The molecular mechanism of SLC34 proteins: insights from two decades of transport assays and structure-function studies

Abstract

The expression cloning some 25 years ago of the first member of SLC34 solute carrier family, the renal sodium-coupled inorganic phosphate cotransporter (NaPi-IIa) from rat and human tissue, heralded a new era of research into renal phosphate handling by focussing on the carrier proteins that mediate phosphate transport. The cloning of NaPi-IIa was followed by that of the intestinal NaPi-IIb and renal NaPi-IIc isoforms. These three proteins constitute the main secondary-active Na⁺-driven pathways for apical entry of inorganic phosphate (P_i) across renal and intestinal epithelial, as well as other epithelial-like organs. The key role these proteins play in mammalian P_i homeostasis was revealed in the intervening decades by numerous in vitro and animal studies, including the development of knockout animals for each gene and the detection of naturally occurring mutations that can lead to P_i-handling dysfunction in humans. In addition to characterising their physiological regulation, research has also focused on understanding the underlying transport mechanism and identifying structure-function relationships. Over the past two decades, this research effort has used real-time electrophysiological and fluorometric assays together with novel computational biology strategies to develop a detailed, but still incomplete, understanding of the transport mechanism of SLC34 proteins at the molecular level. This review will focus on how our present understanding of their molecular mechanism has evolved in this period by highlighting the key experimental findings.

Appendix: Tools and protocols used to characterise electrogenic SLC34 proteins expressed in Xenopus oocytes

• 1. Steady-state and presteady-state assays-characterising electrogenic behaviour at the macroscopic level

Voltage steps applied to a voltage clamped whole Xenopus oocyte expressing electrogenic cotransporters like NaPi-IIa/b reveal two components of membrane current: a presteady state, transient relaxing component and a steady-state component (Fig. 9, left data set). Analysis of both components yields important kinetic information about the transport mechanism. The currents recorded from a whole oocyte are macroscopic and represent the mean electrogenic behaviour of a large (typically ≥ 10¹⁰) population of transporters. Unlike ion channels, whose activity can be resolved at the single molecule level, the slow rate of charge translocation (~ fA or less) of membrane transporters is several orders of magnitude below that which can be presently resolved. However, like ion channels, we assume individual transporters behave independently and the population behaviour can be described in terms of the probability of occupying a particular conformational state. This depends on the membrane potential, substrate availability and the partial reaction rates associated with entering and leaving that state, according to the kinetic scheme (Fig. 1b).

Fig. 9

Two electrode voltage clamp of whole Xenopus oocytes expressing electrogenic NaPi-IIa/b. The voltage step protocol (shown here for 20 mV steps in range − 160 to + 80 mV) is used to characterise the steady-state and presteady-state electrogenic properties of oocytes. Voltage steps evoke a capacitative component (presteady-state) comprising the linear charging/discharging of the endogenous membrane phospholipid and the non-linear transporter-related charge relaxations. In the presence of P_i (right), the steady-state holding current increases and the exogenous charge relaxations are suppressed

The steady-state current component comprises endogenous oocyte currents and, in the absence of external P_i, the constitutive leak of the transporter. In the presence of P_i (Fig. 9, right data set) the downward deflection (arrowed) of the steady-state current indicates inward movement of charge accompanying active cotransport. Subtraction of these data sets from one another eliminates the endogenous component and yields the P_i-dependent current (I_Pi) with a small error due to the constitutive leak that we assume is suppressed by P_i. This component can be independently assayed using the inhibitor PFA (e.g. Fig. 6a). Ignoring the leak can potentially lead to an underestimate of the cotransport activity as well as charge translocation in stoichiometry assays depending on its magnitude (e.g. [2] and see Fig. 6 in [25]). By repeating these measurements with different substrate concentrations, standard phenomenological parameters such as apparent substrate affinity and maximum transport rate are derived as a function of membrane voltage. An important caveat for all oocyte-derived kinetic data is that these parameters are generally obtained at temperatures in the range 18-20 °C and extrapolation to mammalian physiology conditions necessitates taking account of temperature.

The presteady-state component comprises the endogenous charging transient of the oocyte lipid bilayer capacitance upon which the transporter-associated presteady-state current relaxations are superimposed (Fig. 9). The former is a linear function of the test voltage and can be eliminated by procedures such as curve fitting, subtracting matching records in the presence of a blocker or P/n protocols used to resolve ion channel gating currents (e.g. [4]). Analysis of the transporter-associated relaxations focuses on the charge displaced (Q) obtained by numerical integration, and the relaxation time constant (τ) obtained by exponential fitting. Both are important for characterising transporter kinetics. The fidelity and validity of presteady-state data can be tested by confirming charge balance for voltage steps to and from the holding potential, verifying independence of total charge movement from the holding potential and voltage independence of τ for the relaxations in response to the return step to the holding potential and establishing a direct correlation between steady-state transport and the total charge displaced.

The Q-V data show a characteristic sigmoidal relationship with saturation at either or both potential extremes due to the fixed number of charges involved. This is usually fit with a single Boltzmann-type function (Eq. 1) to obtain three parameters: the effective charge (z) per transporter (from the slope), the total charge displaced (Q_max) and the midpoint potential (V_0.5), an equilibrium potential at which half the charge has been displaced. This procedure is valid for a two state model, and with more than one partial reaction contributing to the charge movement, the single Boltzmann function fit is clearly an approximation (e.g. [62]). Despite this limitation, these parameters can be used to characterise and interpret changes in substrate interactions and voltage dependence. The Q-V data are sensitive to external [Na⁺] (Fig. 10) and this relationship is a key biophysical signature for understanding cation interactions. As external [Na⁺] increases, V_0.5 shifts towards depolarizing potentials (arrow), consistent with Na⁺ ions binding more easily according to their availability. The apparent charge associated with the empty carrier and the Na⁺ ion interactions can be estimated from z for each data set. The invariance of Q_max with changing [Na⁺] is expected for a given number of transporters, each of which binds a fixed number of Na⁺ ions. At low [Na⁺] this is less obvious because of limitations of the fitting procedure and at 0 mM Na⁺, charge movement is due to the empty carrier alone. The growth in z with [Na⁺] reflects the increased contribution of Na⁺ interactions to the effective charge movement contributed by each transporter. The limiting slope of V_0.5 vs log₁₀ [Na⁺] (~ 120 mV/decade) is also consistent with a sequential interaction of 2 Na⁺ ions per transporter. These findings are also predicted by deriving an analytical expression for Q-V, and numerically solving for V_0.5 as a function of [Na⁺] (see Eq. 2–4) (e.g. [3, 71, 78]. Estimates of the dissociation constants for each cation interaction can be made and compared with the effect of mutations at the predicted cation binding sites (e.g. [29, 71].

Fig. 10

Analysis of presteady-state relaxations. Transporter-related presteady-state relaxations (a) can be integrated to give the charge movement (Q-V) (b) and fit with a single exponential to give the main relaxation time constant (τ-V) (c). As external [Na⁺] is increased, the midpoint of Q-V and peak of τ-V shift as shown. Fitting the Q-V data with a Boltzmann function (Eq. 1) yields the parameters Q_max, z and V_0.5 which are plotted as functions of [Na⁺] (d). The limiting slope of the relationship between V_0.5 and log₁₀[Na] (d, inset) indicates that 2 Na⁺ ions interact. Examples from representative oocytes expressing flounder NaPi-IIb

In addition, the ratio Q_max/ze (where e is the electronic charge) can also be used to predict (i) the number of transporters (N_t) contributing to the charge movement (and by implication the cotransport activity) (Eq. 5) and, (ii) the turnover rate (R_t) when combined with the steady state cotransport current (e.g. [41] (Eq. 6). However, these should be taken as estimates only, given the inherent assumptions in the method (see [47, 114]).

The dependence of the relaxation time constant on voltage (τ-V) shows a “bell-shaped” form typical for non-linear charge movements associated with membrane proteins (Fig. 10). For a two state system (e.g. the empty carrier transition 0↔1, Fig. 1b), the forward and backward rate constants can be predicted from fits to the τ-V data. For a 3 state model (empty carrier and a Na⁺ binding step) theory predicts two time constants and analytical expressions become more complex. In practice, it may be difficult to resolve more than one relaxations component due to limitations of the curve fitting procedure and bandwidth of the voltage clamp. A single exponential fit to experimental data shows that like the shift of V_0.5, the peak time constant shifts to more depolarizing potentials as [Na⁺] increases (arrow), consistent with having the empty carrier and at least one Na-dependent partial reaction contributing to the charge movement.

• 2. Voltage clamp fluorometry

This technique is readily applied to whole oocytes by combining the two electrode voltage clamp with a basic fluorescence microscope (e.g. [17, 91]) (Fig. 11). The instrumentation used in Zurich to study SLC34 proteins was modified from a design by the Wright group (UCLA) [69]. It has allowed real-time measurement of electrogenic activity (steady-state and presteady-state) and simultaneous fluorescence emissions from oocytes previously labelled with a fluorophore and covalently linked to an engineered cysteine in the transporter. The same voltage step protocols used for steady-state and presteady-state analysis can be used for VCF studies although more signal averaging may be required to obtain an acceptable signal to noise ratio. As all cysteines exposed to the labelling medium will be potentially labelled, an endogenous component will contribute to the background fluorescence. Generally, this is found not vary with membrane potential, but can still potentially compromise the signal resolution if ΔF from the transporter is small, by limiting the useful dynamic range. To overcome this, electronic offset can be applied before data acquisition. Pre-labelling endogenous cysteines and other procedures can also help reduce the background signal [91]. Precautions must also be taken to take account of loss of fluorescence signal during the course of an experiment due to bleaching, washout of fluorophore or internalisation of proteins (e.g. [45, 108]). Like the presteady-state assays, the fluorescence signal is a mean of emissions arising from a population of labelled transporters that at any instant can reside in different conformational states, depending on the probability of state occupancy. For example, if the fluorophore experiences a more polar or charged environment, the fluorescence is quenched and this can then be related to state occupancy. Collisional quenching is commonly assumed to be the underlying mechanism leading to ΔF and has been observed experimentally (e.g. [88]) but other mechanisms should be considered (e.g. [16]). The ΔF-V data are usually fit with a Boltzmann function although a direct correspondence with the Q-V parameters is not always obtained because there may not necessarily be a correlation between charge movement and the change in microenvironment of the fluorophore, depending on the labelling site. Unlike the electrophysiological assays, changes in fluorescence can also be reported by electroneutral processes in response to changing substrate [45].

Fig. 11

Whole oocyte voltage clamp fluorometry. The basic hardware arrangement for whole Xenopus oocyte recording combines a two-electrode voltage clamp and basic fluorescent microscope. Optimal fluorescence signal is obtained with the cell inverted from the normal arrangement in Fig. 9. Filter cube components can be changed for optimal response depending on the fluorophore. Typical simultaneous recordings of I_m and ?F are shown (right) for an oocyte labelled with MTS-TAMRA. The bandwidth for each signal was 500 Hz

• 3. Equations used in presteady-state analysis:

  • a) Boltzmann Q-V relationship

    $$ Q={Q}_{hyp}+\frac{Q_{max}}{1+{\mathit{\exp}}^{\frac{ze\left({V}_{0.5}-V\right)}{kT}}} $$

    (1)

    where V_0.5 is the midpoint voltage, z is the apparent valency/protein, Q_max is the total charge available to move, Q_hyp is the charge at the hyperpolarizing limit and is a function of the holding potential and e, k and T have their usual meanings.

  • b) Derivation of Q-V relationship and V_0.5 vs [Na] derived from the four state kinetic model (states 0,1–3, Fig. 1b).

  The total charge displaced for a voltage step from ∞ (i.e. all transporters in state 0) to V and neglecting charge movement due to transition 0↔7, is given by [3]:

  $$ {Q}_{\infty}^V=-{N}_te\left[\frac{\Big(\left({z}_{01}+{z}_{12}+{z}_{23}\right)+\left({z}_{12}+{z}_{23}\right)\alpha +{Z}_{23}\alpha \beta}{1+\alpha +\alpha \beta +\alpha \beta \gamma}\right] $$

  (2)

  where N_t is the number of transporters, α = k₀₁/k₁₀, β = k₁₂/k₂₁ and γ = k₂₃/k₃₂, (i.e. the ratio of forward to backward rate constants for each partial reaction), and z_ij is the apparent valence for the partial reaction that couples states i and j. Using rate theory, the forward and backward rates (k_ij, k_ji), assuming symmetrical barriers, are expressed as ${k}_{ij}={k}_{ij}^0{\mathit{\exp}}^{-e{z}_{ij}V/2 kT}$; ${k}_{ji}={k}_{ji}^0{\mathit{\exp}}^{e{z}_{ij}V/2 kT}$. For partial reactions involving cation interactions, the forward rates are scaled by the cation concentration.

  To link Eq. 2 to the single Boltzmann function formulation of the Q-V data, the voltage at which 50% of the total charge has been displaced (V_0.5) is found by setting ${Q}_{\infty}^V=0.5{N}_te\left({z}_{12}+{z}_{23}+{z}_{23}\right)$, to give:

  $$ \left({z}_{01}+{z}_{12}+{z}_{23}\right)\left(1-0.5\left(1+\alpha +\alpha \beta +\alpha \beta \gamma \right)\right)+\alpha \left({z}_{12}+{z}_{23}\right)+\alpha \beta {z}_{23}=0 $$

  (3)

  Substituting for α, β, γ, for [Na] large this reduces to:

  $$ {\alpha}^0{\beta}^0{\upgamma}^0\ {\left[ Na\right]}^2\ \left({\exp}^{-\mathrm{e}{V}_{0.5}\left({z}_{01}+{z}_{12}+{z}_{23}\right)/ kT}\right)=1 $$

  (4)

  where α⁰ = k⁰₀₁/k⁰₁₀, β⁰ = k⁰₁₂/k⁰₂₁ and γ⁰ = k⁰₂₃/k⁰₃₂, and the superscript 0 refers to values at V = 0. A plot of V_0.5 vs log_e[Na] yields a limiting slope of 2kT/[e(z₀₁ + z₁₂ + z₂₃)] mV/e-fold change in [Na] (or ≈ 116/(z₀₁ + z₁₂ + z₂₃) mV/10-fold change in [Na]) at 20 °C; i.e. for one net charge translocated across the transmembrane field, the predicted slope is then 116 mV/10-fold change in [Na] [3] as observed experimentally (see Fig. 5e).

  • c) Estimation of number of transporters (N_t) and turnover rate (R_t)

    $$ {N}_{\mathrm{t}}={Q}_{\mathrm{max}}/\mathrm{ze} $$

    (5)
    $$ {R}_{\mathrm{t}}={I_{\mathrm{Pi}}}^{\mathrm{max}}/\left({N}_{\mathrm{t}}{\mathrm{z}}_{\mathrm{t}}\mathrm{e}\right) $$

    (6)

    where I_Pi^max is the maximum transport current and z_t is the charge translocated/cycle. The estimates of R_t require specification of the experimental conditions ([Na⁺], V_m etc.) and are dependent on the assumptions related to estimating N_t from a single Boltzmann fit [114].