Substrate interactions of the electroneutral Na⁺-coupled inorganic phosphate cotransporter (NaPi-IIc)

Abstract

The SLC34 solute carrier family comprises the electrogenic NaPi-IIa/b and the electroneutral NaPi-IIc, which display Na⁺ : P_i cotransport stoichiometries of 3 : 1 and 2 : 1, respectively. We previously proposed that NaPi-IIc lacks one of the three Na⁺ interaction sites hypothesised for the electrogenic isoforms, but, unlike NaPi-IIa/b, its substrate binding order is undetermined. By expressing NaPi-IIc in Xenopus oocytes, isotope influx and efflux assays gave results consistent with Na⁺ being the first and last substrate to bind. To further investigate substrate interactions, we applied a fluorometry-based technique that uses site-specific labelling with a fluorophore to characterize substrate-induced conformational changes. A novel Cys was introduced in the third extracellular loop of NaPi-IIc that could be labelled with a reporter fluorophore (MTS-TAMRA). Although labelling resulted in suppression of cotransport as previously reported for the electrogenic isoforms, changes in fluorescence were induced by changes in extracellular Na⁺ concentration in the absence of P_i and by changes in extracellular P_i concentration in presence of Na⁺. These data, combined with ³²P uptake data, also support a binding scheme in which Na⁺ is the first substrate to interact. Moreover, the apparent P_i affinity from fluorometry agreed with that from ³²P uptake, confirming the applicability of the fluorometric technique for kinetic studies of electroneutral carriers. Analysis of the fluorescence data showed that like the electrogenic NaPi-IIb, 2 Na⁺ ions interact cooperatively with NaPi-IIc before P_i binding, which implies that only one of these is translocated. This result provides compelling evidence that SLC34 proteins share common motifs for substrate interaction and that cotransport and substrate binding stoichiometries are not necessarily equivalent.

Renal reabsorption of inorganic phosphate (P_i) is mediated by secondary active P_i cotransporters that catalyse uphill P_i transport using the prevailing electrochemical Na⁺ gradient (Murer et al. 2008). Two gene products of the SLC34 family, NaPi-IIa (or SLC34A1) and NaPi-IIc (or SLC34A3) (Murer et al. 2004, 2008) are expressed at the renal brush border membrane of proximal tubules. A third member of the SLC34 family, NaPi-IIb (SLC34A2), is probably not expressed in the kidney (Hilfiker et al. 1998) but is present in several other organs such as the small intestine, where it is involved in the absorption of dietary phosphate, as well as lung, testis and liver. The importance of NaPi-IIa/c for renal P_i transport in the kidney is underscored by studies on knockout mouse models (Beck et al. 1998) and the pathophysiology of naturally occurring mutations (Bergwitz et al. 2006; Jaureguiberry et al. 2008).

All three SLC34 isoforms preferentially transport divalent P_i (HPO₄²⁻). NaPi-IIa and NaPi-IIb are electrogenic: they show a strict Na⁺ : P_i cotransport stoichiometry of 3 : 1, which results in the translocation of one net positive charge per transport cycle (Forster et al. 1999). In contrast, NaPi-IIc has a Na⁺ : P_i cotransport stoichiometry of 2 : 1 and is electroneutral (Segawa et al. 2002; Bacconi et al. 2005). Based on the electrogenic kinetics of NaPi-IIa/b expressed in Xenopus oocytes (Forster et al. 1997; Forster et al. 1998), we proposed an ordered substrate binding model in which one Na⁺ ion interaction precedes P_i binding, followed by the binding of the remaining two Na⁺ ions (Forster et al. 1998). This model was recently revised, based on data obtained from voltage clamp fluorometry (VCF) (Virkki et al. 2006b), a technique which combines electrophysiology and time-resolved fluorescence microscopy (e.g. Cha et al. 1998). That study, using the electrogenic NaPi-IIb, yielded compelling evidence for the cooperative binding of two Na⁺ ions before P_i binding and consequently implied that the binding of only one Na⁺ would be the last event before carrier translocation.

In this study, we ask the following question: what is the substrate binding order of the electroneutral NaPi-IIc? As no transport-related electrogenic activity (either steady-state or presteady-state) is detectable with NaPi-IIc, until now its kinetics have only been studied exclusively using radio-labelled isotopes (Segawa et al. 2002; Bacconi et al. 2005). Indeed, the 2 : 1 Na⁺ : HPO₄⁻ stoichiometry would suggest that NaPi-IIc lacks one of the three Na⁺ interactions proposed for NaPi-IIa/IIb; however direct experimental evidence to support this conclusion is lacking. Moreover, only limited kinetic information, such as apparent substrate affinities, is available (Segawa et al. 2002) and, in particular, the substrate binding order is unknown. We therefore addressed these issues, first by determining the substrate binding order on the extracellular side using traditional tracer uptake methods and second by using a fluorometric technique to study putative conformational changes during substrate interactions on individual Xenopus oocytes. In this latter technique, the protein is site-specifically labelled with fluorophore. As the fluorescence emitted by the fluorophore depends on its environment, a change in the fluorescence, induced by changes in membrane potential (in the case of electrogenic interactions) or different substrate concentrations, reflects local potential- or concentration-sensitive conformational changes in the protein. Here we demonstrate the use of fluorometry to document steady-state conformational changes of an electroneutral carrier induced by substrate interactions. Our flux and fluorescence data have allowed us to propose a kinetic scheme for electroneutral NaPi-IIc, which suggests that electrogenic and electroneutral SLC34 proteins share the same cation interactions despite having different transport stoichiometries.

Methods

Solutions and reagents

Standard extracellular solution (ND100) contained (in mm): 100 NaCl, 2 KCl, 1.8 CaCl₂, 10 Hepes, pH 7.4 adjusted with Tris. In Na⁺ substitution experiments NaCl was equimolarly replaced with choline chloride (ND0), whereas for Li⁺ replacing experiments NaCl was replaced with LiCl (LD100). Solutions with intermediate [Na⁺] or [Li⁺] were obtained by mixing ND100 or LD100 with ND0 in appropriate portions to maintain a constant overall molarity. P_i was added from a 1 m K₂HPO₄/KH₂PO₄ stock premixed to give pH 7.4. Phosphonoformic acid (PFA) was added to the test solution from 100 mm stock in water. Modified Barth's solution for storing oocytes contained (in mm): 88 NaCl, 1 KCl, 0.41 CaCl₂, 0.82 MgSO₄, 2.5 NaHCO₃, 2 Ca(NO₃)₂, 7.5 Hepes, pH 7.5 adjusted with Tris and supplemented with 5 mg l⁻¹ doxycyclin and 5 mg l⁻¹ gentamicin.

All standard reagents were obtained from either Sigma-Aldrich or Fluka (Buchs, Switzerland). 2-Aminoethyl methanethiosulfonate hydrobromide (MTSEA) and 2-(trimethylammonium)ethyl methanethiosulfonate bromide (MTSET) were obtained from Toronto Research Chemicals (North York, Ontario, Canada); 2-((5(6)-tetramethylrhodamine) carboxylamino)ethyl methanethiosulfonate (MTS-TAMRA) was obtained from Biotium (Hayward, CA, USA); tetramethylrhodamine-6-maleimide and tetramethylrhodamine-5-maleimide were obtained from Molecular Probes (Invitrogen/Life Technologies, Carlsbad, CA, USA).

Site-directed mutagnesis and cRNA preparation

The S437C mutation was introduced in mNaPi-II© WT using the Quickchange site-directed mutagenesis kit (Stratagene Inc., La Jolla, CA, USA). In brief, 10 ng of plasmid (pT7T3D-Pac) containing the WT mouse NaPi-IIc cDNA was amplified using 2.5 U of Pfu Turbo DNA polymerase (Stratagene Inc.), in the presence of primers (200 nm) containing the mutated codon sequence: GGCTGCCTTAGCCTGCCCTGCAGACATG. After PCR amplification, 10 U of DpnI was added to the amplification reaction and the sample was incubated at 37°C for 1 h to digest parental DNA. XL1-blue competent cells were transformed with the reaction mixture and then plated on LB plates supplemented with Ampicillin. The sequence was verified by sequencing (Microsynth, Balgach, Switzerland) and linearized with NotI, and cRNA was synthesized in the presence of Cap analogue using the T7 Message Machine kit (Ambion, Inc., Austin, TX, USA).

Animal handling and ethical approval

Female X. laevis frogs were purchased from Xenopus Express (Vernassal, France) or African Xenopus Facility (South Africa). Frogs were anaesthetized in 0.1% MS222 (tricaine methansulphonate) in water and portions of ovaries were surgically removed by making a small incision in the abdomen. After suturing the incision, the frog was placed in a separate tank to recover fully and then returned to a larger tank that contained all postoperative animals. A minimum of 8 weeks was allowed before re-operation on the same animal. All animal handling was in full compliance with regulations and recommendations of the University of Zurich (Institut für Labortierkunde) and the Swiss Federal Veterinary Office (FVO, Berne) from whom written approval was obtained. The authors have read the article by Drummond on reporting ethical matters and confirm that the experiments comply with the relevant policies and regulations (Drummond, 2009).

Expression in Xenopus laevis oocytes

Ovaries were cut into small pieces and treated for 45 min with collagenase (crude type 1A) 1 mg ml⁻¹ in ND100 solution (without Ca²⁺) in the presence of 0.1 mg ml⁻¹ trypsin inhibitor type III-O. Healthy stage V–VI oocytes were selected, maintained in modified Barth's solution at 16°C and injected with 10 ng of cRNA. Experiments were performed 4–7 days after injection.

Uptake experiments

Influx assays

Control oocytes and oocytes expressing NaPi-IIc (6–10 oocytes/group) were first allowed to equilibrate in ND100 solution without tracer. After aspiration of this solution, oocytes were incubated in ND100 solution containing 1 mm cold P_i and ³²P (specific activity 10 mCi mmol⁻¹ P_i). Uptake proceeded for 10 min and then oocytes were washed 3 or 4 times with ice-cold ND0 solution containing 2 mm P_i, and lysed individually in 10% SDS. The amount of radioactivity in each oocyte was measured by scintillation counting. The uptake time was chosen to be short enough to assume that initial rate conditions were satisfied so that uptake per unit time is a direct measure of transport velocity, without compromising measurement reliability, based on previous studies (Magagnin et al. 1993). Where indicated, oocytes were incubated with MTSET or MTSEA (1 mm in ND100) or MTS-TAMRA (0.4 mm in ND100) for 15 min prior to the uptake assay.

To determine the effective rate of labelling, oocytes were pre-incubated with ND100 containing MTS-TAMRA 0.4 mm for 5, 10, 15, 20, 30 for 45 min. ³²P uptake that remained after each successive application of MTS reagent was normalized to the value measured at t= 0 and plotted as a function of the exposure time. The data were fitted with a single decaying exponential to determine the effective second order reaction constant using an equation of the form:

(1)

where v^t is the transport rate (pmol oocyte⁻¹ min⁻¹) after incubation in the MTS reagent for a given a cumulative exposure time t, v° is the ³²P uptake rate without exposure to the MTS reagent, v^∞ is the ³²P uptake rate at infinite time, c is the concentration of MTS reagent and k is the effective second order rate constant (e.g. Virkki et al. 2006b).

For kinetic analysis, [P_i] was varied from 0.1 to 1 mm and [Na⁺] was varied from 0 to 125 mm. Data were plotted as a function of [P_i] or [Na⁺] and fitted with the Hill equation:

(2)

where [S] is the concentration of variable substrate (Na⁺ or P_i), v^max is the maximal velocity of transport, is the apparent affinity constant for substrate S, which in general depends on the invariant substrate and H is the Hill coefficient.

Efflux assays

Control and injected oocytes were injected with 50 nl of a solution containing 0.5 m NaCl, 10 mm Hepes (pH 7.4 adjusted with Tris), 10 mm P_i and ³²P. Taking the typical oocyte volume of 450 nl, based on our estimates of the diameter, which varied from 0.8 to 1.1 mm between individual oocytes and between batches, and assuming spherical geometry, injection would result in an intracellular [Na⁺] of >50 mm and [P_i] > 1 mm depending on the distribution of these solutes between free aqueous compartment and organelles that can account for up to 63% of the oocyte volume (e.g. Zeuthen et al. 2002).

After washing in ND0 solution, single oocytes were incubated in 100 μl of ND100 or ND0 with or without cold P_i. Oocytes were gently shaken manually in the vial containing the specified external solution to ensure adequate mixing before sampling. Aliquots (10 μl) were collected at the indicated times and radioactivity was determined by liquid scintillation counting. The values were corrected for the volume of the incubation solution and divided by the total amount of radioactivity present in the oocyte. This amount was calculated for each oocyte by adding the radioactivity in the extracellular sample of the last time point and the radioactivity remaining in the oocyte.

Fluorometry experiments

Equipment

The apparatus for simultaneous voltage clamp and fluorometry of Xenopus oocytes has been described in detail elsewhere (Virkki et al. 2006b). It comprises a two-electrode voltage clamp (OC725C, Warner Instruments, LLC, Hamden, CT, USA) with a laboratory-built fluorescence microscope based on a 10× fluorescence objective (CFI S Fluor, 0.5 N.A., 1.2 mm W.D., Nikon, Switzerland), and a filter set (XF33cube, comprising a 535DF35 excitation filter, 570DRLP dichroic mirror and 605DF50 emission filter; Omega Optical Inc.). Emitted light was measured using a silicon photodiode (S1336-18BQ, Hammatsu, Switzerland) connected to the input of an integrating patch clamp headstage (CV 201, Molecular Devices, Sunnyvale, CA, USA) and processed by a patch clamp amplifier (Axopatch 200A, Molecular Devices). The output of the Axopatch 200A was processed by a differential amplifier/filter unit (LPF-8, Warner Instruments) before digitalization. Fluorescence was excited by a 100 W halogen light source. To avoid photobleaching when not recording, an electronic shutter (VS252T1, Uniblitz, Vincent Associates, Rochester, NY, USA) was mounted between the light source and the optical unit. Oocytes where pre-incubated for 1 h in ND100 solution with MTS-TAMRA, TMR6M or TMR5M 0.4 mm at 20°C.

Protocols

Changes in fluorescence were recorded in response to changing substrate concentration and changing membrane potential. Changes in steady state fluorescence, ΔF, were measured in response to changes in: [Na⁺] (from 0 to 125 mm), [Li⁺] (from 0 to 125 mm), [P_i] (from 0.1 to 1 mm in ND100) or [PFA] (1 mm in ND100 or ND0). Each test substrate concentration application was bracketed with a control solution application (ND100) to allow for correction of a loss of fluorescence (see below). After application of the superfusate the oocyte was allowed to stabilize in the recording chamber for ∼2 min and then the shutter was opened for seven successive 230 ms intervals, to reduce photobleaching. For each change in solution, this protocol was applied 2–3 times. During the shutter opening time, the membrane potential was kept constant at −60 mV (to monitor substrate-dependent ΔF) or was stepped from V_h=−60 mV to −120 mV (to monitor voltage dependent ΔF).

Data analysis

During the course of the experiment, we observed two losses of fluorescence: (i) a systematic decrease in F that occurred even if the oocyte was only exposed once briefly at the start of an experiment, which was attributable to wash out of unbound dye and the internalization of previously labelled membrane protein; (ii) light-dependent photobleaching that was reduced by using the pulsed protocol described above. These losses in F necessitated off-line data correction, as follows. We corrected the light-independent decrease in fluorescence by plotting F at each application of the control solution (ND100) as a function of exposure time after the start of the experiment. These data were best fitted with a single decaying exponential to yield a standardized time course of loss of F in ND100. Interpolation of these data at the time point when we applied the test solution then allowed us to rescale the test measurement according to the predicted loss of F. We corrected the light-dependent decrease in F by determining the change in fluorescence (ΔF) for a defined solution change (ND100 to ND0 for the Na⁺ and Li⁺ dose dependence and ND100 to ND100 + P_i (1 mm) for the P_i dose dependence), plotting ΔF as a function of time and fitting with a single decaying exponential as for the light-independent loss component. These data were interpolated and the test ΔF rescaled appropriately according to the predicted ΔF at application time of the test solution. The underlying assumption for the validity of this second correction is that the photobleaching is a function of light exposure only and does not depend on the chemical composition of the perfusate.

After correction, ΔF obtained in response to changes in substrate concentration was plotted as a function of substrate concentration and fitted with the Hill equation:

(3)

where [S] is the substrate concentration (Na⁺, Li⁺, P_i), ΔF_max is the extrapolated maximum fluorescence, is the concentration of S that gives half-maximum response, H is the Hill coefficient and K is an offset constant. All curve fitting was performed using GraphPad Prism v 3.02/4.02 for Windows (GraphPad Software, San Diego, CA, USA). All data are given as mean ±s.e.m.

Results

Substrate binding order of WT NaPi-IIc determined by ³²P flux assays

We characterized the wild-type (WT) NaPi-IIc by heterologous expression in Xenopus laevis oocytes using standard ³²P uptake assays. In the first report of NaPi-IIc kinetics (Segawa et al. 2002), estimates of the apparent substrate affinities for P_i () and Na⁺ () were determined under saturating conditions for the invariant substrates; however from these data alone it is not possible to infer the binding order. To address this issue, we extended these influx assays by determining the substrate dependence at two concentrations of the invariant substrate. For the WT, the P_i dependence in presence of 100 mm or 50 mm Na⁺ showed the expected Michaelian characteristics with saturation for [P_i] > 0.3 mm (Fig. 1A). We fitted both data sets with the Michaelis–Menten equation (eqn (1), H= 1) to give estimates for V_max and . There was a clear ‘V_max’ dependence on external [Na⁺]: a reduction in [Na⁺] from 100 mm to 50 mm led to a concomitant reduction of V_max from 112 ± 11 to 64 ± 10 pmol oocyte⁻¹ min⁻¹. This behaviour accords with a substrate binding model in which Na⁺ is the last substrate to bind, assuming rapid equilibrium of substrate binding (e.g. Stein, 1990). Consistent with this finding, the estimated increased from 0.08 mm to 0.12 mm, which would also be expected for an ordered substrate interaction (Stein, 1990), although the results were not significant because of uncertainties in the fit (Table 1).

Figure 1. Substrate dependence of ³²P influx for WT NaPi-IIc.

A, P_i dependence of WT mediated ³²P-uptake for different [Na⁺]. ³²P uptake was measured at 0.01, 0.03, 0.1, 0.3 and 1 mm P_i in the presence of ND100 or ND50 (50 mm NaCl and 50 mm choline chloride) and plotted as a function of [P_i]. Data points were fitted with eqn (2) (continuous line). The fit parameters were: mm, V_max= 112 ± 11 pmol oocyte⁻¹ min⁻¹ (100 mm Na⁺) and mm, V_max= 64 ± 10 pmol oocyte⁻¹ min⁻¹ (50 mm Na⁺). Values are means ±s.e.m. (n= 6–8). Data for different oocytes were normalized to the mean uptake rate measured at 1 mm P_i (ND100) and expressed as a percentage. The dashed lines indicate the extrapolated fit beyond the measurement points. B, Na⁺ dependence of WT mediated ³²P-uptake for different phosphate concentrations. P_i uptake was measured at 0, 10, 25, 50, 75 and 100 mm Na⁺ in the presence of 1 mm and 0.1 mm P_i and plotted as a function of [Na⁺]. The fit parameters were: = 43 ± 3, V_max= 114 ± 5 pmol oocytes⁻¹ min⁻¹ (1 mm P_i); = 67 ± 9, V_max= 125 ± 19 pmol oocytes⁻¹ min⁻¹ (0.1 mm P_i). Values are means ±s.e.m. (n= 6–8). Data for different oocytes were normalized to the mean uptake rate measured at 1 mm P_i (ND100) and expressed as a percentage. The dashed lines indicate the extrapolated fit beyond the measurement points. C, efflux assays for WT NaPi-IIc and non-injected (NI) oocytes from the same donor frog for different conditions in the external medium (filled symbols, + 1 mm P_i; open symbols, 0 mm P_i: circles, ND100; squares, ND0). Efflux was measured in ND100 or ND0 solution with or without P_i (1 mm) and plotted as a function of time. Points have been joined by lines for visualising purposes only. Values are means ±s.e.m. (n= 9).

Table 1.

Comparison of phenomenological constants obtained from ³²P assays, fluorescence (ΔF) and electrophysiology (TEVC)

Fit parameter	(mm)		(mm)
Measurement condition	100 mm Na+	50 mm Na+	1 mm Pi	0.1 mm Pi	100 mm/50 mm Na+; 1 mm Pi
WT NaPi-IIc (32P)	0.08 ± 0.03	0.12 ± 0.06	46 ± 8	79 ± 9	0.57 ± 0.06
S437C (32P)	0.08 ± 0.02	0.10 ± 0.03	46 ± 3	79 ± 9	0.42 ± 0.04
S437C (ΔF)	0.07 ± 0.01	nd	79 ± 8	nd	na
WT NaPi-IIc (32P)1	0.07	nd	48 ± 3	nd	nd
WT NaPi-IIa (TEVC)2	0.054	0.27	50.1	89	0.67

Data determined in this study are shown bold; na = not applicable, nd = not determined.

¹Segawa et al. (2002).

²Forster et al. (1998).

To further support this conclusion, we measured Na⁺-dependent ³²P uptake for NaPi-IIc at two [P_i]: close to the expected (0.1 mm) and close to saturation (1 mm). We varied the [Na⁺] in the extracellular uptake solution from 0 mm to 100 mm. These data showed the expected sigmoidicity predicted for a system in which more than one Na⁺ ion interacts with the carrier and we found the most satisfactory fit using the Hill equation (eqn (1)) with the Hill coefficient, H > 1 (Fig. 1B). With 1 mm P_i, the fit gave = 46 ± 3 mm, which is close to the value (48 ± 9 mm) previously reported (Segawa et al. 2002) and H= 2.1 ± 0.2. By decreasing [P_i] from 1 mm to 0.1 mm, we observed that the predicted V_max did not change significantly (1 mm P_i: 115 ± 6 pmol oocyte⁻¹ min⁻¹; 0.1 mm P_i: 125 ± 19 pmol oocyte⁻¹ min⁻¹), which was also consistent with Na⁺ being the last ion that interacts with the transporter before translocation. Moreover, as predicted for an ordered binding model, increased to ∼80 mm (Table 1), whereas the Hill coefficient remained reasonably constant (1.7 ± 0.2).

Although these experiments strongly suggest that Na⁺ is the last substrate to bind before translocation, there are, however, at least two possibilities for the order of the preceding partial reactions: they could be strictly ordered, in which Na⁺ precedes P_i (or vice versa) or random, in which either Na⁺ or P_i interacts depending on the relative activity of the substrate. To investigate these possibilities, we examined the effect of external substrate on ³²P efflux. We injected oocytes with 50 nl aliquots of an injection buffer containing ³²P, so that after ∼10 : 1 dilution in the cytosol, the internal [Na⁺] would be at least 50 mm, which would thereby facilitate establishing an outward Na⁺ gradient. We then examined the effect of four combinations of external substrates on P_i efflux (Fig. 1C) at three time points after incubation in the respective medium. The efflux rates based on the 10 min measurement point were significantly smaller than the corresponding influx rates with comparable driving force. For NaPi-IIc injected oocytes, we measured an efflux rate of 0.18 ± 0.03 pmol oocyte⁻¹ min⁻¹ (n= 4) when incubated in ND0 solution alone compared with typically 40 pmol oocyte⁻¹ min⁻¹ for influx measurements in ND50. Moreover, insignificant efflux was measured from non-injected oocytes under the same external conditions. For NaPi-IIc expressing oocytes, the efflux rate was non-linear for all the conditions and suggested that saturation would occur at times > 60 min. This may result from substrate accumulation due to unstirred layer effects at the external face of the oocyte membrane.

Under all three incubation conditions, the ³²P efflux by control oocytes was negligible (< 1%) compared to those expressing NaPi-IIc (Fig. 1C), which established that the efflux was a direct result of expression of NaPi-IIc. In the absence of external Na⁺, a significant efflux was measured that was unaffected by external P_i. The lack of trans-inhibition of efflux by external P_i in the absence of external Na⁺ is consistent with a scheme that requires Na⁺ to bind before P_i (see Discussion). In contrast, the presence of external Na⁺ (100 mm) in the external medium stimulated efflux approximately 2-fold (at 60 min) and there was a further ∼25% stimulation when P_i was also present in the external medium.

Introducing a Cys at Ser-437 does not alter basic transport properties

To gain further insight into the interactions of substrates with NaPi-IIc, we used a fluorometric approach in which a change in emitted fluorescence that results from changes in a fluorophore's micro-environment are assumed to reflect local, substrate-induced conformational changes. This approach requires introducing a novel cysteine in the protein at a functionally important site that can be conveniently covalently labelled with the fluorophore from the external medium. Based on previous studies, we chose the site Ser-437 at the C-terminal end of the putative, extracellularly oriented, re-entrant segment comprising transmembrane domains 8 and 9 (Fig. 2A). This site is predicted to be a region highly accessible from the extracellular solution and is close to the predicted substrate translocation pathway (Lambert et al. 2001; Kohler et al. 2002). The equivalent Ser–Cys substitution in the electrogenic isoforms (NaPi-IIa, NaPi-IIb) is well-tolerated with only minor changes in transport properties. After incubation in methanethiosulfonate (MTS) reagents, including MTS conjugated with the fluorophore rhodamine, electrogenic cotransport activity and P_i transport are significantly suppressed, yet substrate interactions are retained (Lambert et al. 1999; Virkki et al. 2006b).

Figure 2. Mutant S437C shows similar transport properties to WT NaPi-IIc.

A, topological model of SLC34 proteins indicating the site of the Ser–Cys substitution for mutant S437C (filled square). The model proposes 12 transmembrane domains (numbered), 2 opposed reentrant loops and intracellular C- and N- termini (e.g. Virkki et al. 2007). The sites substituted in the AAD mutant that resulted in a NaPi-IIc electrogenic transporter are also indicated (open squares) (Bacconi et al. 2005) (see Discussion). B, oocytes injected with cRNA of WT (open bars) or S437C mutant (filled bars) were assayed 3 days after injection for ³²P uptake in the presence of standard uptake solution (ND100) or Na⁺-free uptake solution (equimolar replacement of Na⁺ by choline). Values are means ±s.e.m. (n= 5–6). C, ³²P uptake measured for WT and S437C using standard uptake solution (ND100) in the absence or presence of PFA (1 mm). Values are means ±s.e.m. (n= 6–7). Data for different oocytes were normalized to the mean uptake rate measured at 1 mm P_i in the absence of PFA and expressed as a percentage.

As for WT NaPi-IIc, P_i transport by S437C was Na⁺ dependent, and moreover, the Na⁺-dependent transport activity of oocytes expressing WT NaPi-IIc and S437C for oocytes from the same donor animal was very similar (Fig. 2A). This indicated that under these assay conditions, the mutation most likely had not altered the turnover rate. We next repeated the P_i dependence and Na⁺ dependence assays (Fig. 1A and B) and obtained similar estimates for the apparent substrate affinities to the WT (see Table 1). Finally, we investigated the effect on cotransport activity of the Na⁺/P_i cotransport inhibitor, phosphonoformic acid (PFA), which is a competitive inhibitor of P_i transport mediated by SLC34 proteins (Busch et al. 1995; Villa-Bellosta et al. 2007). We confirmed that PFA also inhibits uptake mediated by NaPi-IIc and S437C by performing radio-labelled P_i uptake assays on oocytes incubated with P_i (0.3 mm) or both P_i (0.3 mm) and PFA (1 mm), and a significant decrease in ³²P uptake was documented (Fig. 2C). Taken together, these data confirmed that the Ser–Cys substitution had not affected the basic characteristics of the protein.

Changes in fluorescence reported by mutant S437C after labelling with MTS-TAMRA

We next investigated the properties of S437C after labelling with MTS reagents. Oocytes that expressed either the WT or S437C were assayed for Na⁺-dependent ³²P uptake after incubation with MTSEA, MTSET or MTS-TAMRA (15 min, 1 mm). For the WT, uptake was unaltered (Fig. 3A), as we have previously reported for the WT NaPi-IIa (Lambert et al. 1999; Ehnes et al. 2004). This confirmed that there are no functionally important Cys residues accessible in the WT. In contrast, ³²P for S437C was significantly reduced after preincubation in all three reagents (Fig. 3A). This finding agrees with the behaviour that we have previously reported for the equivalent mutants in the electrogenic rat NaPi-IIa (S460C) (Lambert et al. 1999) and flounder NaPi-IIb (S448C) (Virkki et al. 2006b). It supports our conclusion that this site is critical among all SLC34 gene products. For incubation in MTS-TAMRA, we observed that there was approximately 25% transport activity remaining after Cys modification. This might indicate that the concentration/exposure time was insufficient to complete the modification of all expressed proteins, or that the Cys modification does not cause a complete loss of activity of the protein, as we have previously observed for some Cys mutants (e.g. Lambert et al. 2001; Ehnes et al. 2004).

Figure 3. Effect of MTS on WT and S437C transport activity.

A, oocytes expressing WT or S437C were incubated for 15 min with MTSEA, MTSET or MTS-TAMRA (1 mm) and then ³²P uptake was assayed. Values are means ±s.e.m. (n≥ 6). B, groups of 6–8 oocytes expressing WT (open squares) or S437 (filled squares) were pre-incubated for different times (5, 10, 20, 30, 45 min) in MTS-TAMRA (0.4 mm) and ³²P uptake was measured in the control solution (ND100). ³²P uptake after each MTS-TAMRA exposure was normalized to the value at t= 0 and expressed as a percentage. Data points were fitted with a single decaying exponential (continuous line) to obtain the effective second-order rate constant for Cys modification. For WT, data points are connected by a dotted line to better visualize the absence of effect of MTS-TAMRA on P_i uptake.

To investigate the underlying reasons for this behaviour, we exposed oocytes expressing WT or S437C for longer times and assayed the uptake. WT NaPi-IIc showed no significant alteration in activity for incubation periods up to 45 min, whereas the activity of S437C decreased progressively and reached a plateau corresponding to approximately 10% of the initial response at each test exposure (Fig. 3B). By assuming the MTS reagent to be in excess, we can consider the decrease in activity to result from a pseudo first order reaction and by fitting these data with a decaying exponential (eqn (1)) an effective second order rate constant, k= 0.055 ± 0.001 mm⁻¹ s⁻¹, was obtained (e.g. Karlin & Akabas, 1998). This is an indirect measure of the accessibility of the Cys-437 by MTS-TAMRA and is at least two orders of magnitude smaller than values we have previously reported for the rat NaPi-IIa S460C (Lambert et al. 2001) and the flounder NaPi-IIb S448C (Virkki et al. 2006b). The apparent reduced accessibility may result from the larger size of the conjugated fluorophore compared with the other MTS reagents or a lower accessibility of the site for NaPi-IIc compared with the electrogenic isoforms. The baseline activity for long exposure may result from the insertion of unlabelled protein that we have also observed in the case of the electrogenic isoforms (I. C. Forster and L. V. Virkki, unpublished observations). From these data we decided to routinely incubate S437C in 1 mm MTS-TAMRA for 60 min.

Oocytes incubated with MTS-TAMRA showed a reversible increase in fluorescence when the superfusion solution was changed from ND100 to ND0 (i.e. removal of external Na⁺). We observed no changes of background fluorescence in response to changes in the external medium in non-injected or water-injected control oocytes, which indicated that the fluorescence response is unique to oocytes that express S437C (data not shown). A representative recording (Fig. 4A) shows two additional features. First, there was a decrease in change in fluorescence (ΔF), which we attribute to photobleaching. Although this could be reduced by limiting the time the shutter was open (230 ms), it still accounted for up 50% of the decrease in ΔF. Second, we observed another type of fluorescence loss that was independent of exposure to light (dashed line, Fig. 4A). This fluorescence loss is probably due to the wash-out of unbound dye or to the internalization of labelled protein and it follows a mono-exponential decay.

Figure 4. Fluorescence measurements in the steady state.

A, original fluorescence traces recorded in presence of different Na⁺ concentrations at the holding potential of −60 mV recorded with a Minidigi 1A (Molecular Devices). The ordinate is the change in fluorescence expressed as a percentage of the initial background fluorescence signal when the shutter was first opened. V_h was kept constant at −60 mV and the shutter was open for 7 successive 230 ms pulses followed by a 2 s closed time. For each [Na⁺] concentration, the protocol was repeated 2 or 3 times and to control for loss of F in each test solution (ND0, in this case) the assay was bracketed by ND100. To better visualize the progressive loss of fluorescence, the signal (mean of the 7 samples) recorded in the presence of ND100 is connected by a dashed line. Note that the change in fluorescence (ΔF) induced is smaller for the second set of measurements in ND0 because of photobleaching (see Methods). B, F measured in ND100 (filled squares) is plotted as a function of time and fitted with a single decaying exponential (continuous line) to quantify loss of fluorescence. The effective change in fluorescence (ΔF) induced by a change in substrate concentration is calculated as the difference between the value experimentally obtained and the value predicted from the fit. Error bars smaller than symbol size are not shown.

As described in Methods, we calculated ΔF in response to a solution change from the standard [Na⁺] (ND100) to the test solution as the difference between the value recorded and that predicted from the fit when the oocyte was superfused with ND100. Figure 4B shows the same data replotted as mean F of the individual seven measurements plotted on an absolute time scale. The same time-dependent decrease in F was also found in the absence of Na⁺ (ND0), which confirmed that the effect was not due to substrate interaction with the fluorophore itself (data not shown).

We performed these pilot experiments using MTS-TAMRA that we have previously used for VCF studies on the flounder NaPi-IIb mutants (Virkki et al. 2006a,b;). As MTS-TAMRA is a mixed isomer and may therefore result in a reduced ΔF, we compared the response with that from separate labelling with single isomers (tetramethylrhodamine-6-maleimide and tetramethylrhodamine-5-maleimide). Oocytes were incubated with each of the fluorophores (final concentration 0.4 mm) in the presence of ND100 for 1 h. The main difference among the three fluorophores was the value of background fluorescence: for tetramethylrhodamine-6-maleimide and tetramethylrhodamine-5-maleimide the background fluorescence, taken as the steady-state photodiode current relative to the dark condition, was respectively 6.4-fold and 5.3-fold higher than that observed for MTS-TAMRA (data not shown). This higher background fluorescence was probably due to unspecific interaction with endogenous protein expressed in the oocyte and the consequence is a lower ΔF/F ratio. Therefore, we performed all subsequent experiments with the mixed isomer.

Voltage and cation dependence of fluorescence

P_i translocation mediated by SLC34 proteins is coupled to the electrochemical gradient of Na⁺. Our finding that a change in F accompanies a change in external [Na⁺] provided compelling evidence that Na⁺ ions interact with NaPi-IIc before P_i binding, as we have previously reported for the electrogenic NaPi-IIb isoform (Virkki et al. 2006b). In that study, we showed that the dependence of F on external [Na⁺] is cooperative, which indicated that more than one Na⁺ ion interacts with the protein before P_i binding. Given that NaPi-IIc is electroneutral and involves translocation of two Na⁺ ions per divalent P_i, and if we assume that the last substrate interaction before translocation is Na⁺ dependent (Fig. 1A), we would therefore predict that for an ordered binding scheme, only one Na⁺ ion should interact with S437C in the absence of P_i.

To test this hypothesis, we examined ΔF with increasing [Na⁺] in the absence of P_i (Fig. 5A). We recorded the highest fluorescence in 0 mm Na⁺ (ND0), which was then progressively quenched as [Na⁺] increased. There was clear evidence of saturating behaviour, although we did not increase [Na⁺] above 125 mm because the oocytes did not generally tolerate higher [Na⁺], particularly for long exposure times. These data reveal a sigmoidal relationship between ΔF and [Na⁺] and when fitted with the Hill equation (eqn (2)) the fit predicted an apparent affinity for Na⁺, = 78.7 ± 7.7 mm and a Hill coefficient, H= 1.8 ± 0.1. Thus, although the MTS-TAMRA-labelled S437C no longer transports P_i, Na⁺ ions interact with S437C and in a cooperative manner: the Hill coefficient indicated the cooperative binding of more than one Na⁺ ion in the absence of P_i.

Figure 5. Sodium and lithium dependence of fluorescence change (ΔF) in oocytes expressing S437C.

A, Na⁺ dependence of fluorescence (ΔF). Fluorescence was recorded at different [Na⁺] and data were fitted with eqn (3) (continuous line). Values are given as means ±s.e.m. (n= 4 for [Na⁺]= 125 mm and 75 mm; n≥ 7 for [Na⁺]= 100 mm, 0 mm, 25 mm and 50 mm). Error bars smaller than symbol size are not shown. B, voltage dependence of fluorescence (ΔF). Fluorescence was recorded at V_h=−60 mV, 0 mV and −120 mV and normalized to the value obtained at −60 mV. Values are means ±s.e.m. (n= 5). C, Li⁺ dependence of fluorescence (ΔF). Fluorescence was recorded at different Li⁺ concentrations and data were fitted with eqn (3) (continuous line). The fit gave an apparent affinity constant for Li⁺, = 377 ± 111 mm. Values are means ±s.e.m. (n≥ 4). Error bars smaller than symbol size are not shown. D, ³²P uptake was measured for NaPi-IIc WT in presence of different superfusates as indicated (concentrations in mm). Values are means ±s.e.m. (n= 8). There was no statistical significance between the uptake for 50 mm Na⁺+ 50 mm Li⁺ and 50 mm Na⁺+ 50 mm Chol.

For electrogenic NaPi-IIb, we have also shown that the dependence of ΔF on [Na⁺] is voltage dependent (Virkki et al. 2006a,b;). As it is possible that one or more partial reactions in the transport cycle of electroneutral carrier may still be electrogenic (e.g. Lester et al. 1994), we next investigated if this was the case for S437C. We compared ΔF obtained in ND0 solution at three potentials: 0 mV, −60 mV and −120 mV. We were unable to detect any significant difference in ΔF, which strongly suggested that the partial reactions that reflect the localized conformational changes near Cys-437 are voltage independent (Fig. 5B).

Sodium is the only cation that acts as a cosubstrate for NaPi-II isoforms, although previous studies have shown that H⁺ (Forster et al. 2000; Virkki et al. 2005b) and Li⁺ (Virkki et al. 2006a,b;) interact. We therefore examined the effect of replacing Na⁺ with equimolar Li⁺ and we found that ΔF decreased with increasing [Li⁺] (Fig. 5C). We fitted the Li⁺ dependence with the Hill equation (eqn (3)). The unconstrained fit predicted a Hill coefficient, H= 1.07 ± 0.23, which suggested, as we have previously shown, that only one Li⁺ interacts with the electrogenic NaPi-IIb (Virkki et al. 2006b). When we constrained H= 1 (Fig. 5C) to reduce the fit uncertainty, this procedure yielded an apparent affinity constant for Li⁺, = 377 ± 111 mm which, given the fit error, is comparable to the value reported using VCF (Virkki et al. 2006b). These results indicated that, as observed for NaPi-IIb (Virkki et al. 2006b), Li⁺ ions can also interact with NaPi-IIc, but in an apparently non-cooperative manner.

If Li⁺ ions interact with the transporter as suggested by the above fluorometric findings, can this interaction alter Na⁺-dependent transport rate? To answer this question, we performed ³²P uptake at 50 mm Na⁺, close to the predicted and compared having either choline or Li⁺ as the remaining cation (Fig. 5D). As expected, complete replacement of Na⁺ with either Li⁺ or choline resulted in insignificant transport activity for the WT. Moreover, we observed no significant difference in uptake in the presence of 50 mm Na⁺ when Li⁺ or choline constituted 50% of the main monovalent cations. S437C showed similar behaviour (data not shown). These findings established that Li⁺ appears to have no direct effect on the cotransport activity with saturating P_i.

P_i and PFA induce fluorescence changes in presence of Na⁺ ions

We analysed ΔF due to the presence in the extracellular solution of 1 mm P_i in ND100 or ND0. In the absence of Na⁺ ions in the external medium (ND0), P_i induced no change in F relative to the control condition (ND0). However, in ND100, when we added 1 mm P_i to the superfusate, we observed an increase of F relative to ND100 alone (Fig. 6A). This increase was approximately 30% of the maximal change in fluorescence observed in ND0 alone relative to F in ND100.

Figure 6. P_i and PFA dependence of fluorescence change (ΔF) in oocytes expressing S437C.

A, comparison of the effect of substrate on fluorescence quenching in ND100 and in ND0, in the presence of P_i (1 mm) (open bars) or PFA (1 mm) (hatched bars). Data are expressed as a percentage of the total fluorescence quench when changing from ND0 to ND100 measured in the absence of substrate. Values are means ±s.e.m. (n≥ 5). B, ΔF was recorded at different P_i concentrations (0, 0.1, 0.3 and 1 mm) and data were fitted with eqn (3) (H= 1). The fit gave an apparent affinity constant, mm. Values are means ±s.e.m. (n≥ 5). Error bars smaller than the symbol size are not shown.

As PFA is a competitive inhibitor of Na⁺–P_i cotransport, we would expect it to interact at the same site as P_i. Therefore, we would predict that PFA would cause a similar ΔF as P_i. To test this prediction, we recorded ΔF in the presence of 1 mm PFA, by superfusing in ND100 or ND0, and the data were compared with ΔF recorded in the presence of P_i. Here, we observed that 1 mm of either PFA or P_i in ND100 gave a similar percentage quench of fluorescence and this accounted for approximately 65% of the total quench observed when changing from ND0 to ND100 (Fig. 6A). This suggested that the conformational change induced by the interaction of either substrate with the carrier, in the micro-environment of Cys-437, is the same, in support of our prediction. When the same experiment was performed in the absence of external Na⁺ (ND0), no significant fluorescence quenching was observed apart from the normal loss of F (see Methods). Taken together, these results indicate that both P_i and PFA require Na⁺ in the extracellular solution to interact with the transporter and suggest that they share a common binding site.

Finally, we observed that the change in fluorescence induced by P_i showed a dose dependence that was well described by fitting the data with the Michaelian form of eqn (3) (H= 1) (Fig. 6B). The fit predicted an apparent affinity constant for P_i, mm, which is close to that estimated by means of uptake for the WT or S437C before MTS incubation (Table 1). The similarity of the P_i dose dependence for the normal and modified protein can be readily seen by superimposing the normalized fluorescence and uptake data (Fig. 6B). Taken together, these findings suggested that after incubation with MTS-TAMRA, P_i still interacts with the mutant in a similar manner to the fully functional protein, although P_i cotransport is blocked in the former case.

Discussion

Solute cotransport involves the interaction of one or more cations with the carrier to initiate and facilitate substrate binding and translocation. Ideally there is tight coupling between the driving cations and the driven solute, and translocation can only take place when the carrier is fully loaded. This implies no ‘uncoupled’ leak of any substrate, a fixed number of substrate binding/interaction sites on the protein and strict ‘rules’ for occupancy and accessibility to these sites (e.g. Rudnick, 2006), which accord with the canonical alternating access model for cotransport. It follows that the nature of the substrate–carrier interactions that precede translocation of the loaded carrier is essential to characterise the cotransport mechanism of a given carrier. Several generalised models have been proposed to account for coupled transport and to distinguish between, for example, simultaneous and consecutive translocation of substrates and random or ordered binding. Here, we provide evidence, by integrating the findings from isotope flux assays and a novel fluorometric technique, which supports a kinetic scheme in which Na⁺ is the first and last substrate to interact and bind to the outward facing conformations of the carrier before translocation.

First, based on our influx experiments on the WT NaPi-IIc, our data are consistent with Na⁺ being the last to bind before translocation of the fully loaded carrier because the maximum transport velocity (V_max) shows a Na⁺ dependence under saturating P_i. This conclusion is further supported by the results of the complementary experiment in which the external [Na⁺] was varied at two [P_i] and no ‘V_max’ effect was observed. This behaviour is also consistent with rapid equilibrium conditions applying to at least the P_i interaction with NaPi-IIc (for discussion, see Berteloot, 2003). As summarized in Table 1, for the P_i dependence assays, the ratio of V_max in ND50 to that in ND100 for the WT NaPi-IIc, and mutant S437C were similar whereas, for the equivalent measurement using the electrogenic NaPi-IIa, the ratio was higher (Forster et al. 1998). This may reflect the presence of an additional driving force (−50 mV) and 3 : 1 stoichiometry with respect to Na⁺ ions or isoform-specific differences.

The influx experiments are consistent with at least two ordered kinetic schemes illustrated in Fig. 7A (left panel), in which the empty carrier (state 1) is sequentially loaded with substrates (state 4), and translocates and releases them to assume an inward facing empty carrier conformation (state 6) before returning to state 1. To determine which of these two schemes is the more likely using only influx assays would necessitate differentiating between the predicted dependencies of V_max on [Na⁺] for each scheme. This may be difficult to achieve as the range of [Na⁺] possible with the oocyte system is limited, thereby leading to ambiguous interpretations.

Figure 7. Evolution of a kinetic scheme for NaPi-IIc.

A, influx and efflux experiments suggest an ordered binding scheme for substrate interactions on the outside. Each number represents a conformation (state) that the carrier occupies in sequence during the transport cycle. Preferred partial reactions: bold (influx: black; efflux: grey); unfavourable reactions: dotted. Influx experiments (left panel) allow discrimination between the boxed binding schemes and are consistent with Na⁺ being the last substrate interaction before translocation (lower box). The P_i interaction (2⇌3) is in rapid equilibrium (shaded). We cannot readily distinguish between the two schemes so that first substrate interaction remains undefined. Efflux experiments (right panel) are consistent with Na⁺ being the first substrate to bind. In the absence of external Na⁺, the addition of P_i to the outside does not affect the efflux rate because transition 2→3 cannot occur. The low efflux rate under zero trans conditions occurs via the ‘unprimed’ cycle (boxed). The ‘primed’ cycle occurs when external Na⁺ is available and first involves partial reaction 1→1′ to bind the ‘catalytic’ Na⁺ ion (see text). With external Na⁺ and P_i available, the carrier adopts an exchange mode (4′⇌5′) (shaded). B, scheme comparing the essential differences in substrate interactions for the electroneutral NaPi-IIc (upper) and electrogenic NaPi-IIa/b (lower) isoforms. Two conformations are shown: binding sites externally oriented (left) and internally oriented (right). For NaPi-IIc, the first Na⁺ to interact is postulated to be immobile during the whole cotransport cycle, but acts as a catalytic activator for subsequent substrate binding to give a 2 : 1 cotransport stoichiometry. For NaPi-IIa/b, an additional intrinsic mobile charge establishes a binding site and translocation pathway for the first Na⁺ ion within the transmembrane electric field. In the absence of external P_i, this Na⁺ ion can leak through the protein (uncoupled leak mode), whereas in the presence of external P_i, it participates in the transport cycle and is released along with P_i and 2 more Na⁺ ions to give a 3 : 1 cotransport stoichiometry. Upon release of substrates on the inside, the intrinsic charge senses the transmembrane field and its movement causes a voltage-dependent reorientation of the empty carrier to expose the activating Na⁺ ion binding site once more to the external medium.

Our second experimental strategy involved preloading oocytes with substrate and observing the effect of external substrate interactions on efflux rates (e.g. Turner, 1981, 1984; Beliveau & Strevey, 1988). The efflux rates were at least two orders of magnitude smaller than influx rates with comparable driving force conditions. This may indicate that NaPi-IIc is asymmetrical and optimised for forward mode transport: the apparent substrate affinities on the internal side are low and one or more partial reactions in the reverse cycle are rate limiting. On the other hand, we cannot exclude that the amount of injected P_i (and ³²P) that remains free for transport is significantly reduced due to intracellular sequestering and buffering.

Our findings strongly suggest that Na⁺ is also the first substrate to interact with the empty carrier on the extracellular side. With no external substrates (i.e. zero trans), NaPi-IIc will proceed anticlockwise around the transport cycle, thereby effluxing ³²P (Fig. 7A, right panel, boxed). If P_i were the first substrate to interact with the empty carrier, in the absence of external Na⁺, we would predict that the addition of P_i to the external medium would increase the proportion of transporters in ‘locked’ in state 2 so that the overall efflux rate should decrease. Note that for completion of either the influx or efflux cycles starting arbitrarily at state 1, release of all substrates implies that the carrier returns to state 1, independent of the concentration of substrates on the release side. As we observed no significant difference in efflux whether or not external P_i was present, our data favour a model in which external Na⁺ is required to interact with the carrier before P_i can bind. However, when both Na⁺ and P_i are present, the transporter will favour occupancy of states 4 and 5, thereby exchanging ³²P with cold P_i in the external medium (trans stimulation). This simple scheme nevertheless fails to explain the stimulatory effect of external Na⁺ on ³²P efflux (Fig. 1C). For the ordered model, addition of external Na⁺ in the absence of P_i should result in a trans-inhibition of efflux as a fraction of the transporters will favour occupancy of state 2. The discrepancy between the predicted model behaviour and measured data can be resolved if we assume that an additional Na⁺ ion interacts with the empty carrier with high affinity in the outward oriented conformation (state 1) so that the efflux transport cycle proceeds at a faster rate, as observed experimentally. Evidence for this hypothesis was obtained from the fluorometry data and the properties of the electrogenic isoforms as follows.

We previously applied the fluorometric technique under voltage clamp conditions to the electrogenic NaPi-IIb (Virkki et al. 2006b). In that study, we obtained evidence that two Na⁺ ions interact cooperatively with the empty carrier prior to P_i binding and that one of these interactions is most likely electroneutral (Virkki et al. 2006a). This led to the proposal of a binding sequence for NaPi-IIa/b of Na⁺–Na⁺–P_i–Na⁺ that contrasted with our previously proposed Na⁺–P_i–Na⁺–Na⁺ scheme (Forster et al. 1998). For the electroneutral NaPi-IIc with a 2 : 1 Na⁺ : P_i cotransport stoichiometry, it follows that in accordance with the Na⁺–P_i–Na⁺ binding order (Fig. 7A), we would predict that only one Na⁺ ion should interact before P_i binding and the fluorescence quenching should show a hyperbolic dependence on external [Na⁺].

Like the electrogenic NaPi-IIb (Virkki et al. 2006b), we could label the equivalent site in NaPi-IIc (Cys-437) with a fluorophore and record changes in fluorescence in response to changes in the external substrates. In the absence of external P_i, increasing the external [Na⁺] quenched the fluorescence as we have previously documented for the labelled Cys-448 in NaPi-IIb (Virkki et al. 2006b), but for the electroneutral S437C mutant, this effect was independent of the membrane potential. The Na⁺-dependent fluorescence quenching in the absence of P_i corroborates the efflux studies on the WT NaPi-IIc by showing that under these conditions Na⁺ ions can induce conformational changes of the protein that are reported by the fluorophore linked to Cys-437. The fluorescence data therefore provide independent support for a kinetic scheme in which Na⁺ is the first substrate to interact with the empty carrier (Fig. 7A). Unexpectedly, we found that the change in fluorescence induced by changes in external [Na⁺] reflected a cooperative interaction of >1 Na⁺ ion. The Hill coefficient of ∼2 therefore suggests that at least two Na⁺ ions interact with both electroneutral and electrogenic SLC34 proteins in a similar cooperative manner.

To account for this behaviour in our model, we therefore postulate that in the forward (clockwise) cycle of NaPi-IIc, an additional Na⁺ ion first binds to the empty carrier state 1 (partial reaction 1–1′, Fig. 7A) with high affinity from the external solution. This ion remains bound throughout the ‘primed’ cotransport cycle (1′–2′–3′–4′–5′–6′–1′), thereby acting as a catalytic activator for cotransport (Turner, 1983); however it does not contribute to overall stoichiometrically coupled cotransport. Its presence can then explain the stimulatory effect that external Na⁺ ions have on ³²P efflux (Fig. 1C). In the absence of external Na⁺, with the catalytic site unoccupied, efflux takes place via the less favourable ‘unprimed’ efflux cycle that involves a reorientation of the empty carrier (1→6, Fig. 7A). In contrast, in the presence of external Na⁺, the catalytic Na⁺ binding site is now occupied and we predict that the carrier will then efflux at a higher rate, i.e via the ‘primed’ cycle that now involves a reorientation of the ‘primed’ empty carrier (1′–6′) after substrate release. The identity of the partial reaction(s) in the ‘unprimed’ cycle altered by the presence of this Na⁺ ion remains to be determined. As indicated in Fig. 7A, the binding at the catalytic site by a Na⁺ ion from the internal side (partial reaction 6→6′) is assumed to be unfavourable.

Thus, it is only in the electrogenic isoforms that the first Na⁺ ion binds to a site within the transmembrane field, as evidenced by Na⁺-dependent pre-steady-state relaxations (e.g. Forster et al. 2002). This Na⁺ ion is subsequently translocated – either in the absence of P_i (uncoupled leak mode) (Forster et al. 1998; Andrini et al. 2008) or together with P_i and two additional Na⁺ ions (cotransport mode), as depicted in Fig. 7B. In both cases the translocation within the protein is an electroneutral event. Therefore, in terms of substrate interactions, the electrogenic isoforms (e.g. Virkki et al. 2006b) and the electroneutral NaPi-IIc differ only with respect to the mobility of the first Na⁺ ion. So far, structure–function studies on SLC34 proteins have revealed one key structural difference between the electrogenic and electroneutral isoforms: the absence of a charged aspartic acid at the equivalent site in NaPi-IIc (Gly-195) (Fig. 2A): removal of this charge in NaPi-IIa results in electroneutral cotransport (Virkki et al. 2005a), whereas introduction of this charge in NaPi-IIc restores electrogenic behaviour, including PFA-sensitive leak, presteady-state relaxations and 3 : 1 Na⁺ : P_i stoichiometry (Bacconi et al. 2005).

Although the fluorescence data were obtained from a protein that has been modified and cannot complete the full cotransport cycle, our finding that the apparent obtained from fluorescence measurements was close to that obtained using uptake assays with the whole transport cycle (Table 1) strongly suggests that the partial reactions that determine this phenomenological parameter are unchanged in the labelled S437C. Moreover, we can exclude translocation of the fully loaded carrier as a kinetic determinant of . On the other hand, the apparent affinity for Na⁺ interactions () determined by fluorescence was significantly larger than that determined using flux assays (Table 1). As obtained from fluorescence only relates to the first two Na⁺ interactions, the higher apparent for partial reactions suggests that the final Na⁺ interaction is a strong determinant of the overall apparent . Consistent with our previous studies on electrogenic NaPi-IIa/b in which Cys substitution was made at the equivalent site in NaPi-IIa (Ser-460) (Lambert et al. 1999; Kohler et al. 2002) and NaPi-IIb (Ser-448) (Virkki et al. 2006b), this finding would also suggest that the modified Cys-437 in NaPi-IIc limits conformational changes to those associated with the same partial reactions. However, we are unable to determine if the last Na⁺ interaction (3⇌4) can occur and the Cys modification blocks the final translocation of the fully loaded carrier (4⇌5) (Fig. 7A). The fluorometric measurements on S437C also establish that like P_i, the interaction of the inhibitor PFA with the carrier requires Na⁺. As the increase in fluorescence relative to ND100 alone by both substrates was very similar, this would be consistent with its established competitive inhibitory properties (Szczepanska-Konkel et al. 1986) and it most likely interacts directly at the P_i binding site. Finally, in contrast to Na⁺, Li⁺ ions do not drive P_i cotransport, yet we observe a change in F when [Li⁺] is changed that was qualitatively similar that reported for the electrogenic NaPi-IIb isoform (Virkki et al. 2006b). The lack of a significant change in uptake rate at 50 mm Na⁺, when Li⁺ was substituted for choline, may indicate that there is no direct interaction between these cations in NaPi-IIc.

Conclusions

Our findings suggest that electrogenic and electroneutral SLC34 isoforms probably share a common functional architecture, which includes substrate coordination sites and translocation pathway. These structural features are reflected in the identical transport characteristics (similar , , pH dependence) for these isoforms. They differ only with respect to the role played by the catalytically activating first Na⁺ interaction, which in the case of the electrogenic isoforms will serve to increase the electrochemical driving force for cotransport and the concentrating capacity of the carrier. This study also underscores the notion that transport stoichiometry and substrate binding stoichiometry may not be equivalent among members of the same transporter family and this may be a feature common to coupled transport systems (e.g. Rudnick, 2006; Shi et al. 2008).

Glossary

Abbreviations

BBMV

brush border membrane vesicle

MTS

methanethiosulfonate

MTSEA

2-aminoethyl methanethiosulfonate hydrobromide

MTSET

2-(trimethylammonium)ethyl methanethiosulfonate bromide

MTS-TAMRA

2-((5(6)-tetramethylrhodamine)carboxylamino)ethyl methanethiosulfonate

NaPi-IIc

type IIc Na⁺-coupled P_i cotransporter

PFA

phosphonoformic acid

SLC34

solute carrier family 34 (http://www.genenames.org)

Author contributions

Experimental work: C.G., I.C.F; conception and design, analysis and interpretation of data: C.G., H.M., I.C.F.; drafting and revision of manuscript: C.G., I.C.F.; approval of final manuscript: C.G., H.M., I.C.F. Experiments were performed in the Institute of Physiology at the University of Zurich, Zurich, Switzerland.