Substrate selectivity in arginine-dependent acid resistance in enteric bacteria

Abstract

To successfully colonize the human gut, enteric bacteria must activate acid resistance systems to survive the extreme acidity (pH 1.5–3.5) of the stomach. The antiporter AdiC is the master orchestrator of the arginine-dependent system. Upon acid shock, it imports extracellular arginine (Arg) into the cytoplasm, providing the substrate for arginine decarboxylases, which consume a cellular proton ending up in a C–H bond of the decarboxylated product agmatine (Agm²⁺). Agm²⁺ and the “virtual” proton it carries are exported via AdiC subsequently. It is widely accepted that AdiC counters intracellular acidification by continuously pumping out virtual protons. However, in the gastric environment, Arg is present in two carboxyl-protonation forms, Arg⁺ and Arg²⁺. Virtual proton pumping can only be achieved by Arg⁺/Agm²⁺ exchange, whereas Arg²⁺/Agm²⁺ exchange would produce no net proton movement. This study experimentally asks which exchange AdiC catalyzes, an issue previously unapproachable due to the absence of a reconstituted system mimicking the situation of bacteria in the stomach. Here, using an oriented liposome system able to hold a three-unit pH gradient, we demonstrate that Arg/Agm exchange by AdiC is strongly electrogenic with positive charge moved outward, and thus that AdiC mainly mediates Arg⁺/Agm²⁺ exchange to support effective virtual proton pumping. Further experiments reveal a mechanistic surprise—that AdiC selects Arg⁺ against Arg²⁺ on the basis of gross valence, rather than by local scrutiny of protonation states of the carboxyl group, as had been suggested by Arg-bound AdiC crystal structures.

The extreme acidity of the human stomach (pH 1.5–3.5) presents a lethal environment for the myriad microorganisms taken daily into the digestive tract. Enteric bacteria such as Escherichia, Salmonella, Shigella, and many others breach this host defense by activating specific amino-acid–dependent “extreme acid resistance” systems. Extensive studies have identified key players in two of these systems—the antiporters AdiC and GadC (1), which deliver extracellular arginine and glutamate, respectively, to their corresponding cytosolic decarboxylases and expel the decarboxylated products, agmatine (Agm) and γ-aminobutyric acid, in a strictly coupled one-to-one exchange (2–6). In this and the companion article (7), we address the question: How do these membrane transport activities produce bacterial acid resistance?

The arginine-dependent system, depicted in Fig. 1, concerns us here. Upon encountering low extracellular pH, AdiC becomes activated (5, 8) and imports extracellular arginine (Arg). Once in the cytoplasm, Arg encounters an acid-activated decarboxylase, AdiA, which is dormant at neutral pH but becomes active as the cytoplasm drops below pH 6 (9, 10). The decarboxylation reaction consumes a proton, forming a C–H bond in Agm, which is then transported out of the cell by AdiC along with the “virtual proton.” The antiporter is thus seen to act as a decarboxylation-driven proton efflux pump, continually counteracting intracellular acidification during strong acid challenge to prevent the cytoplasm from falling much below pH 5 (11).

Fig. 1.

AdiC in the gastric environment. AdiC switches between outward- and inward-open conformations to import extracellular Arg while expelling intracellular Agm, which carries the virtual proton in a C–H bond (black circle). At extracellular pH 2.2, 50% Arg is protonated on the α-carboxyl (shadow).

This widely accepted mechanism, however, is potentially problematic. At a typical gastric pH of 2.2, near the α-carboxylate pK_a, ∼50% of Arg is present in carboxyl-protonated form (Arg²⁺), which if imported would deliver a proton into the less acidic cytoplasm to create a proton-neutral antiport cycle (Fig. 1); accordingly, it was proposed (5, 12) that AdiC specifically selects deprotonated Arg⁺ for transport. This idea, however, is based on the teleologically appealing—but shaky—assumption of optimal biological efficiency and has not been directly tested at the level of the protein's transport behavior. It therefore remains unclear how AdiC influences intracellular proton dynamics of bacteria suffering from acid shock. In this study, we address the issue of which Arg form is transported by determining whether Arg/Agm is electrogenic under conditions resembling acid shock, i.e., if it is accompanied by net movement of electric charge in strongly acidic conditions. Electrogenic Arg⁺/Agm²⁺ exchange produces outward positive current and so should be suppressible by a large negative-inside membrane potential, whereas the same potential would exert no effect on electroneutral Arg²⁺/Agm²⁺ exchange. Applying this strategy, we demonstrate that Arg/Agm antiport is strongly electrogenic and thus that Arg⁺ is the preferred substrate for AdiC.

Many antiporters operate via the classic “alternating-access” mechanism (13), whereby the protein switches between outward-open and inward-open conformations to move cargo across the membrane (Fig. 1). We show that the outward-open conformation prefers Arg⁺ over Arg²⁺, whereas the inward-open conformation discriminates poorly if at all between these two forms. A mechanistic surprise also emerges from our experiments: that AdiC crystal structures (12, 14, 15), which strongly suggest specific hydrogen bonding as essential for substrate selection, are frankly misleading; we find instead that gross electrical valence rather than intimate chemical interaction plays the major role in AdiC's specificity toward Arg⁺.

Results

Two difficulties in creating an AdiC-reconstituted system echoing conditions of bacterial acid stress previously prevented us from directly tackling substrate selectivity (12). First, the liposomes used for substrate-transport assays must hold a large pH gradient on the experimental timescale near pH 2, but to our knowledge, proton leakage across liposome membranes has not been previously examined under extreme acid conditions. Accordingly, we used a pH-sensitive dye to follow intraliposomal acidification in the presence of a large pH gradient (outside pH 2.2/inside pH 5.0). The liposomes can indeed maintain this gradient for several minutes, time enough for the transport experiments contemplated here, but only when appropriate lipids, salts, buffers, and protein densities are used, as discussed in SI Methods, Fig. S1. Second, all functional antiporters must be in the “outside-out” orientation, with the extracellular side of the protein exposed to the external side of the liposome. AdiC inserts randomly into liposomes, with roughly half the protein in the desired outside-out orientation, and half inside-out. By using a cysteine-substituted AdiC construct and thiol-modifying maneuvers, either population may be completely silenced to yield functionally oriented outside-out or inside-out reconstituted proteoliposomes (8).

This oriented system was tested in the presence of a pH gradient that bacteria typically experience in the human stomach (outside pH 2.2/inside pH 5.0). Experiments (Fig. 2A) comparing Arg_ex/Agm_in antiport in three sets of liposomes—randomly oriented, outside-out, and inside-out—show that transport characteristics in the outside-out and randomly oriented systems are similar, whereas the inside-out liposomes are almost completely (>95%) inhibited. Thus, the inside-out protein contributes negligibly (<5%) to overall substrate transport, which can be explained by the strong inhibition suffered by AdiC when exposed to extreme acidity on its cytoplasmic face (Fig. 2B). This is a fortunate result indicating that the asymmetric pH setup intended to mimic acid-shock conditions also enforces an outside-out oriented system automatically, neatly obviating the need of using cysteine-substituted AdiC protein for the thiol-modification strategy. Therefore, except where specifically indicated, we will simply apply pH gradients to establish outside-out orientation of wild-type (WT) AdiC in the following experiments.

Fig. 2.

Sidedness of AdiC established by asymmetric pH. (A) ¹⁴C-Arg_ex/Agm_in exchange using AdiC-S26C reconstituted in randomly oriented, outside-out, or inside-out liposomes (indicated by icons), with pH gradient conditions shown. (B) Effect of cytoplasmic pH on AdiC activity. Liposomes with pH 5 inside were oriented inside out so that extraliposomal pH is experienced by the cytoplasmic side of AdiC. Uptake of ¹⁴C-Arg_ex in exchange of Agm_in at the indicated pH was recorded 4 min after initiation of transport.

With this system in hand, we evaluate AdiC activity under acid-shock–mimetic conditions with an antiport flux assay. Radiolabeled substrate is added to the outside of liposomes preloaded with unlabeled substrates. Accumulation of counts in the liposomes coupled to the exit of cold substrate is recorded over several minutes, as in Fig. 3A, which follows AdiC-mediated exchange between external Arg (50% , 50% , pH 2.2) and internal Agm (100% , pH 5.0). These experiments also include a 1,000-fold outward K⁺ gradient so that a large hyperpolarized (negative-inside) membrane potential can be imposed by adding the K⁺ ionophore valinomycin (Vln). This maneuver reduces transport activity appoximately fivefold (Fig. 3A), indicating that at least 80% of the exchange observed without hyperpolarization is associated with net efflux of positive charge , whereas at most, 20% is electroneutral . Because and are equally populated at the prevailing pH, but the former is at least fourfold more likely to be imported, we conclude that the extracellular-open conformation of AdiC chooses Arg⁺ over fully protonated Arg²⁺ with at least fourfold selectivity.

Fig. 3.

Substrate selectivity of AdiC. Liposomes reconstituted with WT AdiC (inside pH 5.0, outside pH 2.2, and a 1,000-fold outward K⁺ gradient) were used to quantify (A) ¹⁴C-Arg_ex/Agm_in, (B) ¹⁴C-Agm_ex/Agm_in, and (C) ¹⁴C-Agm_ex/Arg_in exchange, in the presence (open squares) or absence (filled squares) of Vln.

Three potential artifacts must be dismissed to firm up this conclusion. First, nonspecific effects of Vln unrelated to the large membrane potential itself are ruled out because the ionophore does not affect Arg_ex/Agm_in exchange in symmetrical K⁺, which clamps liposomes at zero voltage (Fig. S2). Second, the possibility that hyperpolarization drives AdiC into nonfunctional conformations is ruled out because electroneutral exchange is little affected by this maneuver (Fig. 3B), whereas exchange, which generates inward current, is strongly stimulated by it (Fig. 3C). Third, any contribution from the unwanted inside-out orientation is negligible, because essentially identical results as in Fig. 3A are obtained (Fig. 4A) with inside-out, cysteine-substituted protein rendered explicitly nonfunctional by thiol modification (8). This comparison further confirms that the pH-gradient conditions here lead to WT transporters functionally oriented outside-out.

Fig. 4.

Conformation-specific substrate selectivity. Voltage-dependent ¹⁴C-Arg_ex/Agm_in exchange was assessed using cysteine-substituted AdiC oriented outside out (A) or inside out (B). pH and K⁺ gradients were imposed as in previous experiments, with Vln (open squares) used to create a negative-inside potential.

A question naturally arises as to whether the inward-open conformation of AdiC also selects Arg⁺ against Arg²⁺. This question may be approached by using reconstituted cysteine-modified AdiC in the fully inside-out orientation. In these experiments, the inward-facing conformation opens to the liposome outside (pH 2.2) containing equally populated Arg⁺ and Arg²⁺. Because these inside-out transporters are so strongly inhibited by the acidic extraliposomal solution (Fig. 2), quantifying transport is technically challenging because of the low rates involved. We nonetheless managed to observe that strong hyperpolarization inhibits transport ∼40% (Fig. 4B), indicating that at zero voltage, ∼40% of substrate exchange is electrogenic (Arg⁺/Agm²⁺) and ∼60% is electroneutral (Arg²⁺/Agm²⁺). Thus, the inward-open conformation of AdiC lacks significant selectivity between Arg⁺ and Arg²⁺.

The results above provide a purely phenomenological description of AdiC's substrate selectivity characteristics. By what mechanism does the antiporter's outward-open conformation achieve this selectivity? High-resolution structures of AdiC in Arg-bound states (14, 15) reveal close proximity between the substrate's α-carboxylate and the hydroxyl of Ser26 (Fig. S3). It is thus natural to infer a localized substrate-recognition mechanism, wherein Arg⁺ and Arg²⁺ are differentiated via the stronger H bonding of deprotonated carboxylates (16). To test this idea, we assessed the selectivity of Arg_ex/Agm_in exchange in a mutant, S26A, lacking the presumably critical H bond. In contrast to structural expectation, we find that this mutant, which is fully transport competent, responds to hyperpolarization quantitatively as in the WT protein (Fig. S3). This is a striking result, demonstrating that AdiC's ability to distinguish Arg⁺ from Arg⁺² is not compromised by the H bond's absence.

To follow up this unexpected result on the role of α-carboxyl protonation, we tested external citrulline (50% , 50% , pH 2.2), a transported Arg analog with an uncharged, isosteric side chain (Fig. 5), in exchange for internal Arg (>99% , pH 5.0). If it is α-carboxyl protonation that makes Arg²⁺ a poor substrate, the similarly protonated Cit⁺ should be poorly transported compared with deprotonated Cit⁰. If, in contrast, Arg⁺ is selected mainly on the basis of its net charge, Cit⁺ should be preferred for transport. The result is unambiguous: uptake of citrulline is barely affected by negative potential (Fig. 5A), a sign that exchange is electroneutral. Thus, protonated Cit⁺ is actually preferred over deprotonated Cit⁰, a result starkly refuting the hypothesis that AdiC excludes Arg²⁺ by chemically sniffing out the protonation state of its carboxyl group.

Fig. 5.

Selectivity of Arg analogs. (A) Citrulline transport in gastric pH. A total of 500 μM citrulline was added to the extraliposomal solution (pH 2.2) to expel ¹⁴C-Arg previously loaded in liposomes (pH 5.0), in the presence (open squares) or absence (filled squares) of Vln. The sidedness of WT AdiC was established by the pH gradient. (B–E) pH-dependent transport of Arg analogs. Transport of Arg analogs was examined using the expulsion assay at indicated extraliposomal pH and inside pH 5.0. Trapped ¹⁴C-Arg was expelled by 1 mM of the indicated analogs, except for Agm, which was used at 0.1 mM. Because the inside-out protein is not adequately suppressed by extraliposomal pH higher than 3.0, the thiol-modification strategy is used to enforce the outside-out orientation.

An additional finding buttresses the idea that Arg⁺ and Arg²⁺ are distinguished by overall charge. The outside-out, cysteine-substituted AdiC system was used to examine import of substrate analogs devoid of carboxyl groups and nontitratable over the pH range 2.2–5. As the protein's extracellular side is acidified, transport of monovalent substrates (5-aminopentanol and N-carbamoylputrescine, Fig. 5 B and C) increases, whereas that of divalent substrates (Agm and argininamide, Fig. 5 D and E) declines. Thus, the protein appears to possess an intrinsic acid-activated mechanism that rejects extracellular divalent substrates, such that at the extreme acidity of the stomach, monovalent substrates in general—and Arg⁺ in particular—are preferred for import.

Discussion

The first bacterial antiporter shown to exchange the substrate and product of an intracellular decarboxylase was the oxalate/formate antiporter OxlT (17). As the decarboxylation reaction consumes one intracellular proton to form a new C–H bond in the product, each transport cycle of OxlT was proposed to drive the removal of a single intracellular proton, and the antiporter was dubbed a “virtual” proton pump used to energize the bacterial membrane under anaerobic conditions (18). An analogous mechanism is widely accepted as the basis for Arg-dependent extreme acid resistance in enteric bacteria, which relies upon AdiC operating as a virtual proton pump at extracellular pH as low as 1.5 (1, 3, 4). This physiological picture requires that AdiC must select against Arg²⁺, which would otherwise smuggle protons into the less acidic cytosol and cancel out the proton-removal effect of Arg decarboxylation. If both forms of Arg were transported equally well, AdiC would pump out only 0.5 H⁺ per cycle at gastric pH 2.2 and 0.15 H⁺ at the strongest gastric challenge of pH 1.5.

By using oriented, reconstituted proteoliposomes maintaining pH gradients that mimic extreme acid resistance conditions, we demonstrate directly that AdiC mediates efficient proton extrusion by virtue of the selectivity of its outward-open conformation for deprotonated Arg⁺ as the exchange partner for intracellular Agm²⁺. This selectivity results in at least 0.8 H⁺ being removed from the cytosol in each exchange cycle at pH 2.2. The ∼4-fold selectivity reported here differs from the 30-fold selectivity estimated in our previous work (8, 12), where transport of argininamide, used as a proxy of protonated Arg²⁺, was ∼30-fold less robust than Arg⁺. However, that interpretation was based on an assumption, possibly unjustified, that argininamide is a perfect mimic of Arg²⁺. We also note that previous work demonstrated electrogenicity of Arg/Agm exchange, but because those experiments were carried out at pH 6, where Arg²⁺ is essentially absent, they did not address the issue of Arg⁺/Arg²⁺ selectivity.

In contrast to the outward-open conformation, which must choose between protonation forms of Arg, in the physiological context the inward-open conformation sees negligible Arg, and needs only to export Agm²⁺, the sole form of this substrate. It is therefore expected from evolutionary considerations that this conformation might not differentiate substrate protonation states, as is indeed implied by the lack of selectivity between Arg⁺ and Arg²⁺ shown here. We hasten to note, however, that Arg⁺/Arg²⁺ selectivity for the inward-open conformation is necessarily assessed at a cytoplasmic-side pH of 2.2, which is far more acidic than ever experienced by the bacteria cytoplasm. For this reason, we cannot rule out that this deviation from physiological pH may undermine our conclusion about the inward-facing conformation’s nonselective character.

Current ideas for substrate selection have been shaped by crystal structures of Arg-occupied AdiC, which offer a compelling mechanism for a specific H bond donated by the Ser26 hydroxyl group to preferentially stabilize the deprotonated α-carboxyl of Arg⁺ in its binding site (14, 15). However, we show here that these structures are mechanistically misleading, because elimination of the H-bond donor by mutation disturbs neither substrate transport rate nor selectivity against Arg²⁺. This surprising fact motivated experiments (Fig. 5) with transported Arg analogs that are readily explained by an electrostatic mechanism in which AdiC recognizes the overall substrate charge to differentiate protonation states of the α-carboxyl group, and not the carboxylate itself. We consider the strong preference for transport of protonated over deprotonated citrulline an especially powerful refutation of the H-bonding hypothesis. Of course, chemical features other than net charge, such as size, shape, and polarity, will also influence substrate binding affinity and transport rate, as seen in the range of absolute transport rates seen for the various Arg analogs. However, the central point remains that the biologically crucial selectivity of AdiC toward Arg⁺ is governed not by H bonding, but by charge. A cursory inspection of an Arg-occupied crystal structure of AdiC (15) suggests where the source charges governing this electrostatic selectivity may be located. In this conformation, bound substrate lies at the bottom of an outward-open “cavern” ringed with five positively charged sidechains and no negative charges (because all externally exposed carboxyls would be protonated during acid shock). This ring of charges at the cavern’s rim could produce a local potential of 30–50 mV at the substrate site 20–25 Ǻ below; a potential of this magnitude would preferentially destabilize divalent substrates over monovalent, naturally accounting for the observed ∼5-fold selectivity. This idea is susceptible to future experimental and computational examination.

A charge-based mechanism of substrate selectivity makes sense when considered in light of the extreme acid resistance response. Beyond excluding Arg²⁺ to avoid proton-neutral Arg²⁺/Agm²⁺ antiport, AdiC's outward-open conformation must also efficiently unload Agm²⁺ into the extracellular solution. Physiology thus demands that this conformation renders its substrate site inhospitable to both Arg²⁺ and Agm²⁺. A mechanism that recognizes the +2 valence shared by both molecules, rather than rejecting the protonated α-carboxyl, which Agm²⁺ lacks, is a simple and elegant way for the protein to fulfill this dual purpose.

Methods

Biochemical Procedures.

¹⁴C-arginine was acquired from Perkin-Elmer, and ¹⁴C-agmatine was synthesized by enzymatic decarboxylation, as described (8). N-carbamoylputrescine in Cl⁻ form was synthesized and purified according to established procedures (19), recrystallized twice from water, and confirmed by NMR. Detailed AdiC expression and purification procedures were documented in our previous report (8). For reconstitution, a mixture of POPE [1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine) and POPG (1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1'-rac-glycerol)] (3:1, wt/wt) in chloroform (Avanti Polar Lipids) was blown dry under N₂ and solubilized in 35 mM CHAPS in reconstitution buffer (RB, 100 mM K₂SO₄, 50 mM citric acid, KOH pH 5.0) at 20 mg lipid/mL. Purified protein in 10–20 mM decyl maltoside (DM) was then added to the micellar solution to a final density of 10 μg protein/mg lipid. The mixture was dialyzed against 1,000 vol RB, which was replaced twice every 16–24 h. Liposomes were collected after a 3-d dialysis and stored at −80 °C.

Orientation of AdiC in Liposomes.

Liposomes reconstituted with AdiC transporters fully oriented in either outside-out or inside-out directions were produced as described in detail (8). Briefly, we used an AdiC construct containing a single cysteine (S26C) susceptible to complete inhibition by treatment with thiol-directed methanethiosulfonate reagents from the extracellular side of the protein. AdiC is incorporated randomly into liposomes formed by dialysis, with roughly half of the proteins exposing the cytoplasmic side to the outside solution (inside-out orientation), and half in the alternative, outside-out orientation. To produce the fully inside-out system, proteolipsomes were treated with membrane-impermeant 2-sulfonatoethyl methanethiosulfonate to silence all of the outside-out proteins. For the outside-out system, all transporters were treated with membrane-permeant (2-(trimethylammonium)ethyl) methanethiosulfonate, and then the outside-out proteins were resurrected by treatment with tris (2-carboxyethyl) phosphine, a membrane-impermeant reductant.

Transport Assays.

Antiport activity of AdiC was quantified by monitoring the uptake of ¹⁴C-labeled substrates into liposomes or, in a few “expulsion” experiments, the efflux of preloaded ¹⁴C-Arg (12). In a typical uptake experiment, liposomes were loaded with 5 mM unlabeled Arg or Agm in reconstitution buffer by three freeze–thaw cycles, sonicated briefly, and then spun through Sephadex G-50 columns equilibrated with 100 mM Na₂SO₄, 1 mM citric acid, pH 5.0. The flow through was added to two volumes of flux buffer (100 mM Na₂SO₄, 0.1 mM K₂SO₄ adjusted to pH 2.2–3 by 25 mM glycine-H₂SO₄ or pH 4–5 by 25 mM citric acid NaOH) in the presence or absence of 1 μg/mL Vln immediately before the beginning of the time course. Transport was initiated by adding 50 μM ¹⁴C-labeled Arg or Agm (∼1 µCi/mL) to the extraliposomal solution and terminated by filtration of the sample through G-50 columns at desired time points. For expulsion experiments, ¹⁴C-labeled Arg was first concentrated into liposomes for several minutes using the uptake procedure above, but with extraliposomal pH 5.0. Upon completion of uptake, ¹⁴C-Arg was removed by G-50 spin columns. Flux buffer (two volumes) was then added to adjust the solution to the desired pH, and 0.1–1 mM unlabeled substrate was added to expel trapped ¹⁴C-Arg. In all figures, each point represents the mean ± SE of three to five independent experiments.