The Human Sodium-Glucose Cotransporter (hSGLT1) Is a Disulfide-Bridged Homodimer with a Re-Entrant C-Terminal Loop

Abstract

Na-coupled cotransporters are proteins that use the trans-membrane electrochemical gradient of Na to activate the transport of a second solute. The sodium-glucose cotransporter 1 (SGLT1) constitutes a well-studied prototype of this transport mechanism but essential molecular characteristics, namely its quaternary structure and the exact arrangement of the C-terminal transmembrane segments, are still debated. After expression in Xenopus oocytes, human SGLT1 molecules (hSGLT1) were labelled on an externally accessible cysteine residue with a thiol-reactive fluorophore (tetramethylrhodamine-C5-maleimide, TMR). Addition of dipicrylamine (DPA, a negatively-charged amphiphatic fluorescence "quencher") to the fluorescently-labelled oocytes is used to quench the fluorescence originating from hSGLT1 in a voltage-dependent manner. Using this arrangement with a cysteine residue introduced at position 624 in the loop between transmembrane segments 12 and 13, the voltage-dependent fluorescence signal clearly indicated that this portion of the 12-13 loop is located on the external side of the membrane. As the 12-13 loop begins on the intracellular side of the membrane, this suggests that the 12-13 loop is re-entrant. Using fluorescence resonance energy transfer (FRET), we observed that different hSGLT1 molecules are within molecular distances from each other suggesting a multimeric complex arrangement. In agreement with this conclusion, a western blot analysis showed that hSGLT1 migrates as either a monomer or a dimer in reducing and non-reducing conditions, respectively. A systematic mutational study of endogenous cysteine residues in hSGLT1 showed that a disulfide bridge is formed between the C355 residues of two neighbouring hSGLT1 molecules. It is concluded that, 1) hSGLT1 is expressed as a disulfide bridged homodimer via C355 and that 2) a portion of the intracellular 12-13 loop is re-entrant and readily accessible from the extracellular milieu.

Fig 1. Principle of hVoS FRET quenching.

Dipicrylamine (DPA, grey hexagon) distribution on either side of the membrane is controlled via the membrane potential (V_m). Changes in V_m (ΔV_m) thus reduce/increase the mean distance between DPA and fluorescent probe (Alexa 488, AL488, as green hexagon and Tetramethylrhodamine, TMR, as orange hexagon). Two possibilities are considered. A) If only monomers are present, FRET between AL488 and TMR does not occur. Thus, only DPA quenching of AL488 (using the donor filter set) and TMR (using the acceptor filter set) is observed, but no fluorescence signal is detected using the FRET filter set. B) If multimeric states are present, FRET occurs between AL488 and TMR. A voltage-dependent fluorescent signal can be seen using the 3 filter sets.

Fig 2. hVoS FRET quenching results.

Typical fluorescence traces for oocytes expressing either wt hSGLT1 (panel A through C) or the hSGLT1 mutant C255A (panel D through I) which is known from a previous study [3] to present a cystein residue (C511) that is freely accessible for labelling from the external solution. Oocytes are simultaneously labeled with Alexa 488 (AL488, the donor) and Tetramethylrhodamine (TMR, the acceptor). In the absence of dipicrylamine (DPA, a negatively-charged amphiphatic fluorescence quencher), changing the membrane potential from -50 mV to +50 mV (black traces) or to -150 mV (red traces) produces very little change in the fluorescence intensities measured for C255A expressing oocytes using either one of the 3 filter sets (panel D to F). In the presence of DPA, oocytes expressing C255A present strong voltage-dependent fluorescent signals (panel G to I). This effect is specific for the fluorophores attached to C511 as wtSGLT1 (panel A to C) display much weaker voltage-dependent fluorescent signal. Panels J to L compare the mean voltage-dependent signal $\frac{F_{+50mV}-F_{-150mV}}{F_{-50mV}}$ (mean ± SE) using the 3 filter sets for oocytes expressing wt SGLT1 (n = 5) or the mutant C255A (n = 15 to 17). Stars (*) denote statistical significance (C255A vs wt, p<0.05).

Fig 3. Control experiments performed in the absence of either the donor or the acceptor fluorophore.

Typical fluorescence traces for oocytes expressing mutant C255A cotransporters in the presence of dipicrylamine (DPA) using 3 filter sets (Donor, Acceptor and FRET). When oocytes are only labelled with AL488 (panel A-C), membrane voltage pulses from -50 mV to +50 mV (black traces) and from -50 mV to -150 mV (red traces) produces a large change in the AL488 fluorescence signal which can be somewhat observed using the FRET filter set. The signal recorded with the FRET filter set (F_+50mV − F_−150mV) averaged only 3.8 ±0.9% (mean ±SE, n = 6) of the signal directly measured with the Donor filter set. When C255A expressing oocytes are only exposed to the TMR (the acceptor) (panel D-F), the ratio between the FRET and the acceptor voltage-dependent fluorescence signals is 0.8 ±0.2% (mean ±SE, n = 5).

Fig 4. Western blot of wt and mutant myc—tagged hSGLT1.

A) Effect of absence/presence of β-Mercaptoethanol (β-ME) on the migration profiles of the wt, the single cysteine mutants C255A and C511A and of the double cysteine mutant C255/511A. In the absence of β-ME, most of the cotransporters migrate at a level corresponding to a dimeric configuration. B) Absence of effect of β-ME on the migration profile of Aquaporin 2 (AQP2). C) Single cysteine mutants in the absence of β-ME. A band corresponding to a putative dimer is present for the wt and all tested single mutant, except C355A. D) Multiple mutants in absence of β-ME. A band corresponding to the putative dimer is present for 2xCys (C351/361A), and 3xCys (C345/351/361A), but not for 4xCys (C345/351/355/361A). E) β-ME has no effect on the migration profile for the mutants C355A and 4xCys. For all panels presented, experiments have been repeated 3 times and representative examples are shown.

Fig 5. hVoS quenching on E624C.

A) Typical fluorescence trace from wt hSGLT1 expressing oocytes labeled with Tetramethylrhodamine (TMR) in the presence of dipicrylamine (DPA). DPA is displaced by changing the membrane potential from -50 mV to +50 mV (black traces) and from -50 mV to -150 mV (red traces). B) Typical fluorescence trace from E624C expressing oocytes, labeled with TMR in the presence of DPA. C) Average voltage-sensitive fluorescence signal for wt hSGLT1 (mean±SE, n = 9) and for the mutant E624C (n = 8).*** denote statistical significance (p<0.001).

Fig 6. Expected location of C355 and comparison of the 12–13 loops from vSGLT and hSGLT1.

A sequence alignment between vSGLT and hSGLT [12] showed that the position of C355 in hSGLT1 corresponds to N328 in vSGLT. Using vSGLT crystal structure (PDB# 3DH4) as a template, C355 would be located between TMS7 and 8, more precisely between extracellular helix 7a and 7b that are colored in yellow in the present figure (LeuT numbering of TMS’s). A top view shows that C355 would be located 10Ǻ away from the closest membrane-cotransporter interface. The figure also shows the difference in the loop linking TMS12 (in pink) and TMS13 (in purple). The 12–13 loop of vSGLT (in magenta) is 24 amino acids long and the corresponding loop in hSGLT1 (in blue) is 90 amino acids long. The position of E624 in the hSGLT1 12–13 loop is indicated. When mutated to a cysteine, that position can be labelled from the extracellular solution and hVoS FRET quenching suggested that this position should be located clearly above the membrane center.