Active sugar transport in health and disease


Ernest M. Wright and Bruce A Hirayama, Department of Physiology, The David Geffen School of Medicine at UCLA, Los Angeles CA 90095-1751, USA.
(fax: 310 206 5661; e-mails: and


Secondary active glucose transport occurs by at least four members of the SLC5 gene family. This review considers the structure and function of two premier members, SGLT1 and SGLT2, and their role in intestinal glucose absorption and renal glucose reabsorption. Genetics disorders of SGLTs include Glucose-Galactose Malabsorption, and Familial Renal Glucosuria. SGLT1 plays a central role in Oral Rehydration Therapy used so effectively to treat secretory diarrhoea such as cholera. Increasing attention is being focused on SGLTs as drug targets for the therapy of diabetes.


Glucose is the fuel that provides energy for normal activity in humans and the major source is the carbohydrate in food. The amount of glucose stored in the body is ∼450 and ∼250 g turns over daily. The brain alone consumes ∼125 g day−1. The normal Western diet provides ∼180 g day−1 and the gap between the amount in food and the amount utilized is bridged by stored glucose and gluconeogenesis mainly in the liver and kidneys. Homeostatic mechanisms maintain blood glucose within narrow limits, 4 and 12 mm, to ensure that an adequate amount of energy is provided first to the brain and then to the other organs of the body. Glucose in food is usually in the form of complex carbohydrates, which are broken down and absorbed completely in the small intestine. Glucose is not an essential food as people can thrive on a carbohydrate free diet. In these individuals, gluconeogenesis is adequate to maintain their blood glucose at essentially normal levels thereby maintaining the supply of glucose to the brain and the rest of the body. The kidneys assist in keeping blood glucose levels normal by reabsorbing the ∼180 g filtered each day at the glomeruli. Virtually no glucose appears in the urine (<0.5 g day−1) because of presence of glucose transporters in the proximal tubule.

Glucose transport

Transporters in two gene families are responsible for the absorption of glucose across the small intestine, the reabsorption of glucose from the glomerular filtrate, brain uptake across the blood–brain barrier, and the uptake and release of glucose from all cells in the body. These are the facilitated glucose transporters, the GLUT or SLC2 gene family [1], and the sodium-coupled glucose cotransporters, the SGLT or SLC5 gene family [2]. The GLUTs are responsible for the downhill, passive transport of glucose across cell membranes, i.e. these transporters speed up or facilitate the equilibration of the sugar across a membrane. Prime examples include GLUT1 involved in the transport of glucose across the endothelial cells of the blood–brain barrier, and GLUT4 responsible for insulin-stimulated glucose uptake into skeletal muscle. In the case of the SGLTs, the most well-known member is SGLT1, which is responsible for the ‘active’ transport of glucose across the brush border membrane of the small intestine.

Figure 1 illustrates how the dual expression of SGLT1 and GLUT2 in enterocytes accounts for the complete absorption of glucose from food in the small intestine. SGLT1 is expressed in the brush border membrane of the enterocyte, where it couples the transport of two sodium ions and one glucose molecule across the brush border membrane. The energy stored in the sodium electrochemical potential gradient across the brush border membrane provides the energy to drive glucose accumulation in the enterocyte against its concentration gradient. Sodium that enters the cell along with glucose is then transported out into blood by the Na/K-pump in the basolateral membrane, thereby maintaining the driving force for glucose transport. As the sugar accumulates within enterocytes, this sets up a driving force to transport glucose out of the cells into the blood via GLUT2 expressed in basolateral membranes. A fraction of the intracellular glucose appears to be taken up into endosomes, as glucose-6-phosphate, and delivered to the blood by exocytosis through the basolateral membrane [1]. The net result is that 1 mole of glucose and two moles of sodium are transported across the enterocyte from gut lumen into blood, and this is followed by two moles of anions to ensure electroneutrality, and water. The energy for the overall process comes from the ATP consumed by the basolateral Na/K-pump.

Figure 1.

 A model for glucose (and galactose) absorption across the small intestine. Shown is a cartoon of a mature enterocyte on the upper villus of the small intestine. Glucose (and galactose) is transported across the brush border membrane by Na-cotransport (SGLT1), and the Na is then transported out across the basolateral membrane by the Na/K-pump. Glucose (and galactose) accumulates within the cell and then diffuses out into blood across the basolateral membrane through GLUT2. Some glucose is phosphorylated within the cell and accumulated in endosomes before dephosphorylation and exocytosis into blood. Modified from Ref. [8].

Na and sugar cotransport by SGLT1 is referred to as secondary active transport because the driving forces – Na gradients – are maintained by the primary active Na/K-pump, or Na/K-ATPase. Quite simply, the direction and rate of glucose transport by SGLT1 are a function of the direction and magnitude of the Na-gradients across the plasma membrane. In normal cells, it is the Na/K-pump that sets the direction and magnitude of the sodium gradients. This review will focus on the structure and function of two major secondary active glucose transporters, SGLT1 and SGLT2, in health and disease.

SLC5 genes

Although there are well over 230 members of the Sodium Substrate Symporter gene Family (SSSF), we have only identified 11 in the SLC5 human gene subfamily [2]. These include proteins that transport sugars (SLC5A1, 2, 9 and 11), myo-inositols (SLC5A3 and 11), anions (SLC5A5 and SLC5A8), vitamins (SLC5A6) and choline (SLC5A7), one orphan (SLC5A10), and one glucose sensor (SLC5A4). Table 1 lists the SLC5 sodium/glucose cotransporters along with their substrates, apparent affinities, and expression profiles. Although SGLT1 is prominently expressed in the small intestine, the gene is also transcribed in the renal proximal tubule and other organs such as the brain and heart. The natural substrates are glucose and galactose with apparent affinities (K0.5) of 0.5 mm. Mutations in the SGLT1 gene causes malabsorption of glucose and galactose (see below). SGLT2 is also expressed in various tissues throughout the body including the renal proximal tubule, where it appears to play a major role in the reabsorption of glucose from the glomerular filtrate. Functionally, this transporter is different from SGLT1 in that it has a lower affinity for glucose and it does not transport galactose. Transporters favouring other substrates may also be important in glucose metabolism. SGLT4 appears to be a low affinity transporter for mannose and glucose and the gene is found in a variety of tissues including the pancreas. SGLT5, an orphan with an unknown function, is almost exclusively expressed in the renal cortex. Although SGLT6 (SMIT2) is a high-affinity myo-inositol transporter with a K0.5 close to the plasma concentration, 120 μm, it also behaves as a low-affinity glucose transporter. Given the normal plasma concentrations of myo-inositol and glucose and the finding that both substrates have the same maximum transport rate, it is clear that SGLT6 should play a significant role in glucose uptake in those cells in the brain, kidney and intestine where this gene is expressed. Finally, it should be noted that SGLT3, which is expressed in the enteric nervous system and at the neuromuscular junction, is not a glucose transporter but instead a glucosensor [3].

Table 1. SLC5 genes responsible for Na/glucose cotransport
TransporterSubstrateK0.5 mmDistribution
  1. Substrate specificity, affinity (K0.5 for αMDG) and RNA expression of SGLT (SLC5) genes. Substrate specificity and αMDG transport were measured by using heterologous expression systems [29, 53–55 and M. Coady pers. comm.]. RNA distribution is based on RNase protection assays (M. Bing, M.G. Martin and E.M. Wright, unpubl. data) and Northern blots (SMIT1 and SMIT2) [2].

SGLT1 (SLC5A1)Glucose, galactose0.5Intestine, trachea, kidney, heart, brain, testis and prostate
SGLT2 (SLC5A2)Glucose
Not transported
Kidney, brain, liver, thyroid, muscle and heart
SGLT4 (SLC5A9)Glucose and mannose2.4Intestine, kidney liver, brain, lung, trachea, uterus and pancreas
SGLT5 (SLC5A10)Not knownNot knownKidney
SGLT6 (SMIT2,SLC5A11)Myo-Inositol, glucose0.12
Brain, kidney, intestine
SMIT1 (SLC5A3)Myo-inositol, glucose0.055
Brain, heart, kidney and lung

As of this time, we do not have an atomic structure for SGLT proteins, but we have developed some structural insights. The SLC5 genes code for ∼75 kDa proteins with >59% amino acid identity. A secondary structure model for SGLT1 is shown in Fig. 2. This model, based on both experiments and computer modelling [2–6], shows a polytopic membrane protein with the N- and C-termini facing the extracellular side of the plasma membrane. Highlighted on the model are the structural motifs for the large SSSF family and the smaller eukaryote SGLT and SMIT family members. Studies with SGLT chimeras and truncated SGLT1 proteins have led to the identification of the sugar-binding/transport domain in the C-terminal portion of the protein. The truncated protein, containing transmembrane helices 10–14, facilitates downhill, Na-independent sugar transport in both the Xenopus laevis oocyte expression and in proteoliposomes [7].

Figure 2.

 A secondary structure model for SGLT1. The polytopic protein has 14 transmembrane helices with extracellular N- and C- termini. Shown are the glycosylation site, the SGLT1/SMIT and SSSF motifs, and the two putative glucose binding sites in the C-terminal domain. Modified from Ref. [8].

Detailed analysis of the sequence alignments of 30 members of the mammalian SLC5 gene family has been performed: including 9 SGLT1s, 3 SGLT2s, 3 SGLT3s, 2 SGLT4s, 2 SGLT6s, 4 sodium/myoinositol cotransporters (SMITs), 4 sodium/iodide cotransporters (NISs), 2 sodium/multivitamin cotransporters and 1 choline transporter (CH1) [8]. This analysis, together with a survey of the structure of 60 known sugar-binding proteins [9] and the structure of the one known sugar transporter, lactose permease [10], provides insights into the structure of the glucose binding and translocation domain of SGLT1. The predominant features of sugar-binding proteins are that (i) an extensive network of hydrogen bonds exists between the polar residues of the sugar and polar side-chains of residues forming the binding site, and (ii) additional interaction energy is frequently provided by stacking of an aromatic residue with the apolar face of the pyrananose ring [11]. Thirty-one charged and aromatic residues are conserved in transmembrane helices 10–14 of 9 SGLT1s that are neither conserved amongst members of the large gene family, nor the nonsugar transporters of the SLC5 sub-family [2, 8]. There is remarkably little variation in the 31 conserved residues in the C-terminal of SGLT1 and the other SGLTs (SGLT2, 3, 4 and 6). One prominent exception is that Q457 is changed to E457 in the SGLT3s (see below). Based on the distribution of the conserved polar and aromatic residues, we have postulated that there are two sugar-binding sites in the C-terminal of SGLT1, one on the extracellular face and another on the intracellular face of the protein (Fig. 2). Efforts are presently underway to obtain the structure of SGLT sugar-binding sites to complement the structural information we have obtained from functional studies.

Functional studies

We have exploited the power of the X. laevis oocyte expression system to study the mechanism of Na and glucose transport by SGLTs using radioactive tracer, electrophysiological, electron microscopic and optical methods [12–19]. In addition, we have developed biochemical methods to study recombinant SGLT proteins produced in bacteria [6, 20–22]. Much of this work has been reviewed recently [23].

In summary, we have demonstrated that SGLTs are fully functional as monomers [5,24] and that the rate and direction of glucose transport simply depend on the magnitude and direction of the sodium electrochemical potential gradients, i.e. SGLTs function as fully reversible molecular motors [18, 22]. Nevertheless, the transporter is asymmetrical in that under identical driving conditions (i) the sugar affinity (1/K0.5) is orders of magnitude lower for reverse than forward transport, (ii) the sugar selectivity for reverse transport is different from that for forward transport, and (iii) phlorizin, the nontransported competitive inhibitor of Na/sugar cotransport, is much more potent on the external side of the membrane than on the internal side. There is a fixed stoichiometry of Na to glucose transport (2 Na/1 glucose [5]) and transport occurs by an ordered mechanism with the two Na ions binding before the sugar [25–28]. The essential six-state model for Na/glucose cotransport is presented in Fig. 3, where two intermediate states between state 1 and 6 are not shown for simplicity. A single set of parameters for the eight-state model accounts for the kinetic behaviour of the transporter in the forward and reverse modes [28]. The maximum turnover number of the transporter is ∼30 per second at 22 °C. The rate limiting step for transport in the forward direction at saturating ligand concentrations and voltage is the electroneutral dissociation of Na from the cytosolic face of the transporter, steps 5–6.

Figure 3.

 A 6-state kinetic model for Na/glucose cotransport by SGLT1. Two Na ions bind first on the external surface (1–2) to produce a conformation change that permits glucose to bind (2–3). The two Na ions (in yellow) and one glucose molecule are transported across the membrane (3–4), where glucose and the ions dissociate in turn due to the low intracellular Na. The empty binding sites then re-orientate from the inner to the external surface to complete the cycle. The protein is negatively charged with a valence of −2 (red symbols). Under physiological conditions with high extracellular and low intracellular Na and a membrane potential of −50 mV, the maximum turnover number of the transporter at saturating external glucose concentrations (>5 mm) is in the range of 100 per second at 37 °C. Not shown in this figure is the inward Na transport recorded in the absence of glucose (2–5). This occurs at a rate of 5–10% of the maximum rate of Na/glucose cotransport. The simplified model is in essence similar to that proposed earlier [26] and does not include the two intermediate states observed between six and one [19]. External phlorrizin competes with glucose for the sugar-binding site to inhibit transport (not shown).

Protons can substitute equally well for Na to drive glucose transport but the affinity for sugar is about an order of magnitude lower than in Na. This may be of physiological significance in the proximal duodenum, where the chyme has an acid pH and high sugar concentration. d-glucose and d-galactose are transported equally well by SGLT1 with K0.5’s of 0.5 mm, but fructose is not transported. Sugar selectivity has been examined by transport assays of sugar analogues [29]. Although the anomers of d-glucose are both transported well, as are the methyl-glucosides, the importance of the C # 1 oxygen as a hydrogen bond acceptor is indicated by the 10-fold loss of affinity by 1-deoxy-d-glucose. The equatorial −OH groups at carbons # 2 and # 3 are required for transport as the K0.5’s for 2-deoxy-, 3-deoxy-d-glucose. d-mannose and d-allose are >100 mm. At Cs # 4 and # 6 the presence or absence of −OH groups are relatively unimportant, but the C # 6 −CH2 group is favourable in that the pentose d-xylose is poorly transported. The pyranose oxygen is essential in that the imino sugars and inositols are poorly accepted. However, a sulfur is accepted as 5-thio-glucose is transported with a K0.5 of 6 mm.

These results indicate that hydrogen bonding between SGLT1 and the C #1 −O−, the equatorial −OHs at C #s 2 and 3, and the pyranose −O−, and hydrophobic bonding with the C # 6 −CH2 are central in determining sugar specificity. Notable is the finding that beta-, but not alpha-glucosides with phenyl, napthyl and indole groups are transported or bind to SGLT1. For example, indican (indoxyl-beta-d-glucopyranoside) is transported with a K0.5 of 60 micromolar but only at 14% of the maximum rate of d-glucose [30]. This indicates that (i) glucosides with large ring structures in the same plane as the pyranose ring are well accommodated at the sugar-binding site, (ii) large conformation changes of SGLT1 are required to transport substrates as large as 20 × 12 × 5 A° across the membrane, and (iii) large substrates reduce the transporter turnover rate by slowing Na/substrate translocation across the membrane (steps 3–4, Fig. 3). The fact that the beta-glucoside phlorizin is a potent competitive inhibitor of Na/glucose transport, Ki 200 nm, further indicates that there is strong hydrophobic bonding of the glucoside with the protein some 6 A° away from the external sugar-binding site [31].

Novel insights into both the sugar-binding and the conformational changes underlying Na/glucose cotransport are obtained from a combination of cysteine scanning mutagenesis and functional studies [17, 29]. The basic strategy is to mutate residues at or near the putative external sugar-binding site to cysteine, express the mutants in oocytes, and then measure (i) sugar selectivity, (ii) functional effects of –SH reagents such as MTSEA (2-aminoethyl methanethiosulfonate hydrobromide) in the different conformational states (Fig. 3) and (iii) the accessibility of the cysteines to fluorescent functional reagents, e.g. tetramethyl-rhodamine-6-maleimide (TMRM6).

Figure 4 summarizes results obtained so far with single cysteine mutants located in transmembrane helices 10–13. The starting point was Gln457Cys-hSGLT1, where we demonstrated that pretreatment with MTSEA in Na-buffer (conformation state 2, Fig. 3), blocked Na/glucose cotransport [17]. In contrast, there was no inhibition after pretreatment with MTSEA in (i) Na-free buffer (conformation state 1 or 6, Fig. 3), or (ii) in Na-buffer in the presence of phlorizin (conformation state 3), or (iv) in Na-buffer in the presence of saturating glucose (conformation state 5). In Fig. 4, the accessibility of Q457C in the presence and absence of Na is schematically represented by the tilting of helix 13 from the normal plane to open the sugar-binding site. Similar results to those for Q457C were also obtained for cysteine mutants on helix 10 (439 and 445), helix 11 (460 and 468) and helix 12 (499). For those cysteine mutants where the MTS reagents produced no functional effects, we determined access by recording the level of TMRM6 labelling in Na in the presence and absence of sugar or phlorizin, and in the presence and absence of Na. Overall, these results demonstrate that external Na produces a conformational change that permits access to residues lining the hydrophilic cavity bounded by transmembrane helices 10–13. Access to this cavity is blocked either by phlorizin or sugar. It is also noted that phlorizin or glucose increases the accessibility of residue 443 on helix 10 and 528 on helix 13 to TMRM6. This indicates a spatial change in the positions of the external ends of helices 10 and 13 relative to the surrounding transmembrane helices, i.e. sugar and phlorizin induce global conformational changes in the protein.

Figure 4.

 Na-dependent conformational changes in transmembrane helices 10–13 of SGLT1. Residues 439, 445, 457, 460, 468, 499 and 507 (in red) line the hydrophilic cavity involved in glucose binding and translocation. In the absence of external Na the cavity is closed and inaccessible to either glucose or hydrophilic group-specific reagents. In the presence of external Na, the residues become accessible as shown schematically by the rotation and tilt of transmembrane helix 13. Phlorizin reduces the accessibility of these residues to the group-specific reagents but increases the accessibility of residues 443 and 528 (in yellow). See text for further discussion.

The strongest evidence for sugar binding in the hydrophilic cavity formed by helices 10–13 emerges from a detailed examination of the sugar selectivity of the wild-type protein, Q457C, Q457E, Q457R SGLT1 mutants, and the Q457C mutant before and after modification by a panel of neutral, positively and negatively charged cysteine reagents [29]. The results clearly demonstrate that Q457 interacts with glucose oxygens at C #1 and in the pyranose ring. These studies have been extended to the cysteine mutants in Fig. 4 and these point to the involvement of residues 439, 445, 460, 499 and 507 in glucose binding and/or translocation.

Functional disorders of SGLTs


There are two inherited disorders of Na/glucose cotransport: intestinal glucose-galactose malabsorption (GGM, MIM 182380) and familial renal glucosuria (FRG, MIM 233100). The former is due to mutations in the SGLT1 gene whereas the latter are associated with mutations in the SGLT2 gene. Individuals with GGM only have a minor renal glucosuria and individuals with FRG have no defect in intestinal sugar absorption, as expected from the expression pattern of the two genes. GGM was reported simultaneously by two groups in Sweden and Belgium in 1962 and pioneering clinical studies by Meeuwisse et al. at Lund established that it was an inherited diseases probably caused by a defect in the ‘active’ transport of glucose and galactose [32]. The evidence included the findings that the severe diarrhoea is accompanied by glucose and galactose malabsorption, but not fructose malabsorption, and that the diarrhoea ceases immediately after removal of glucose, galactose and lactose from the diet. The underlying cause of GGM was further enhanced by sugar uptake studies on duodenal biopsies from a GGM patient and controls [33]. These demonstrated that the defect was due to a failure of the epithelium to accumulate sugar across the brush border membrane. The therapy for the GGM is to completely remove the offending carbohydrates from the diet. There appears to no long-term consequences of GGM and/or a sugar-free diet as our oldest subject, 55 years old, is living a full and healthy life. This subject has no functional SGLT1 proteins in his body and the diarrhoea promptly returns on the consumption of glucose containing carbohydrate.

A FRG is a rare disorder of the kidney characterized by an isolated defect in the reabsoption of glucose. The level of glucose excretion in the urine ranges from 1 up to 170 g of glucose per day. At the higher level of excretion, there is a complete absence of glucose reabsorption from the glomerular filtrate [34, 35]. Long-term follow up of a subject with a complete failure to reabsorb glucose indicates that he is in good clinical condition with no kidney complication [36]. FRG is considered a rather benign disorder.

After cloning SGLT1 in 1987, we focused our attention on the molecular basis of GGM [37, 38]. We are aware of over 300 GGM subjects world-wide and have conducted genetic tests on some 85 patients in 75 different families. In all but three patients with GGM, we have identified the mutations responsible for the defect in intestinal transport. Around 65% of our subjects are the result of a consanguineous union, i.e. they have the same mutations on both alleles, and the remainder is compound heterozygotes with different mutations inherited from their mother and their father. In addition to 34 different missense mutations, we have identified six nonsense, seven frame shift and seven splice-site mutations that result in premature stop codons and truncated nonfunctional proteins. The residues that were mutated are highly conserved in nine different species of SGLT1 and 10 other members of the SGLT family genes. The mutated residues are distributed widely throughout the protein including 10 of the 14 transmembrane helices (Fig. 5). Somewhat surprising is the large number of conservative mutations, e.g., five Ala/Val and one Ile/Leu mutation.

Figure 5.

 Glucose-galactose malabsorption variants in SGLT1. The residues in red have been tested for function in a heterologous expression system whilst those in blue have not. Conservative Ala/Val mutations are highlighted in black and Gln457Arg is highlighted as this is the only variant expressed at normal levels in the plasma membrane. Revised from Ref. [55].

We have tested the functional effects of 24 of the 34 missense mutations shown in Fig. 5 using the oocyte expression system. All but three exhibited a serious defect in Na/glucose cotransport. The transport defect in most mutants is largely due to a failure to insert the mutant protein into the plasma membrane. This appears to be actual cause of the phenotype in the small intestine as we have six subjects with little or no SGLT1 protein in the brush border membrane. In one case, the mutant Asp28Asn/Gly appears to only affect trafficking to the membrane and not function, a purified truncated hSGLT1 missing residues 1–28 is fully functional when reconstituted into proteoliposomes [21]. Western blots and immunocytochemistry reveals that the level of mutant proteins in oocytes and enterocytes is at least equal to that of the wild type protein, most are core glycosylated, and are found between the nucleus and plasma membrane. We suggest that correct targeting to the brush border membrane is due to changes in the global conformation of the protein such that it is not recognized correctly by the molecular motors responsible for trafficking of the vesicles containing SGLT1 cargo.

In general, every kindred with a GGM subject have a private mutation and only one mutation was found in multiple (three) unrelated families. In the overwhelming majority of cases, mutations have been identified in the SGLT1 gene that account for their severe diarrhoea, and this firmly suggests that SGLT1 is the major transporter involved in glucose and galactose transport across the intestinal brush border. In five of the six GGM subjects that carry a missense mutation that do not impair glucose transport, e.g. Asn51Ser, we have identified other mutations that cause the defect. We have screened 552 alleles in the Coriell Institute genomic DNA collection for variants in the SGLT1 gene for variants and have found that 9% of the European–American population carries this Asn51Ser polymorphism ( Two other variants are found in the European–American population at the same frequency (Ala411Thr and His611Gln). None of the GGM mutations have been found in the 552 alleles tested. Given the relatively high incidence of GGM in Sweden [32], it would be interesting to determine the frequency of GGM mutations in this population.


Several groups have screened the SGLT2 gene for mutations in subjects with FRG [39–44]. As anticipated for a rare autosomal recessive disorder, more than half of the subjects have homozygous mutations and in most of the remainder are compound heterozygotes. The mutations include 17 missense, three nonsense, three frame- shift and one abnormal splice site variant. Unlike GGM, four subjects are found with deletions. So far, every FRG kindred have its own private mutation. Despite the homology between SGLT1 and 2, 59% amino acid identity, there is only one mutation in common, Arg137His. Unfortunately, it has not yet been possible to determine the functional effects of the 17 missence mutations because of the difficulty in expressing SGLT2 in heterologous expression systems. Whilst there is a reasonable correlation between the FRG phenotype and the SGLT2 genotype [41], there are glaring exceptions. For example, although one subject with a premature stop codon on both alleles has severe glucosuria, another unrelated subject with the same genotype has a much lower glucose excretion rate. This suggests that another transporter plays a major role in glucose reabsoption, e.g. SGLT4, 5 or 6. The fact that GGM is only accompanied by a mild renal glucosuria indicates that SGLT1 only plays a minor role.

Oral rehydration therapy

SGLT1 plays a central role in what The Lancet described in 1978 as ‘potentially the most important medical advance in this (20th) century’, namely oral rehydration therapy (ORT) for the treatment of acute diarrhoea [45]. It is estimated that the introduction of ORT decreased the number of deaths of children with severe diarrhoea by 60% from 1980 to 2000, from 4.6 to 1.6 million. ORT is a simple, inexpensive and effective therapy for the dehydration caused by diarrhoea. Toxins produced by bacteria stimulate a massive intestinal fluid secretion, up to 120 mL h−1 for 2–4 days in a 10 kg child with cholera. This fluid loss can be balanced simply by administering an oral salt and glucose solution. The recipe for the oral rehydration calls for a solution of 75 mm glucose and 75-mm NaCl with sufficient potassium and citrate to bring the total osmolarity to 245 milliosmoles L−1. How does ORT work? The glucose stimulates sodium uptake across the brush border membrane of intestinal enterocytes through SGLT1 and this leads to the net absorption of sugar and salt and water from the gut lumen into blood (Fig. 1). Each mole of glucose absorbed is accompanied by two moles of NaCl and 4–6 L of water [46, 47]. Although the precise nature of the link between Na/glucose cotransport and water transport across the brush border is controversial, there is common agreement that SGLT1 activity, either directly or indirectly, can account for the glucose stimulated water absorption. We suggest that there is cotransport of water through SGLT1 (1 glucose/2 Na/264 H2O) and that cotransport in addition sets up an osmotic gradient to pull water into the epithelium across the brush border membrane through both SGLT1 and the plasma membrane. Critics contend that all water transport is osmotic through both SGLT1 and the membrane [48].

SGLTS and diabetes

There has been a long-standing interest in the role of the small intestine in glucose homeostasis in diabetes. Perhaps the most definitive study so far is that of Shirazi-Beechey's group who reported on the expression of SGLT, GLUT5 (the brush border fructose transporter), and GLUT2 in duodenal biopsies taken from control and Type 2 diabetic subjects [49]. They found that brush border Na/glucose cotransport, SGLT1 and GLUT5 protein levels were three- to fourfold higher in diabetics than in controls. In addition, they found approximately twofold increase in the level and activity of brush border enzymes and a twofold increase in level of core structural proteins, and threefold increase GLUT5 and GLUT2 mRNA levels in the mucosa. They concluded that there is an increased capacity to absorb glucose and fructose in diabetic subjects and that this is due to a both a rise in transporter expression and structural changes in the brush borders. The pharmaceutical industry has used a two-prong approach to control intestinal glucose absorption in diabetic subjects: one is to produce glucosidase inhibitors to delay the final digestion of complex carbohydrates and blunt the postprandial rise in blood glucose level [50]; and the other is to develop phlorizin-like SGLT1 inhibitors (three patents issued, see January 2006). It is not clear how patients will tolerate SGLT1 inhibitors owing to the accompanying diarrhoea.

There is a growing interest in controlling blood glucose levels in diabetic patients by blocking the reabsorption of glucose in the proximal tubule predicated on the benign nature of FRG. Some 20 patents have been issued and several companies are reported to have drugs in Phase 2 clinical trials (see January 2006. See also link at A prime example is the oral prodrug T-1095, a phlorizin analogue, which is absorbed from the intestine, converted to the active form, and freely filtered by the kidney where it blocks glucose reabsorption by SGLT2 and SGLT1. In dogs at least, an oral dose of 10 mg kg−1 T-1095 suppressed postprandial hyperglycemia [51]. However, it appears that this prodrug disappeared from the development pipeline in 2004, perhaps because of side effects because of inhibition of SGLTs expressed in other parts of the body (Table 1).


From our perspective, there are two sets of questions that remain to be answered about the SGLT sugar transporters. The first are basic ones about the structural basis of coupled Na and glucose cotransport. Simply, how does this molecular machine actually couple energy stored in the sodium electrochemical potential gradient across the membrane to drive uphill sugar transport? Related problems are how SGLT1 behaves as a water and urea channel and is our hypothesis correct that SGLT1 actually cotransports water. We believe that it is first necessary to obtain high-resolution structural information about SGLT1 in different conformations to gain insight into these questions, but for a fuller understanding we will need dynamic information about how ligands and voltage induce the conformational changes underpinning the transport of Na, sugar and water. Human physiology questions focus on the function of SGLTs in tissues other than the kidney and intestine; for example, in the hippocampus of the brain, muscle and lungs (see Table 1). This is a complex question to tackle in view of the multifunctional properties of SGLTs as Na/glucose cotransporters, water and urea channels, and glucosensors. It would be informative to have access to a data base containing comprehensive, longitudinal, clinical evaluations of GGM and FRG subjects.

Conflict of interest statement

No conflict of interest was declared.


We wish to thank all of our former students, and fellows for their contributions to the SGLT story. Long-term financial support from the National Institutes of Health is gratefully acknowledged. More recently, contributions from the Yamanouchi (USA) Foundation and UCLA Endowments (Mellinkoff Fund) have facilitated pilot and feasibility studies.