Transmembrane Transport of Bicarbonate Unravelled**

Graphical Abstract

A new assay to study bicarbonate transport is reported, which is based on a europium complex that is encapsulated inside liposomes and reports changes in the intravesicular concentration of bicarbonate in a direct and highly sensitive way. This assay was used to reveal that synthetic anion transporters can act via different mechanisms: 1) actual transport of HCO₃⁻ anions; 2) CO₂ diffusion coupled to dissipation of the resulting transmembrane pH gradient.

Abstract

Anion receptors can be used to transport ions across lipid bilayers, which has potential for therapeutic applications. Synthetic bicarbonate transporters are of particular interest, as defects in transmembrane transport of bicarbonate are associated with various diseases. However, no convenient method exists to directly observe bicarbonate transport and study the mechanisms involved. Here, an assay is presented that allows the kinetics of bicarbonate transport into liposomes to be monitored directly and with great sensitivity. The assay utilises an encapsulated europium(III) complex, which exhibits a large increase in emission intensity upon binding bicarbonate. Mechanisms involving CO₂ diffusion and the dissipation of a pH gradient are shown to be able to lead to an increase in bicarbonate concentration within liposomes, without transport of the anion occurring at all. By distinguishing these alternative mechanisms from actual bicarbonate transport, this assay will inform the future development of bicarbonate transporters.

Introduction

The transport of bicarbonate is crucial to many biological processes, such as the regulation of pH^{1, 2} and the removal of metabolic waste.³ The development of synthetic HCO₃⁻ transporters could contribute to the study and understanding of various diseases linked to mutations in HCO₃⁻ transporting proteins, such as haemolytic anaemia, renal diseases, congenital chloride diarrhoea, and glaucoma.³ Furthermore, HCO₃⁻ transporters have potential therapeutic applications and were reported to restore the properties of airway surface liquid in cystic fibrosis airway epithelial tissue.^{4, 5}

Despite the importance of bicarbonate transport in health and disease, most research on synthetic anion transporters to date has focussed on chloride transport.^6-10 This is mainly driven by the ease by which Cl⁻ transport can be studied compared to that of HCO₃⁻. Whereas Cl⁻ transport through the membranes of liposomes can be readily monitored by fluorescent probes or by ion selective electrodes (ISE),⁸ no equivalent methods for the study of HCO₃⁻ transport exist. The pH sensitive probe HPTS has been widely used to study transport of many different anions and cations;¹¹ however, this method cannot be readily adapted for the study of HCO₃⁻ transport, due to the inherent pH variations in HCO₃⁻ solutions over time.

Monitoring HCO₃⁻ transport across lipid membranes remains a significant challenge. In the homeostasis of biological systems, the transmembrane transport of HCO₃⁻ anions and the spontaneous diffusion of CO₂ through membranes are two closely associated processes, which have clearly distinct roles.³ In model systems such as unilamellar vesicles (LUVs), it is not possible to distinguish between the actual transport of HCO₃⁻ and mechanisms based on CO₂ diffusion using current assays. Consequently, the exact mechanism(s) by which synthetic HCO₃⁻ transporters operate remains ambiguous. Nonetheless, numerous reports on HCO₃⁻ transporting anionophores exist,^12-21 of which the first described a series of isophthalamides and the natural compound prodigiosin.¹² However, indirect methods were employed to study the kinetics of HCO₃⁻ transport by these and other compounds (Figure 1). In most cases, either the efflux of Cl⁻ out of liposomes was monitored with a chloride selective electrode,^12-17 or the influx of Cl⁻ was followed with the fluorescent probe lucigenin^18-20 or SPQ.²¹ From the observed Cl⁻ transport it was concluded that an antiport process with HCO₃⁻ must have taken place. Consequently, these methods are restricted to the study of Cl⁻/HCO₃⁻ antiport only, and do not permit the study of exchange with any other anions, nor HCO₃⁻ uniport. This limits the possibilities of studying and understanding HCO₃⁻ transport.

Figure 1

Existing and new methods to study HCO₃⁻ transport.

The only existing method for the direct study of HCO₃⁻ transport relies upon ¹³C NMR spectroscopy in combination with NaH¹³CO₃ and a paramagnetic species, to distinguish interior from exterior isotopically labelled bicarbonate.^{12-14, 21, 5} The major disadvantage of this method is the difficulty in monitoring transport processes over time (requiring 3–5 minutes per NMR spectrum), which precludes the accurate measurement of transport kinetics.²² More recently, an osmotic assay was reported where the efflux of HCO₃⁻ by an anionophore is accompanied by the efflux of a cation (by a cationophore), resulting in an osmotic efflux of water, which can be observed as a change in the scattering intensity of the liposome dispersion.²³ This is a promising strategy for studying HCO₃⁻ uniport; however, the assay suffers from a low sensitivity and requires relatively large concentrations of transporter to be present in the membranes (∼10 mol%).

The limitations associated with current methods clearly call for a new assay that can report on HCO₃⁻ transport directly, accurately and with high sensitivity. Fluorescence-based transport assays offer a great sensitivity and can be readily implemented.^24-26 A fluorescence-based assay in which the influx of HCO₃⁻ can be monitored directly would thus surmount the current limitations, enabling an accurate comparison and quantification of rates of HCO₃⁻ transport, and verification of the results obtained by indirect methods. Crucially, it would enable the mechanisms of transport to be elucidated unequivocally, including i) exchange processes of HCO₃⁻ with different anions (antiport), ii) uniport of HCO₃⁻, and iii) identification of actual transport of HCO₃⁻ versus mechanisms based on CO₂ diffusion. Such an assay requires a water-soluble probe whose emission intensity changes in response to HCO₃⁻ levels, whilst being unaffected by the presence of other anions and cations in the assay.

The cationic europium complex [Eu.L¹]⁺ previously developed by Butler for the purpose of detecting fluoride ions,²⁷ satisfies these requirements. [Eu.L¹]⁺ binds reversibly to HCO₃⁻ in aqueous solution and shows an increase in Eu(III) emission intensity upon binding, particularly within the emission band centred at 615 nm. In contrast, [Eu.L¹]⁺ has negligible responses to Cl⁻ and NO₃⁻ and this made it an ideal candidate for the development of the transport assay. We present here the use of this emissive probe encapsulated in liposomes, to directly monitor the transport of HCO₃⁻ across the lipid bilayers by fluorescence spectroscopy. We have used this new assay to study HCO₃⁻ transport by a series of highly potent synthetic anionophores (1–3, Figure 2) and natural product prodigiosin (4), for which transport was previous observed indirectly using the lucigenin assay (Figure S1).^{18, 20, 28}

Figure 2

Structures of anionophores 1–4.

This novel HCO₃⁻ assay allows the study of the kinetics and mechanisms of HCO₃⁻ transport by ionophores in unprecedented detail, as well as the comparison of various antiport and uniport processes. We have established that transporters 1–4 operate in different ways, and that only 1 is a “pure” HCO₃⁻ carrier, transporting the anion without interference from other processes. Our results raise the distinct possibility that reported HCO₃⁻ transporters might not transport the HCO₃⁻ anion, but rather dissipate the pH gradient induced by CO₂ diffusion. The assay represents a significant step forwards for identifying pure HCO₃⁻ transporters and providing the mechanistic insight required to develop their potential biological applications.

Results and Discussion

Assay to monitor transport of bicarbonate directly

The cationic Eu(III) complex [Eu.L¹]⁺ (Figure 3c) is based on a cyclen scaffold possessing two pendant quinoline arms that absorb UV light around 330 nm and transfer energy efficiently to the central Eu(III) ion, which emits red light in the range 570–720 nm.²⁷ The Eu(III) probe has an open coordination site, occupied by a single water molecule in aqueous solution which quenches the Eu(III) emission significantly. In the presence of HCO₃⁻, the coordinated water molecule is displaced upon binding of the hard oxyanion, resulting in a large enhancement in emission intensity (especially around 615 nm) and changes in spectral form (Figure 3a).²⁹ The probe responds to physiologically relevant (millimolar) concentrations of HCO₃⁻ and exhibits high selectivity over poorly coordinating anions that are commonly used in anion transport assays, including Cl⁻ and NO₃⁻.²⁷ The complex is also sensitive to hydroxide ions, and thus to pH, but this can be controlled readily with the use of a buffer (see ESI).

Figure 3

Transport of HCO₃⁻ by anionophore 1 preincorporated in LUVs with the probe [Eu.L¹]⁺ encapsulated (50 μM), suspended in 225 mM NaCl with 5 mM HEPES at pH 7.0 (interior and exterior), upon addition of 10 mM NaHCO₃ after 30 seconds and lysis of the LUVs after 10 minutes: a. Emission spectra of [Eu.L¹]⁺ recorded during the transport by 1 (at 1 : 25 k transporter to lipid ratio); b. Emission intensity at 615 nm monitored over time for the transport as in a.; c. Schematic representation of EuL1 assay to study transport of HCO₃⁻; d. Normalised transport curves for anionophore 1 preincorporated at various anionophore to lipid ratios.

In order to use [Eu.L¹]⁺ to monitor the transport of HCO₃⁻, we encapsulated this probe into large unilamellar vesicles (LUVs) consisting of the lipids POPC and cholesterol in a 7 : 3 ratio and extruded these liposomes through a membrane with 200 nm pores, to obtain LUVs with an average hydrodynamic diameter of 183 nm (Figure S2). Liposomes of this diameter are routinely used for transport experiments by fluorescence spectroscopy and can be prepared reliably with a high degree of unilamellarity, in contrast to much larger vesicles, as used in the ¹³C NMR assay.¹² An aqueous solution of 225 mM NaCl was present both interior and exterior to facilitate HCO₃⁻/Cl⁻ exchange (antiport), which also contained 5 mM HEPES buffer to adjust the pH to 7.0 (Figure 3c). Anionophore 1 was preincorporated in the membrane of the LUVs and a NaHCO₃ solution was added to create a HCO₃⁻ concentration gradient of 10 mM (Figure S3). An increase in the intensity of the different emission bands of [Eu.L¹]⁺ was observed (Figure 3a) upon the addition of NaHCO₃. We chose to monitor the ΔJ=2 emission band around 615 nm (see Figure 3b), as this showed the largest increase, in agreement with observations in titrations of [Eu.L¹]⁺ with HCO₃⁻.²⁷ From here on, we will refer to these experimental conditions as the EuL1 assay.

The increase in the emission intensity over time following the addition of NaHCO₃ indicates that HCO₃⁻ has entered the liposomes. Since hardly any change in the emission intensity was observed in the absence of anionophore 1 (Figure 3d, black curve), we can conclude that bambusuril 1 transports HCO₃⁻ into the liposomes and that the new EuL1 assay allows this process to be monitored. The concentration of 1 was varied (Figure 3d) and a clear increase in the rate of transport was observed for increasing concentrations of anionophore 1. This shows that the EuL1 assay is highly sensitive and can be used to study the kinetics of HCO₃⁻ transport. Furthermore, these results reinforce our previous findings that bambusuril 1 is a very potent HCO₃⁻/Cl⁻ transporter,²⁰ showing activity even at 1 : 250,000 ratio, which corresponds to 1.6 nM concentration and an average of two bambusurils per LUV.

Differentiating the mechanisms of bicarbonate transport

The processes by which actual and apparent HCO₃⁻ transport can occur are schematically represented in Figure 4. The simplest mechanism for HCO₃⁻ transport is the antiport process with another anion, such as Cl⁻ (Figure 4, mechanism A). However, we should consider that addition of a pulse of NaHCO₃ to the exterior of the liposomes at pH<8 does not only create a gradient of HCO₃⁻, but also of its conjugate acid H₂CO₃²³ and of CO₂, formed upon dehydration (Scheme 1).³⁰ At equilibrium the concentration of CO₂ is almost 1000-fold higher than that of H₂CO₃ in aqueous salt solutions.³¹ Furthermore, it is well known that CO₂ can diffuse spontaneously across the membranes of cells that play important roles in HCO₃⁻ homeostasis,³² such as red blood cells and renal epithelial cells.³ Upon the addition of the HCO₃⁻ pulse, CO₂ could thus diffuse across the membranes of our liposomes. This increase in the concentration of CO₂ inside the liposomes would result in an acidification of the interior, causing a pH gradient to build up, which would stop the diffusion of CO₂. However, when transporters that can dissipate pH gradients are present in the membrane, the diffusion of CO₂ can continue, leading to a net increase in HCO₃⁻ concentration inside the liposomes, without this anion crossing the membrane (Figure 4B−D).

Figure 4

Different mechanisms by which apparent transport of HCO₃⁻ could occur. In mechanism A, anionophore (a) exchanges HCO₃⁻ for another anion – we refer to this as actual HCO₃⁻ transport. Mechanisms B−D rely on the diffusion of CO₂ coupled to transport of H⁺ or OH⁻ by cationophores (c) or anionophores to result in the net increase in HCO₃⁻ concentration, without this anion crossing the membrane.

Scheme 1

Representation of the interconversion between bicarbonate, carbonic acid, and carbon dioxide.

We indeed found that the addition of the cationophore monensin (H⁺/M⁺ antiporter)^{33, 34} to liposomes with [Eu.L¹]⁺ encapsulated gave a clear response upon addition of a HCO₃⁻ pulse (Figure 5, red curve). A similar response was observed when the combination of K⁺ transporter valinomycin³⁵ and the protonophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP)³⁶ were added (Figure 5, green curve), while those transporters added individually gave no significant response. Monensin gives similar transport curves in MCl, MNO₃, and M₂SO₄ solutions (where M⁺ is Na⁺ or K⁺, see Figure S8 and 6a), in agreement with the anion independent CO₂ diffusion mechanism B (Figure 4).

Figure 5

Increase in interior HCO₃⁻ concentration as monitored by the EuL1 assay in 112 mM K₂SO₄ with 5 mM HEPES at pH 7, upon addition of 10 mM KHCO₃ after 30 seconds. Different cation transporters were added to the LUVs at a transporter to lipid ratio of 1 : 1000.

No systematic differences were observed between the experiments using either sodium or potassium salts. This could mean either that monensin performs H⁺/Na⁺ and H⁺/K⁺ antiport at identical rates, or that the formation³⁰ and diffusion of CO₂ are rate limiting in mechanism B. To distinguish between these possibilities, we varied the concentration of monensin. While decreasing the monensin to lipid ratio from 1 : 1000 to 1 : 10,000 gave a lower rate of transport, increasing to a ratio of 1 : 100 did not significantly impact the rate of transport (Figure S9). This confirms that the diffusion of CO₂ is rate limiting in the observed increase in HCO₃⁻ concentration within the liposomes and not the H⁺/M⁺ antiport by monensin.

To verify if the pH inside the LUVs changes as expected upon diffusion of CO₂ (in the absence of ionophores) or upon dissipation of the pH gradient by monensin, transport experiments were performed in which the pH sensitive probe HPTS was encapsulated instead of the bicarbonate sensitive probe [Eu.L¹]⁺. All other conditions were identical to those used in the standard EuL1 assay (i. e., 225 mM NaCl, 5 mM HEPES, pH 7.0). The results in Figure 6d show that the addition of 10 mM NaHCO₃ to LUVs with monensin (1 : 1000 ratio) results in a rapid increase of the pH to 7.4 (red curve), indicating the equilibration of the pH gradient caused by the addition of the basic solution of NaHCO₃. In contrast, addition of NaHCO₃ to LUVs without transporters results in an acidification of the interior (black curve), in agreement with the formation of carbonic acid upon diffusion of CO₂. LUVs with a very low concentration of monensin (1 : 50,000) show an initial acidification of the interior due to CO₂ diffusion, followed by a slow increase of the pH due to the H⁺/Na⁺ antiport by monensin. These experiments with HPTS confirm that the apparent transport of HCO₃⁻ by monensin can be attributed to mechanism B. Furthermore, the pH equilibration by monensin at 1 : 1000 ratio (Figure 6d) is much faster than the apparent HCO₃⁻ transport revealed by the EuL1 assay (Figure 6a), providing further evidence that CO₂ diffusion (and/or formation³⁰) is rate limiting in the apparent transport of HCO₃⁻ by monensin (at 1 : 1000 ratio).

Figure 6

Increase in interior HCO₃⁻ concentration as monitored using the EuL1 assay (a–c) or change of the interior pH as monitored using the probe HPTS (d–f) in 225 mM NaCl with 5 mM HEPES at pH 7, upon addition of 10 mM NaHCO₃ after 30 seconds and lysis of the LUVs 10 minutes after that, to study transport by monensin (a,d), bambusuril 1 (b,e) and urea 2 (c,f). Monensin (1 : 1000 transporter to lipid ratio) was added to the experiments with anionophores 1 and 2.

Determining the transport mechanisms of different anionophores

Next, we studied the HCO₃⁻ transport by anionophores 1–4 in the EuL1 assay in NaCl (blue curves in Figure 6b,c and S12). A clear increase of the emission intensity was observed for all anionophores, even at the relatively low concentration of 1 : 25,000 (transporter to lipid ratio). A key question we wished to address was whether the observed increase in HCO₃⁻ in the liposomes is due to HCO₃⁻/Cl⁻ antiport transport mechanism A, or rather by CO₂ diffusion and pH gradient dissipation, as in mechanisms C and D (Figure 4).

Urea and thioureas with acidic N−H groups have been reported to not only transport anions, but also H⁺ (or OH⁻), and prodigiosin 4 is a known H⁺Cl⁻ transporter as well.³⁷ Indeed, rapid pH equilibration was observed for anionophores 2–4 (blue curve in Figure 6f and S21). In contrast, bambusurils have an electron deficient cavity formed by twelve polarised methine C−H groups, which can neither be readily deprotonated nor interact strongly with OH⁻.³⁸ Upon addition of NaHCO₃ to liposomes with bambusuril 1, a gradual increase in pH was observed, resembling the kinetics of the transport of the basic HCO₃⁻ anion into the LUVs (blue curve in Figure 6e vs 6b). This result concurs with our previous finding that 1 is unable to dissipate pH gradients by HCl symport or Cl⁻/OH⁻ antiport.²⁰ This excludes mechanisms C and D for this compound, leaving HCO₃⁻/Cl⁻ antiport (A) as the only possible transport mechanism for 1.

To distinguish the mechanisms involved in the apparent HCO₃⁻ transport by the other compounds, we have made use of the limited rate of CO₂ diffusion in the EuL1 assay, as observed from the experiments with monensin. For bambusuril 1, addition of monensin gives a clear increase in the rate of transport as seen from the comparison of the green to the blue curves in Figure 6b and S14. This increase can be understood from the combined effect of mechanism A by 1 and mechanism B by monensin, leading to a higher rate of apparent transport of HCO₃⁻ than by either of these two processes alone. In contrast, addition of monensin to LUVs with anionophores 2–4 did not increase the rate of transport, as shown for 2 in Figure 6c and for 3 and 4 in Figure S12. Because pH equilibration by these compounds is nearly instantaneous (Figure 6f and S21), the addition of a second pathway to dissipate the pH gradient (by monensin) will have no effect, as the overall rate of (apparent) HCO₃⁻ transport will remain limited by CO₂ diffusion. From this observation we can conclude that anionophores 2–4 primarily act via mechanism C or D.

To test if urea 2 and thiourea 3 can perform any HCO₃⁻/Cl⁻ transport (mechanism A), we have increased the concentrations of 2 and 3 in the membranes of the liposomes to 1 : 2500 (transporter to lipid ratio). The light blue curves in Figure 6c and S12a show that this ten-fold increase in transporter concentration leads to a significantly faster rate of (apparent) HCO₃⁻ transport, and that this overall rate clearly exceeds rates of transport that are limited by CO₂ diffusion (as observed in the curves for monensin ≥1 : 1000 ratio, see also Figure S13). From this we can conclude that HCO₃⁻/Cl⁻ antiport mechanism A also takes place. These compounds dissipate the pH gradient faster than they transport HCO₃⁻ and as a result C or D is the main mechanism, up to the point that CO₂ diffusion becomes rate limiting, after which mechanism A contributes to the apparent HCO₃⁻ transport. It is clear from these data that bambusuril 1 is the only “pure” HCO₃⁻ transporter studied, which functions without interference from other processes.

Bicarbonate uniport and antiport with nitrate

After demonstrating that our EuL1 assay can distinguish between actual and apparent HCO₃⁻ transport mechanisms, we used the assay to study the exchange of HCO₃⁻ with other anions, or uniport of HCO₃⁻. Commonly employed indirect methods to study HCO₃⁻ transport rely on the monitoring of Cl⁻ concentrations (Figure 1), preventing their use for studying exchange of HCO₃⁻ and NO₃⁻, or the uniport of HCO₃⁻. In contrast, [Eu.L¹]⁺ can operate in various salt solutions to study other processes than HCO₃⁻/Cl⁻ exchange. Hence, we utilised the EuL1 assay in NaNO₃ solution, to monitor HCO₃⁻/NO₃⁻ exchange (Figure 7a–c), and in a potassium gluconate (KGluc) solution in the presence of the K⁺ cationophore valinomycin, to study the uniport of HCO₃⁻ (Figure 7d–f).

Figure 7

Increase in interior HCO₃⁻ concentration in the presence of anionophores 1–4 monitored by the EuL1 assay in different salt solutions: a–c exchange with nitrate in 225 mM NaNO₃ with 5 mM HEPES at pH 7, and d–f uniport in 100 mM KGluc with 5 mM HEPES at pH 7 in presence of valinomycin. 10 mM NaHCO₃ (in b,c) or KHCO₃ (in e,f) was added after 30 seconds and the LUVs are lysed after 10 minutes. The schematic representations in a and d only show the mechanisms based on actual HCO₃⁻ transport, while mechanisms based on CO₂ diffusion could also take place.

Compounds 2–4 were found to exhibit efficient (apparent) transport of HCO₃⁻ in NaNO₃ (Figure 7c) and the rates do not change upon addition of monensin (see Figure S16), similar to the results obtained for these compounds in NaCl, indicating that the same combination of mechanisms is occurring. However, bambusuril 1 showed no transport at a 1 : 25,000 ratio, and only very slow transport was observed when using a 10-fold higher concentration of 1 (1 : 2500, Figure 7b). This slow HCO₃⁻/NO₃⁻ exchange by 1 resembles previous results reported for Cl⁻/NO₃⁻ exchange, which was found to be 100-fold slower than Cl⁻/HCO₃⁻ exchange by this bambusuril.²⁰ This large difference in Cl⁻/HCO₃⁻ and Cl⁻/NO₃⁻ exchange rates was explained by the very high affinity of 1 for NO₃⁻ (K_a=5×10¹¹ M⁻¹ in acetonitrile), which could prevent the release of this anion. In addition, it was proposed that simultaneous binding of a Cl⁻ and a HCO₃⁻ anion in the bambusuril could facilitate the exchange of these anions.²⁰ Even though the formation of an equivalent complex with NO₃⁻ and HCO₃⁻ simultaneously is possible, this does not appear to increase the rate of the exchange of these two anions by 1. Instead, the very strong binding of NO₃⁻ is the most probable cause for the low rates of HCO₃⁻/NO₃⁻ exchange by 1 (see also Figure S17).

This was further confirmed by the HCO₃⁻ uniport experiment in KGluc (Figure 7e), where HCO₃⁻ was efficiently transported by bambusuril 1. In this experiment valinomycin transports K⁺ to compensate for the displacement of charge associated to the HCO₃⁻ uniport, while the highly hydrophilic gluconate anion is not readily transported.²³ Under these conditions, it is highly unlikely that an anion exchange process takes place and instead bambusuril 1 will have to release the strongly bound HCO₃⁻ and return through the membrane without an anion bound.

In contrast, apparent transport of HCO₃⁻ by thiourea 3 and prodigiosin 4 was much slower when tested in uniport conditions (Figure 7f), compared to in the presence of Cl⁻ or NO₃⁻ (Figure 7c and S12). In NaCl and NaNO₃ the apparent HCO₃⁻ transport by these compounds was mainly attributed to H⁺/Cl⁻ or H⁺/NO₃⁻ cotransport in combination with CO₂ diffusion (mechanism C, or equivalent mechanism D). The poor rates of transport in KGluc indicate that these compounds are less efficient protonophores (H⁺ or OH⁻ transporters) than H⁺/Cl⁻ or H⁺/NO₃⁻ cotransporters and that they are not efficient HCO₃⁻ uniporters either. This result corroborates with other reports in which prodigiosin 4 was found to be a poor protonophore.³⁷ The rate of apparent HCO₃⁻ transport by urea 2 in KGluc is higher than those observed for 3 and 4 and the increase in the rate of transport with a higher concentration of 2 (Figure S18) indicates that 2 is able to perform actual HCO₃⁻ uniport (see SI for further mechanistic discussions). Nonetheless, bambusuril 1 is clearly the most efficient HCO₃⁻ uniporter tested.

Quantification of transport rates

To verify the qualitative trends and comparisons described above, we fitted the transport data from the EuL1 assay with single and double exponential functions, to obtain half-lives and initial rates respectively (see SI for details). Due to the slight differences observed in the equilibration levels of the different transport curves after normalisation and the effect of pH on the emission levels (see SI for a discussion), half-lives are more reliable to compare transport data, as these values indicate the time required to reach half of the final transport level and are thus a measure of how fast equilibrium is reached, independent of absolute emission values. The obtained values for the half-lives are given in Table 1 (see Table S1 for additional data). In NaCl, the comparison of half-lives of transport by 1 with and without monensin clearly shows that equilibrium is reached much faster in the presence of monensin (Table 1 and Figure S15), confirming the additivity of mechanisms A and B as discussed above. In contrast, the half-lives for 2–4 are nearly identical in the presence and absence of monensin, both in NaCl and in NaNO₃.

Table 1. Performance of anionophores 1–4 in the EuL1 assay.

Salt	Aniono- phore	Concentration (anionophore: lipid)	Half-life (s)^[a] without monensin	Half-life (s)^[a] with monensin
NaCl	None		*	82
	1	1 : 2500	10	4
	1	1 : 25 k	64	21
	2	1 : 2500	12	11
	2	1 : 25 k	51	50
	3	1 : 2500	12	11
	3	1 : 25 k	46	47
	4	1 : 25 k	59	74
NaNO3	None		*	81
	1	1 : 2500	*	67
	1	1 : 25 k	*	85
	2	1 : 2500	45	40
	2	1 : 25 k	89	85
	3	1 : 2500	16	14
	3	1 : 25 k	65	59
	4	1 : 25 k	61	68
KGluc^[b]	None		*	124
	1	1 : 25 k	83	38
	2	1 : 25 k	140	45
	3	1 : 25 k	180	42
	4	1 : 25 k	*	n.d.

• [a] Calculated from a single exponential fit of the transport curve, see ESI for details. [b] Transport in KGluc was studied in presence of valinomycin. * Transport was absent or too slow to quantify. n.d.=not determined.

Table 1 also shows that the overall half-lives obtained from the apparent HCO₃⁻ transport by anionophores 1–4 (in absence of monensin) are rather similar in NaCl. However, the different pH profiles could affect this comparison (see SI Section 2.10) and it would thus be better to compare the different transporters in the presence of monensin. Under those conditions, CO₂ diffusion-based mechanisms contribute to the transport for all the compounds, but as this process has a limited and thus constant rate, the differences in half-lives between anionophores 1–4 (in presence of monensin) can be attributed to the differences in rates of HCO₃⁻/Cl⁻ antiport (mechanism A) by the anionophores. In this comparison, bambusuril 1 is clearly the most active ionophore for HCO₃⁻/Cl⁻ antiport. Bis-urea 2 and bis-thiourea 3 show similar rates of transport and are slightly more active than prodigiosin 4, for which the half-life is very close to that of transport by monensin alone.

In contrast, in NaNO₃, transport by bambusuril 1 was too slow to be quantified, while addition of monensin resulted in half-lives that were identical or slightly lower than for monensin alone (for the lower and higher concentration of 1, respectively). Half-life values also indicate that apparent HCO₃⁻ transport in NaNO₃ by 2 is a bit slower than in NaCl, and that this is not the case for 3 and 4, which show similar rates in both salt solutions.

The values of half-lives obtained in KGluc and in presence of valinomycin clearly demonstrate that 1 is a much better HCO₃⁻ uniporter than any of the other transporters. Again, transport by 1 can be enhanced by adding both valinomycin and monensin due to an additivity of mechanisms. In contrast, while rates of 2 and 3 in presence of both valinomycin and monensin are faster than with valinomycin only (Table 1), these rates are the same as with monensin only (see Table S2), indicating that these compounds could perform HCO₃⁻/H⁺ symport (see SI for further discussion).

Comparison of the EuL1 assay with existing methods

Our EuL1 assay for monitoring HCO₃⁻ transport directly overcomes numerous disadvantages of existing methods, while combining their advantages, to offer a complementary tool in anion transport research (Figure 1). The only other direct HCO₃⁻ transport assay is based on ¹³C NMR spectroscopy,¹² which suffers from low sensitivity and poor time resolution. Furthermore, we observed that the cationophore monensin also gives a positive response in the ¹³C NMR assay for HCO₃⁻/Cl⁻ transport (Figure S22), which demonstrates that this assay cannot distinguish between CO₂ diffusion-based mechanisms and actual HCO₃⁻ transport. Indirect assays, such as ISE and lucigenin assays,^12-20 have not been able to provide mechanistic insights, nor allow comparisons between various HCO₃⁻ antiport and uniport processes. The osmotic HCO₃⁻ uniport assay is the only method reported so far that showed diffusion of neutral HCO₃⁻-based species in combination with H⁺ transport by monensin,²³ in agreement with our findings. However, the drawbacks of the osmotic assay are the very high concentrations of ionophores required due to the low sensitivity and the lower compatibility of the assay with preincorporated lipophilic transporters in the membrane during the preparation of the LUVs.

We have exploited the attractive features of the EuL1 assay to discover that bambusuril 1 can efficiently perform HCO₃⁻/Cl⁻ antiport and HCO₃⁻ uniport, while bisurea 2, thiourea 3, and prodigiosin 4 mainly combine CO₂ diffusion and pH gradient dissipation, leading to apparent HCO₃⁻ transport. Compounds 1–4 have previously been shown to act as HCO₃⁻ transporters in the lucigenin assay^{18, 20} and prodigiosin 4 also in the ISE and ¹³C NMR assays.¹² However, those experiments could not distinguish between actual and apparent transport of HCO₃⁻ anions. Notably, most of the HCO₃⁻ transporters reported in the literature resemble compounds 2–4 and can transport H⁺ or OH⁻.³⁷ This transport activity combined with CO₂ diffusion could be the mechanism of apparent HCO₃⁻ transport for many reported compounds. It is striking that selectivity for transport of Cl⁻ and NO₃⁻ over HCO₃⁻ has been reported for only two compounds, a biotinuril macrocycle and a bis-triazole,^{39, 40} which do not have acidic protons and are thus likely to be poor H⁺ and OH⁻ transporters.

Our results imply that only for compounds that show apparent HCO₃⁻ transport without dissipating pH gradients (such as 1) and for very potent anion transporters for which the rate of total apparent HCO₃⁻ transport surpasses the limited rate of CO₂ diffusion (2 and 3), we can conclude with certainty that these can act as actual anionophores for HCO₃⁻. This should be taken into account in the future development of HCO₃⁻ anionophores and can be readily verified with the EuL1 assay.

Conclusion

We have developed a new assay to directly monitor the transport of HCO₃⁻ into liposomes by fluorescence spectroscopy, using the encapsulated europium complex [Eu.L¹]⁺ that provides a luminescence increase upon binding HCO₃⁻. This assay provides a rapid and highly sensitive signal that enables anion transport kinetics to be determined and low concentrations of anionophores to be used. By combining anionophores with monensin in this direct and sensitive assay, it was possible to distinguish actual transport of HCO₃⁻ anions from alternative mechanisms based on CO₂ diffusion, which lead to an increase of HCO₃⁻ concentration in the liposomes without this anion crossing the membrane.

Our assay provides unprecedented insight into the mechanisms of HCO₃⁻ transport by anionophores and leads to the conclusion that we should doubt if many of the reported HCO₃⁻ transporters are actually capable of transporting this anion, or that they rather operate by dissipating the pH gradient resulting from CO₂ diffusion. Furthermore, the versatility of the assay compared to all existing assays was demonstrated by comparing HCO₃⁻/Cl⁻ and HCO₃⁻/NO₃⁻ antiport and HCO₃⁻ uniport processes for the first time.

We are convinced that the new opportunities provided by this assay to study transport of HCO₃⁻ efficiently and in new mechanistic detail will contribute to the further development of HCO₃⁻ transporters for biomedical purposes, such as channel replacement therapies.^{6, 41} As the anion binding properties of lanthanide probes such as [Eu.L¹]⁺ can be tuned for different anions,^{29, 42} the assay developed in this work could also be readily adapted to study transport of other anions. Furthermore, our work will also inform the future design of Eu(III) probes capable of monitoring spatio-temporal HCO₃⁻ dynamics within living cells. Indeed, a structurally related Eu(III) complex has already been shown to enter living cells and localise to specific subcellular compartments.⁴³ This feature, combined with the long luminescence lifetime of [Eu.L¹]⁺ and its derivatives augurs well for cellular imaging of HCO₃⁻ transport with high signal-to-noise, using time-gated fluorescence microscopy.

Acknowledgements

The results reported here are part of a project that has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (Grant agreement No. 802727). H.V. is a research associate of the Fonds de la Recherche Scientifique – FNRS. L.M.C. and H.V. also thank the ULB, “Fonds Van Buuren” and “Fonds Defay” for grants that enabled the purchase of the fluorescence spectrometer. S.J.B. and S.H.H. acknowledge the support of the EPSRC (EP/S032339/1) and the Wellcome Trust (204500/Z/16/Z). V.Š. thanks the Czech Science Foundation (No. 20-13922S).

Conflict of interest

The authors declare no conflict of interest.