions between open and closed conformations, in agreement with thecomputational results.The crystallographic inward facing structure shows the IC gate in a wide open position ,corresponding to an inward open structure. The MD simulations with the thick and both thingates closed showed that a second occluded state might exist (Figure 10B,C), as predictedby a pore with three gates (Figure 6).The MD simulations sampled one conformation predicted for a triple-gated pore, bring-ing the total observed to 4 out of 7 (discounting the leak state). They also suggest that someother conformations are not observable because of coupling between thick and thin gates.For example, opening of the thick gate while the EC gate was open was experimentallyruled out (Simmons et al. 2014) (see also the discussion in Section 5.1.); conversely theMD suggested that EC gate will remain closed when the thick gate is open. These obser-vations rule out two more putative states (EC open/thick open/IC open and EC open/thickopen/IC closed). The remaining state (EC closed/thick close/IC open) seems unlikely basedon the MD, which showed that the IC gate is locked when the thick gate is closed. Basedon this analysis, only 4 out of 8 possible states, namely the three crystallographic confor-mations (OF open, OF occluded, IF open) and the MD-based prediction for IF occluded,should be the only observable conformations for Mhp1.An analysis based on gate states is general but limited in important details and mustbe augmented with additional information about the relative stability of the states. Forexample, under physiological conditions, the inward facing conformation is more prevalentthan the outward facing one although this can be changed with the addition of substrate(Kazmier, Sharma, Islam, et al. 2014; Calabrese et al. 2017). The DEER experimentssuggest that some conformations such as the inward facing occluded might be much shorterlived than other ones (Kazmier, Sharma, Islam, et al. 2014). Such quantitative informationis crucial in order to interpret kinetic reaction diagrams based on the predicted states.
6. Unified transport cycle model
In our enumeration of gate states we have implicitly assumed that these states do not de-pend on the binding of driving ions or substrate. Such an assumption is not warranted. Forinstance, the symporter in Figure 1A must avoid leak cycles that involve a slippage transi-tion between the inward- and outward-facing transporter conformations when only the ionis bound or it would just dissipate the ion gradient. Mechanistically, the absence of thesubstrate when the ion is present must be changing the free energy landscape of the tran-sition (Läuger 1980) in such a way that the ion-only transition faces a much higher barrierthan the fully loaded transporter. In other words, binding of an ion and a substrate unlocksthe transporter and enables the conformational transition to occur as experimentally found
7. The crystals that produced the inward facing open structure with PDB ID 2X79 only formed when cellswere grown on minimal medium with seleno- L -methionine (Shimamura et al. 2010). The electron density mapof 2X79 contains a blob of unidentified density in the inward facing cavity that appears to have wedged openthe IC gate and stabilized the IF conformation. It is possible that the unidentified molecule(s) held the IC gatein an especially wide open position from which it relaxes in the MD. (cid:105) “ms” — 2019/12/16 — 1:47 — page 24 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
24 O. Beckstein and F. Naughton
ABC
Figure 11. Unified picture of transporter function for a hypothetical transporter with a 1:1stoichiometry between driving ion (red circle) and substrate (blue square) and two gates.All combinations of conformational states with bound ion and/or substrate are listed. Leakstates are omitted for simplicity. When an ion or substrate is shown, the corresponding bind-ing or dissociation reaction is implied. Depending on the physiological function (symbolon the right), only certain sequences of states are visited (in the idealized case) while others(grayed out) are not part of the cycle. A Symporter. (The cycle drawn here correspondsto the one in Figure 1A; the alternative binding sequences 1 ↔ ↔ ↔ ↔
12 mightalso occur.) B Antiporter. (The cycle corresponds to the one in Figure 1B; the alternativebinding sequences 2 ↔ ↔ ↔ ↔
11 might also occur.) C Uniporter. (cid:105) “ms” — 2019/12/16 — 1:47 — page 25 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 25for the aspartate-sodium symporter GltPh (Akyuz et al. 2015). A similar argument can bemade for the antiporter cycle in Figure 1B where the transition between the empty inwardand outward-facing states needs to be suppressed to avoid leak cycles that only dissipate theion. In this case, binding of the ion or the substrate unlocks the transporter. In both casesit is clear that presence or absence of bound ion and/or substrate corresponds to a proteinwith different energetics from, say, the unloaded conformation.Consequently, the number of available states of a transporter will be the product ofthe number conformational states n C with the number of ways to bind driving ions andsubstrates. We can estimate n C from the gate model as n C = 2 N − for N gates anda leak state excluded. Assuming an ion:substrate stoichiometry of ν I : ν S and making thesimplifying assumption that there are ν I ion binding sites and ν S substrate binding sites,there are n j = (cid:80) ν j k =0 (cid:0) ν j k (cid:1) = 2 ν j ways to distribute zero to ν j particles over ν j bindingsites. Hence the number of states in the model is Ω( N, ν I , ν S ) = n C n I n S = (2 N −
1) 2 ν I + ν S (11)(If ions and substrates compete for overlapping binding sites then Ω is less than the valuegiven by Eq. 11.)For example, for the transporters in Figure 1, ν I = ν S = 1 and with N = 2 , Ω =3 × different states should be considered, as shown in Figure 11. In principle,all theoretically possible transitions between states (arrows in Figure 11) contribute to anoverall transport process (Hill 1989). In our example, occluded states are included forcompleteness. However, occluded states (numbers 5–8) cannot exchange with each otherand their effect could be replaced with an effective rate constant between the outward facingand inward facing conformations that are connected by the occluded state.If we follow Zuckerman (2019) and consider three different idealized processes, inwhich certain transitions are suppressed by virtue of low kinetic rate constants (Hill andEisenberg 1981) (grayed out in Figure 11), then we recognize the transport cycle for a sym-porter (Figure 11A) and for an antiporter (Figure 11B). Furthermore, by suppressing tran-sitions that involve the driving ion and only retaining a cycle that contains the alternating-access transition with either substrate bound or the empty transporter, a simple uniportermodel emerges (Figure 11C). In this case, no energy coupling occurs and the substrate willmove down its electrochemical gradient by facilitated diffusion.The idealized cycles could be made more realistic by retaining all transitions relatedto ion and substrate binding (1–4 and 9–12 in Figure 11) while still suppressing the un-desirable conformational transition (e.g., the ion leak pathway 2 ↔
11 for the symporter inFigure 11A). With these additional transitions included, different sequences of binding ordissociation reactions would be included in the models for the symporter and antiporter. Auniporter might be able to bind a driving ion but not transport it. Overall, the three physi-ologically very different types of transporters only differ in which of the alternating-accesstransitions is forbidden (or strongly suppressed). One can imagine that mutations may dif-ferentially change the transition rates and so switch the function of transporter from, say,a symporter to a uniporter (as seen in the MFS sugar transporters (Madej et al. 2014)) oran antiporter to a symporter as reviewed by Henderson, Fendler, and Poolman (2019). Oc-cluded states are now seen as a possible control point to tune the function: for instance, (cid:105) “ms” — 2019/12/16 — 1:47 — page 26 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
26 O. Beckstein and F. Naughtonin order to block an outward ↔ inward transition, only one partial transition to or from theoccluded state needs to be blocked.
7. Conclusion
We provided a perspective on broad and general principles that apply to secondary activetransporters. Transporters are seen as catalysts or “physical enzymes” that enable transportacross the cell membrane against an electrochemical gradient by transducing free energyfrom the electrochemical gradient of a driving ion. Energy transduction requires cyclicalreactions that include both driving ion and transported substrate. The alternating accessmodel provides a simple scheme through which such cycles can be established. The proteinmust exist in at least two distinct conformations in which the binding sites are exposed onlyto either the outside or the inside solution. Importantly, no energy transduction is possible ifa continuous pore is established. The two protein conformations that are needed for the twoalternating access states are related by a structural two-fold pseudo symmetry that originatesin inverted repeats in the protein’s genetic sequence. A description of transporters as gatedpores is fruitful in many cases because gates (two state switches) can be identified withstructural elements in the transporter. Enumerating all distinct gate states naturally includesoccluded states in the alternating access picture and also suggests what kind of proteinconformations might be observable. By connecting the possible conformational states andion/substrate bound states in a kinetic model, a unified picture emerges in which symporter,antiporter, and uniporter function are extremes in a continuum of functionality.Many open questions remain. For example, the molecular mechanism of coupling be-tween conformational changes and ion/substrate binding and allosteric interactions betweengates need to be evaluated for most known transporters. General theories of allosteric cou-pling will likely be helpful to define the specific quantitative questions that need to be asked(LeVine et al. 2016). Occluded states were explained as a consequence of the existence ofgates in transporters, so a natural question to ask is if all transporters have occluded states,and if so, are they ultimately a consequence of the symmetries of the inverted repeats? It istempting to speculate that occluded states are the fully symmetrical high energy conforma-tions whose energy is lowered by symmetry breaking. There remain classes of transportersfor which we do not have sufficient structural evidence to answer the question althoughsimulations (as shown here) have started filling this gap. The unified model indicates thattransporter function forms a continuum. However, how difficult is it to move through thiscontinuum, what are the minimal changes to change physiological function? Could suchchanges be achieved with allosteric modulators (small molecules) or changes in externalconditions such as membrane tension, pressure, temperature, or transmembrane voltage?More broadly speaking, it has also been recognized that channels and transporters forma spectrum (Gadsby 2009; Henderson, Fendler, and Poolman 2019), or as expressed byLäuger (1980): “Channel and carrier [transporter] models should therefore not be regardedas mutually exclusive possibilities, but rather as limiting cases of a more general mecha-nism.” There seems to be value in stepping back and asking what the general principles areunder which a class of proteins has to operate. (cid:105) “ms” — 2019/12/16 — 1:47 — page 27 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
General Principles of Secondary Active Transporter Function 27
Acknowledgments
We thank Philip Fowler for providing us with the data for Figure 9.Research reported in this work was supported by the National Institute Of General MedicalSciences of the National Institutes of Health under Awards Number R01GM118772.
References
Abramson, Jeff, Irina Smirnova, Vladimir Kasho, Gillian Verner, H Ronald Kaback, and SoIwata. 2003. “Structure and mechanism of the lactose permease of Escherichia coli.”
Science
301 (5633): 610–615. doi: .Abramson, Jeff, and Ernest M. Wright. 2009. “Structure and function of Na + -symporterswith inverted repeats.” Curr. Opin. Struct. Biol.
19 (4): 425–432. doi: .Accardi, Alessio. 2015. “Structure and gating of CLC channels and exchangers.”
The Jour-nal of Physiology
593 (Pt 18): 4129–4138. doi: .Akyuz, Nurunisa, Elka R. Georgieva, Zhou Zhou, Sebastian Stolzenberg, Michel A. Cuen-det, George Khelashvili, Roger B. Altman, Daniel S. Terry, Jack H. Freed, Harel We-instein, Olga Boudker, and Scott C. Blanchard. 2015. “Transport domain unlockingsets the uptake rate of an aspartate transporter.”
Nature
518 (7537): 68–73. doi: .Bai, Xiaoyun, Trevor F. Moraes, and Reinhart A. F. Reithmeier. 2017. “Structural biologyof solute carrier (SLC) membrane transport proteins.”
Molecular Membrane Biology
34 (1-2): 1–32. doi: .Boudker, Olga, and Grégory Verdon. 2010. “Structural perspectives on secondary activetransporters.”
Trends Pharmacol Sci
31 (9): 418–426. doi:
10 . 1016 / j . tips .2010.06.004 .Buch-Pedersen, M J, B P Pedersen, B Veierskov, P Nissen, and M G Palmgren. 2009.“Protons and how they are transported by proton pumps.”
Pflügers Arch
457 (3): 573–9. doi: .Calabrese, Antonio N., Scott M. Jackson, Lynsey N. Jones, Oliver Beckstein, FlorianHeinkel, Joerg Gsponer, David Sharples, Marta Sans, Maria Kokkinidou, Arwen R.Pearson, Sheena E. Radford, Alison E. Ashcroft, and Peter J. F. Henderson. 2017.“Topological Dissection of the Membrane Transport Protein Mhp1 Derived from Cys-teine Accessibility and Mass Spectrometry.” PMID: 28726379,
Analytical Chemistry
89 (17): 8844–8852. doi: .Cameron, Alexander D., Oliver Beckstein, and Peter J. F. Henderson. 2013. “The five-helixinverted repeat superfamily of membrane transport proteins.” In
Encylopedia of Bio-physics, edited by Gordon Roberts, 1481–1485. Springer. doi: . (cid:105) “ms” — 2019/12/16 — 1:47 — page 28 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
28 O. Beckstein and F. NaughtonChang, Yung-Ning, and Eric R. Geertsma. 2017. “The novel class of seven transmembranesegment inverted repeat carriers.”
Biological Chemistry
398 (2): 165–174. doi:
10 .1515/hsz-2016-0254 .Crisman, Thomas J., Shaogang Qu, Baruch I. Kanner, and Lucy R. Forrest. 2009. “Inward-facing conformation of glutamate transporters as revealed by their inverted-topologystructural repeats.”
Proceedings of the National Academy of Sciences
106 (49): 20752–20757. doi: .DeFelice, Louis J, and Tapasree Goswami. 2007. “Transporters as channels.”
Annu RevPhysiol .Diallinas, George. 2014. “Understanding transporter specificity and the discrete appearanceof channel-like gating domains in transporters.”
Frontiers in Pharmacology .Dill, Ken A., and Sarina Bromberg. 2003.
Molecular Driving Forces.
New York: GarlandScience.Drew, David, and Olga Boudker. 2016. “Shared Molecular Mechanisms of MembraneTransporters.”
Annual Review of Biochemistry
85 (1): 543–572. doi:
10 . 1146 /annurev-biochem-060815-014520 .Fang, Yiling, Hariharan Jayaram, Tania Shane, Ludmila Kolmakova-Partensky, Fang Wu,Carole Williams, Yong Xiong, and Christopher Miller. 2009. “Structure of a prokary-otic virtual proton pump at 3.2 A resolution.”
Nature
460 (7258): 1040–1043. doi: .Forrest, Lucy R. 2013. “Structural biology. (Pseudo-)symmetrical transport.”
Science .. 2015. “Structural Symmetry in Membrane Proteins.”
Annual Review of Biophysics
44 (1): 311–337. doi: .Forrest, Lucy R., Reinhard Krämer, and Christine Ziegler. 2011. “The structural basis ofsecondary active transport mechanisms.”
Biochim Biophys Acta .Forrest, Lucy R., and Gary Rudnick. 2009. “The rocking bundle: a mechanism for ion-coupled solute flux by symmetrical transporters.”
Physiology
10 .1152/physiol.00030.2009 .Forrest, Lucy R., Yuan-Wei Zhang, Miriam T. Jacobs, Joan Gesmonde, Li Xie, Barry H.Honig, and Gary Rudnick. 2008. “Mechanism for alternating access in neurotransmit-ter transporters.”
Proceedings of the National Academy of Sciences
105 (30): 10338–10343. doi: . (cid:105) “ms” — 2019/12/16 — 1:47 — page 29 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 29Fowler, Philip W., Marcella Orwick-Rydmark, Sebastian Radestock, Nicolae Solcan, Patri-cia M. Dijkman, Joseph A. Lyons, Jane Kwok, Martin Caffrey, Anthony Watts, Lucy R.Forrest, and Simon Newstead. 2015. “Gating topology of the proton-coupled oligopep-tide symporters.”
Structure
23 (2): 290–301. doi:
10 . 1016 / j . str . 2014 . 12 .012 .Fuster, Daniel G., and R. Todd Alexander. 2014. “Traditional and emerging roles for theSLC9 Na + /H + exchangers.” Pflügers Archiv - European Journal of Physiology .Gadsby, David C. 2009. “Ion channels versus ion pumps: the principal difference, in prin-ciple.”
Nat Rev Mol Cell Biol
10 (5): 344–52. doi: .Gouaux, Eric. 2009. “The molecular logic of sodium-coupled neurotransmitter trans-porters.”
Philos Trans R Soc Lond B Biol Sci
364 (1514): 149–54. doi:
10 . 1098 /rstb.2008.0181 .Henderson, Ryan K., Klaus Fendler, and Bert Poolman. 2019. “Coupling efficiency of sec-ondary active transporters.”
Current Opinion in Biotechnology
10 .1016/j.copbio.2018.11.005 .Henzler-Wildman, Katherine. 2012. “Analyzing conformational changes in the transport cy-cle of EmrE.” Folding and binding/Protein-nucleic acid interactions,
Current Opinionin Structural Biology
22 (1): 38–43. doi: .Henzler-Wildman, Katherine, and Dorothee Kern. 2007. “Dynamic personalities of pro-teins.”
Nature
450 (7172): 964–72. doi: .Hill, Terrell L. 1989.
Free energy transduction in biology and biochemical cycle kinetics.
New York: Springer.Hill, Terrell L., and Evan Eisenberg. 1981. “Can free energy transduction be localized atsome crucial part of the enzymatic cycle?”
Q Rev Biophys
14 (4): 463–511. doi: .Hille, Bertil. 2001.
Ion Channels of Excitable Membranes.
Nature
478 (7369):408–11. doi: .Huang, Yafei, M. Joanne Lemieux, Jinmei Song, Manfred Auer, and Da-Neng Wang. 2003.“Structure and mechanism of the glycerol-3-phosphate transporter from Escherichiacoli.”
Science
301 (5633): 616–620. doi: .Hunte, Carola, Emanuela Screpanti, Miro Venturi, Abraham Rimon, Etana Padan, and Hart-mut Michel. 2005. “Structure of a Na + /H + antiporter and insights into mechanismof action and regulation by pH.” Nature
435 (7046): 1197–11202. doi:
10 . 1038 /nature03692 . (cid:105) “ms” — 2019/12/16 — 1:47 — page 30 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
30 O. Beckstein and F. NaughtonJackson, Scott M, Ekaterina Ivanova, Katie Simmons, Simon G Patching, Simone Weyand,Tatsuro Shimamura, Florian Brückner, So Iwata, David J Sharples, Stephen A Bald-win, Mark PS Sansom, Oliver Beckstein, Alexander D Cameron, and Peter JF Hen-derson. 2013. “Na + -Hydantoin Membrane Transport Protein, Mhp1.” In Encylopediaof Biophysics, edited by Gordon Roberts, 1514–1521. Springer. doi: .Jardetzky, O. 1966. “Simple allosteric model for membrane pumps.”
Nature
211 (5052):969–970. doi: .Jeschke, Gunnar. 2013. “A comparative study of structures and structural transitions ofsecondary transporters with the LeuT fold.”
Eur Biophys J
42 (2-3): 181–197. doi: .Kaback, H. R., R. Dunten, S. Frillingos, P. Venkatesan, I. Kwaw, W. Zhang, and N. Er-molova. 2007. “Site-directed alkylation and the alternating access model for LacY.”
Proceedings of the National Academy of Sciences
104 (2): 491–494. doi: .Kaback, H. Ronald, Irina Smirnova, Vladimir Kasho, Yiling Nie, and Yonggang Zhou.2011. “The alternating access transport mechanism in LacY.”
J Membr Biol
239 (1-2):85–93. doi: .Kazmier, Kelli, Derek P. Claxton, and Hassane S. Mchaourab. 2017. “Alternating accessmechanisms of LeuT-fold transporters: trailblazing towards the promised energy land-scapes.”
Current Opinion in Structural Biology .Kazmier, Kelli, Shruti Sharma, Shahidul M. Islam, Benoît Roux, and Hassane S.Mchaourab. 2014. “Conformational cycle and ion-coupling mechanism of theNa + /hydantoin transporter Mhp1.” Proceedings of the National Academy of Sciences
111 (41): 14752–14757. doi: .Kazmier, Kelli, Shruti Sharma, Matthias Quick, Shahidul M Islam, Benoît Roux, HarelWeinstein, Jonathan A Javitch, and Hassane S Mchaourab. 2014. “Conformationaldynamics of ligand-dependent alternating access in LeuT.”
Nat Struct Mol Biol
21 (5):472–9. doi: .Keller, Rebecca, Christine Ziegler, and Dirk Schneider. 2014. “When two turn into one:evolution of membrane transporters from half modules.”
Biol Chem
395 (12): 1379–88. doi: .Klingenberg, Martin. 1979. “The ADP,ATP shuttle of the mitochondrion.”
Trends in Bio-chemical Sciences .. 2007. “Transport viewed as a catalytic process.”
Biochimie
89 (9): 1042–1048.doi: . (cid:105) “ms” — 2019/12/16 — 1:47 — page 31 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 31Korkhov, Vladimir M., and Christopher G. Tate. 2009. “An emerging consensus for thestructure of EmrE.”
Acta Crystallographica Section D
65 (2): 186–192. doi: .Krishnamurthy, Harini, and Eric Gouaux. 2012. “X-ray structures of LeuT in substrate-free outward-open and apo inward-open states.”
Nature .Krishnamurthy, Harini, Chayne L. Piscitelli, and Eric Gouaux. 2009. “Unlocking themolecular secrets of sodium-coupled transporters.”
Nature
459 (7245): 347–355.doi: .Latorraca, Naomi R., Nathan M. Fastman, A.J. Venkatakrishnan, Wolf B. Frommer, Ron O.Dror, and Liang Feng. 2017. “Mechanism of Substrate Translocation in an AlternatingAccess Transporter.”
Cell
169 (1): 96–107.e12. doi: .Läuger, P. 1980. “Kinetic properties of ion carriers and channels.”
J Membr Biol
57 (3):163–178. doi: .Law, Christopher J, Peter C Maloney, and Da-Neng Wang. 2008. “Ins and outs of majorfacilitator superfamily antiporters.”
Annu Rev Microbiol .Lespine, A, J Dupuy, M Alvinerie, C Comera, T Nagy, P Krajcsi, and S Orlowski. 2009.“Interaction of macrocyclic lactones with the multidrug transporters: the bases of thepharmacokinetics of lipid-like drugs.”
Curr Drug Metab
10 (3): 272–288.LeVine, Michael V, Michel A Cuendet, George Khelashvili, and Harel Weinstein. 2016.“Allosteric Mechanisms of Molecular Machines at the Membrane: Transport bySodium-Coupled Symporters.”
Chem Rev
116 (11): 6552–87. doi: .Li, Jing, Po-Chao Wen, Mahmoud Moradi, and Emad Tajkhorshid. 2015. “Computationalcharacterization of structural dynamics underlying function in active membrane trans-porters.”
Current Opinion in Structural Biology
31 (0): 96–105. doi: .Madej, M Gregor, Linfeng Sun, Nieng Yan, and H Ronald Kaback. 2014. “Functional ar-chitecture of MFS D-glucose transporters.”
Proc Natl Acad Sci U S A
111 (7): E719–E727. doi: .Marger, Michael D., and Milton H. Saier Jr. 1993. “A major superfamily of transmem-brane facilitators that catalyse uniport, symport and antiport.”
Trends in BiochemicalSciences
18 (1): 13–20. doi: .Miller, C. 2006. “ClC chloride channels viewed through a transporter lens.”
Nature . (cid:105) “ms” — 2019/12/16 — 1:47 — page 32 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
32 O. Beckstein and F. NaughtonMilo, Ron, Paul Jorgensen, Uri Moran, Griffin Weber, and Michael Springer. 2010.“BioNumbers–the database of key numbers in molecular and cell biology.”
NucleicAcids Res
38 (Database issue): D750–3. doi: .Mitchell, Peter. 1967. “Translocations Through Natural Membranes.” Chap. 2 in
Advancesin Enzymology and Related Areas of Molecular Biology, edited by F. F. Nord, 29:33–87. Advances in Enzymology and Related Areas of Molecular Biology. New York:John Wiley & Sons, Inc. doi: .Morrison, Emma A., Gregory T. Dekoster, Supratik Dutta, Reza Vafabakhsh, MichaelW. Clarkson, Arjun Bahl, Dorothee Kern, Taekjip Ha, and Katherine A. Henzler-Wildman. 2012. “Antiparallel EmrE exports drugs by exchanging between asymmetricstructures.”
Nature .Morth, J Preben, Bjørn P Pedersen, Mads S Toustrup-Jensen, Thomas L-M Sørensen, JannePetersen, Jens Peter Andersen, Bente Vilsen, and Poul Nissen. 2007. “Crystal structureof the sodium-potassium pump.”
Nature
450 (7172): 1043–9. doi: .Myers-Turnbull, Douglas, Spencer E. Bliven, Peter W. Rose, Zaid K. Aziz, PhilippeYoukharibache, Philip E. Bourne, and Andreas Prli´c. 2014. “Systematic Detection ofInternal Symmetry in Proteins Using CE-Symm.”
Journal of Molecular Biology .Naftalin, Richard J. 2010. “Reassessment of models of facilitated transport and cotrans-port.”
J Membr Biol
234 (2): 75–112. doi: .Newstead, Simon, David Drew, Alexander D Cameron, Vincent L G Postis, XiaobingXia, Philip W Fowler, Jean C Ingram, Elisabeth P Carpenter, Mark S P Sansom,Michael J McPherson, Stephen A Baldwin, and So Iwata. 2011. “Crystal structure ofa prokaryotic homologue of the mammalian oligopeptide-proton symporters, PepT1and PepT2.”
EMBO J .Ottolia, Michela, Scott John, Yi Xie, Xiaoyan Ren, and Kenneth D Philipson. 2007. “Shed-ding light on the Na+/Ca2+ exchanger.”
Ann N Y Acad Sci
10 .1196/annals.1387.044 .Padan, Etana. 2014. “Functional and structural dynamics of NhaA, a prototype for Na + andH + antiporters, which are responsible for Na + and H + homeostasis in cells.” Biochim-ica et Biophysica Acta (BBA) - Bioenergetics .Perez, Camilo, Belinda Faust, Ahmad Reza Mehdipour, Kevin A. Francesconi, Lucy R.Forrest, and Christine Ziegler. 2014. “Substrate-bound outward-open state of the be-taine transporter BetP provides insights into Na + coupling.” Nat Commun .Perez, Camilo, Caroline Koshy, Ozkan Yildiz, and Christine Ziegler. 2012. “Alternating-access mechanism in conformationally asymmetric trimers of the betaine transporterBetP.”
Nature . (cid:105) “ms” — 2019/12/16 — 1:47 — page 33 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 33Radestock, Sebastian, and Lucy R. Forrest. 2011. “The alternating-access mechanism ofMFS transporters arises from inverted-topology repeats.”
Journal of Molecular Biol-ogy
407 (5): 698–715. doi: .Ressl, Susanne, Anke C Terwisscha van Scheltinga, Clemens Vonrhein, Vera Ott, and Chris-tine Ziegler. 2009. “Molecular basis of transport and regulation in the Na + /betainesymporter BetP.” Nature
458 (7234): 47–52. doi: .Reyes, Nicolas, Christopher Ginter, and Olga Boudker. 2009. “Transport mechanism of abacterial homologue of glutamate transporters.”
Nature
462 (7275): 880–885. doi: .Rolfe, D. F., and G. C. Brown. 1997. “Cellular energy utilization and molecular origin ofstandard metabolic rate in mammals.”
Physiol Rev
77 (3): 731–58.Sazanov, Leonid A. 2015. “A giant molecular proton pump: structure and mechanism ofrespiratory complex I.”
Nat Rev Mol Cell Biol
16 (6): 375–88. doi: .Schushan, Maya, Abraham Rimon, Turkan Haliloglu, Lucy R. Forrest, Etana Padan, andNir Ben-Tal. 2012. “A Model-Structure of a Periplasm-facing State of the NhaAAntiporter Suggests the Molecular Underpinnings of pH-induced ConformationalChanges.”
Journal of Biological Chemistry
287 (22): 18249–18261. doi: .Schwartz, Steven D, and Vern L Schramm. 2009. “Enzymatic transition states and dynamicmotion in barrier crossing.”
Nat Chem Biol .Schweikhard, Eva S., and Christine M. Ziegler. 2012. “Amino acid secondary transporters:toward a common transport mechanism.” Chap. 1 in
Co-Transport Systems, edited byMark O. Bevensee, 70:1–28. Current Topics in Membranes. Elsevier Inc. doi:
10 .1016/B978-0-12-394316-3.00001-6 .Selvam, Balaji, Shriyaa Mittal, and Diwakar Shukla. 2018. “Free Energy Landscape of theComplete Transport Cycle in a Key Bacterial Transporter.”
ACS Cent Sci .Shi, Yigong. 2013. “Common Folds and Transport Mechanisms of Secondary Active Trans-porters.”
Annual Review of Biophysics
42 (1): 51–72. doi:
10 . 1146 / annurev -biophys-083012-130429 .Shimamura, Tatsuro, Simone Weyand, Oliver Beckstein, Nicholas G Rutherford, JonathanM Hadden, David Sharples, Mark S P Sansom, So Iwata, Peter J F Henderson,and Alexander D Cameron. 2010. “Molecular basis of alternating access membranetransport by the sodium-hydantoin transporter Mhp1.”
Science
328 (5977): 470–473.doi: . (cid:105) “ms” — 2019/12/16 — 1:47 — page 34 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105)
34 O. Beckstein and F. NaughtonSimmons, Katie J., Scott M. Jackson, Florian Brückner, Simon G. Patching, Oliver Beck-stein, Ekaterina Ivanova, Tian Geng, Simone Weyand, David Drew, Joseph Lannigan,David J. Sharples, Mark S. P. Sansom, So Iwata, Colin WG. Fishwick, A. Peter John-son, Alexander D. Cameron, and Peter JF Henderson. 2014. “The molecular mecha-nism of ligand recognition by membrane transport protein, Mhp1.”
EMBO J .Slotboom, Dirk J. 2014. “Structural and mechanistic insights into prokaryotic energy-coupling factor transporters.”
Nat Rev Microbiol
12 (2): 79–87. doi:
10 . 1038 /nrmicro3175 .Smart, Oliver S., Joseph G. Neduvelil, Xiaonan Wang, B. A. Wallace, and Mark S. P. San-som. 1996. “HOLE: A program for the analysis of the pore dimensions of ion channelstructural models.”
J. Mol. Graph.
Proceedings of the National Academy of Sciences
115 (50):12716–12721. doi: .Stelzl, Lukas S., Philip W. Fowler, Mark S. P. Sansom, and Oliver Beckstein. 2014. “Flex-ible gates generate occluded intermediates in the transport cycle of LacY.”
J Mol Biol .Suzuki, Shun’ichi, and Peter J. F. Henderson. 2006. “The hydantoin transport protein from
Microbacterium liquefaciens .” J Bacteriol
188 (9): 3329–3336. doi: .Tanford, C. 1983. “Translocation pathway in the catalysis of active transport.”
Proc NatlAcad Sci U S A
80 (12): 3701–3705.Vandenberg, Robert J., Shiwei Huang, and Renae M. Ryan. 2008. “Slips, leaks and channelsin glutamate transporters.”
Channels
Proteins
82 (10):2797–811. doi: .Vergara-Jaque, Ariela, Cristina Fenollar-Ferrer, Desirée Kaufmann, and Lucy R. Forrest.2015. “Repeat-swap homology modeling of secondary active transporters: updatedprotocol and prediction of elevator-type mechanisms.”
Frontiers in Pharmacology .Weyand, Simone, Tatsuro Shimamura, Oliver Beckstein, Mark S. P. Sansom, So Iwata, PeterJ. F. Henderson, and Alexander D. Cameron. 2011. “The alternating access mechanismof transport as observed in the sodium-hydantoin transporter Mhp1.”
J. SynchrotronRad.
18 (1): 20–23. doi: . (cid:105) “ms” — 2019/12/16 — 1:47 — page 35 — (cid:105)(cid:105)(cid:105) (cid:105) (cid:105)(cid:105) General Principles of Secondary Active Transporter Function 35Weyand, Simone, Tatsuro Shimamura, Shunsuke Yajima, Shun’ichi Suzuki, Osman Mirza,Kuakarun Krusong, Elisabeth P Carpenter, Nicholas G Rutherford, Jonathan M Had-den, John O’Reilly, Pikyee Ma, Massoud Saidijam, Simon G Patching, Ryan JHope, Halina T Norbertczak, Peter C J Roach, So Iwata, Peter J F Henderson, andAlexander D Cameron. 2008. “Structure and molecular mechanism of a nucleobase-cation-symport-1 family transporter.”
Science
322 (5902): 709–713. doi: .Yamashita, Atsuko, Satinder K Singh, Toshimitsu Kawate, Yan Jin, and Eric Gouaux. 2005.“Crystal structure of a bacterial homologue of Na + /Cl − -dependent neurotransmittertransporters.” Nature (Department of Biochemistry)(Molecular Biophysics and.) 437(7056): 215–223. doi: .Yan, Nieng. 2013. “Structural advances for the major facilitator superfamily (MFS) trans-porters.”
Trends Biochem Sci
38 (3): 151–9. doi: .Zhou, Yunpeng, and John H. Bushweller. 2018. “Solution Structure and Elevator Mecha-nism of the Membrane Electron Transporter CcdA.”
Nature structural & molecularbiology
25 (2): 163–169. doi: .Zuckerman, Daniel M. 2019.
Physical Lens on the Cell. online.