Molecular Genetics and Metabolism 86 (2005) 473–477

www.elsevier.com/locate/ymgme

Brief communication

Elements of diabetic nephropathy in a patient with GLUT2 deWciency

Gerard T. Berry a,¤, John W. Baynes b, Kevin J. Wells-Knecht b, Benjamin S. Szwergold c, René Santer d

a Department of Pediatrics, Divisions of Human Genetics and Molecular Biology and Endocrinology and Diabetes, The Children’s Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia, PA 19104, USA

b Department of Chemistry, The University of South Carolina, Columbia, SC, USAc Department of Medicine, Dartmouth University School of Medicine, Hanover, NH, USA

d Department of Pediatrics, University Children’s Hospital, University of Hamburg, Hamburg, Germany

Received 22 July 2005; received in revised form 15 September 2005; accepted 15 September 2005Available online 8 November 2005

Abstract

The Fanconi–Bickel syndrome is caused by homozygosity or compound heterozygosity for mutations of the facilitated glucose trans-porter 2 gene (GLUT2). Glycogen accumulates in renal tubular cells and they fail to reabsorb multiple Wltered solutes because of impair-ment in GLUT2-mediated eZux of glucose. We describe a 10-year-old male child with GLUT2 deWciency who produced massiveamounts of 3-deoxyfructose (3-DF) in the kidneys. Since 3-DF is a detoxiWcation product of a potent glycating agent, 3-deoxyglucosone,a precursor of advanced glycation end-products, this suggests a massive accumulation of glucose within tubular cells probably as a conse-quence of GLUT2 deWciency. The level of 3-DF in the urine of this atypical patient, who also manifested renal glomerular hyperWltration,microalbuminuria, and glomerular mesangial expansion, was higher than in any patient examined with diabetes mellitus. Elevated levelsof glucose and/or its metabolites in renal tubular cells may be necessary but not suYcient for the development of both the renal tubulop-athy and diabetic-like glomerular disease in GLUT2 deWciency. 2005 Elsevier Inc. All rights reserved.

Keywords: GLUT2; Fanconi–Bickel syndrome; Glycogen storage disease; Fanconi syndrome; Diabetes mellitus; Diabetic nephropathy; Glucose trans-port; Advanced glycation end-product; Maillard reaction; 3-Deoxyfructose

Introduction

The Fanconi–Bickel syndrome is caused by a homozy-gosity or compound heterozygosity for mutations of thegene for the facilitated glucose transporter 2 (GLUT2;SLC2A2, glucose–galactose transporter) [1–4]. It is char-acterized by hepatorenal glycogen deposition, galactoseintolerance, fasting hypoglycemia, massive glycosuria,and the renal Fanconi syndrome. We describe a child withthis autosomal recessive disorder of carbohydrate metab-olism who had an unusually high urinary 3-deoxyfructose(3-DF) concentration. This fructose metabolite has been

* Corresponding author. Fax: +1 215 923 9519.E-mail address: gerard.berry@jeVerson.edu (G.T. Berry).

1096-7192/$ - see front matter 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.ymgme.2005.09.010

detected at increased levels in plasma and urine ofpatients with diabetes mellitus [5]. It is a detoxiWcationproduct of 3-deoxyglucosone (3-DG), a potent glycatingagent and a precursor of advanced glycation end-products(AGEs) [6], which are formed by non-enzymatic Maillardreactions [7] between sugars and proteins. Both intra-andextracellular AGEs have been hypothesized to play a rolein the development of diabetic complications, includingnephropathy [5–7]. Our Wndings suggest that renal tubularcell glucose levels are elevated because of GLUT2 deW-

ciency, as 3-DG and, subsequently, 3-DF synthesisdepends on the glucose concentration. The resultant dis-turbance in intracellular metabolism may contribute tothe pathogenesis of renal disease in Fanconi–Bickelsyndrome.

474 G.T. Berry et al. / Molecular Genetics and Metabolism 86 (2005) 473–477

Materials and methods

Patient description

The patient is a 10-year-old white male with short stature,massive hepatomegaly, and inability to walk as a result ofhypophosphatemic metabolic bone disease. The clinical andlaboratory Wndings that established the diagnosis of Fan-coni–Bickel syndrome were the subject of a previous report[8]. Hypoglycemia, occurring within 5 h after a meal, wasoriginally detected at the age of 7 months and necessitatedfrequent feedings and the use of nighttime cornstarch. Dur-ing hypoglycemic episodes, the blood glucose levels werebelow 50mg%, and, when measured, the serum insulin levelswere appropriately reduced. Medications used to treat renalsolute wastage included: sodium bicarbonate, potassium–sodium phosphate, and 1,25-dihydroxycholecalciferol. Toeliminate hypergalactosemia and galactosuria, the patientwas maintained on a lactose-restricted diet. His weight was16.8kg and height 100 cm. As the patient drank liquids, pri-marily fruit juices, every 2–4 h during the day and night, a sig-niWcant amount of his caloric intake was from fructose. Heconsumed approximately 50g of total fructose per day, or3.0 g fructose/kg/day. It was estimated that the mean dailyintake of total fructose per kilogram body weight of 7- to 10-year-old children in the US is approximately 1.2g fructose/kg/day [9]. The patient’s blood pressure was normal. His cog-nitive skills and school performance were outstanding.

Histological examination of liver tissue revealed glyco-gen accumulation within hepatocytes. Glycated hemoglo-bin (HbA1c) ranged from 4.5 to 5.8% (Nl: 4.2–6.8% of totalhemoglobin) and serum fructosamine was 228�M (Nl:174–286 �M). Increased renal glomerular Wltration and uri-nary excretion of albumin were Wrst detected at 5 years ofage. Creatinine and 99mTc-DTPA clearances were greaterthan 200 ml/min/1.73 m2, with peak values of 402 and 281,respectively. A continuous inulin infusion test indicated arenal clearance of 180 ml/min/1.73 m2 (Nl GFR: 95–105 ml/min/1.73 m2). The excretion of albumin ranged from 95 to287�g/min (Nl: <15 �g/min). Histological examination ofthe kidney revealed a mild to moderate increase in glomeru-lar mesangial nuclei in 10 of 14 examined glomeruli. Threeof the fourteen identiWed glomeruli were small and imma-ture, and one had capsular and partial tuft Wbrosis. Elec-tron microscopy of the glomeruli revealed mild patchyswelling of epithelial foot processes. There was some micro-villus formation by these cells. The lamina rara interna ofthe glomerular basement membrane was irregularly wid-ened. An apparent increase in the mesangial matrix wasdetected in one glomerulus. There was no evidence of glyco-gen accumulation in glomeruli using the standard uranylacetate–lead citrate stain. With the same stain, a moderateamount of glycogen was detected in the hyaloplasm oftubular epithelium.

The father and mother are of Eastern European andItalian ancestry, respectively, and are not related. Neitherhad developed diabetes mellitus by the 5th decade.

Control and diabetic subjects

There were 29 healthy non-diabetic control subjects(age D 36 § 10 years) and 35 diabetic patients (age D 43 § 17years). This included 25 patients with type 1 diabetes and 10patients with type 2 diabetes mellitus. Patients were treatedby various drug regimens, including insulin, sulfonylureas,and metformin. Glycated hemoglobin was 4.5 § 0.4% in thecontrol group, compared to 8.8 § 2.4% in the diabeticgroup. Diabetic patients were being treated for the follow-ing complications: cataracts (2), nephropathy (7), neuropa-thy (4), and retinopathy (background, 6; proliferative, 6);some patients had multiple complications. The data onplasma and urine 3-DF and 3-DG levels are reproducedfrom [5].

Molecular genetic analysis

The GLUT2 gene mutational analysis in the familymembers was performed as previously described [3]. High-molecular-weight genomic DNA was prepared from leuko-cytes. Sense (sn) and antisense (asn) primers used for PCRampliWcation of exons 1–10 of GLUT2 with Xanking intronsequences were chosen from the published gene sequence[4,10]. Nucleotides of the GLUT2 cDNA sequence are num-bered from the transcription start site (+1), located 312 ntupstream of the Wrst nucleotide of the ATG initiationcodon [10].

Biochemical determinations

The metabolites, 3-DG, and 3-DF, were measured by gaschromatography–mass spectrometry (GC/MS) as their per-O-acetyl derivatives following reduction to 3-deoxyhexitolwith NaBH4 or NaBD4 [5,6].

Statistical analysis

The data are expressed as means § SD. The signiWcanceof the diVerences in means was determined using a 2-tailedStudent’s t test.

Results

In the patient, two mutations within the GLUT2 genewere detected. The Wrst is a C ! T at nucleotide 469 in exon3 and results in a premature stop codon (R53X). This muta-tion was also detected in the father and was previouslynoted in a patient with Fanconi–Bickel syndrome fromWarsaw, Poland [11]. The second mutation is a G ! A atnucleotide 1264 in exon 6 and predicts a substitution ofarginine for glycine at codon 318 (G318R). This highly con-served amino acid has not been changed in any of theknown human facilitated glucose transporters that havebeen studied. It was also detected in the mother, and previ-ously, in a patient from Milan, Italy [11]. Thus, our patientis a compound heterozygote.

G.T. Berry et al. / Molecular Genetics and Metabolism 86 (2005) 473–477 475

As reported previously (5), the urinary excretion of 3-DF in diabetic subjects (11.1 § 6.1 �mol 3-DF per mmolcreatinine, n D 14) was increased approximately 2-fold(P < 0.01), relative to normal subjects (5.7 § 1.9, n D 12);however, in the patient with Fanconi–Bickel syndrome, itwas increased about eightfold (44.9 § 3.0, n D 3 diVerenturine samples) (Fig. 1A). In fact, this is an underestimationof 3-DF production since 3-DF excretion is expressed permmol creatinine and the patient’s creatinine excretion wasalso extremely elevated because of renal secretion. In con-trast to the levels in urine, the patient’s plasma 3-DF con-centration (1.4 �mol/L) was only slightly above the rangefor normal subjects (0.67 § 0.26, n D 18) and within therange for diabetic subjects (0.99 § 0.28, n D 19) (Fig. 1B).The mean plasma 3-DF concentration was increased in dia-betic subjects compared to normals (P D 0.001). The meanplasma 3-DG concentration in normals was0.042 § 0.006 �mol/L (n D 17) (Fig. 1C). The mean plasma3-DG was increased in diabetic patients (0.094 § 0.022,n D 18, P < 0.0001), comparable to the increase in urinary 3-DF; however, plasma 3-DG was present in the normalrange in the patient with GLUT2 deWciency (Fig. 1C) andwas not detectable in his urine.

Discussion

Like patients with type 1 (von Gierke’s) glycogen stor-age disease (GSD, type 1; glucose-6-phosphatase (G6Pase)deWciency) [12,13], this child with compound heterozygosityfor two mutations of the GLUT2 gene developed renal glo-merular disease. This is most unexpected, as unlike GSD,type 1, glycogen does not accumulate in the glomerularcells of patients with GLUT2 deWciency. Yet, glycogenaccumulation is observed in renal tubular cells in both Fan-coni–Bickel syndrome and GSD, type 1 [12,13] and is pri-marily due to intracellular glucose or glucose-6-phosphateaccumulation as a result of GLUT2 or G6Pase deWciency,respectively. Although expression of GLUT2 in the kidneyis limited to tubular cells, glomerular hyperWltration, micro-albuminuria, and mesangial expansion, point to the pres-

ence of glomerular dysfunction in our patient whosephenotype is apparently atypical. The important questionfrom a genetics perspective is whether his phenotype isrelated to an environmental factor such as dietary fructoseintake or a modiWer gene. Apropos of the lessons learnedwith GSD, type 1, there is also the possibility that this phe-notype is not so atypical given the fact that few patientswith GLUT 2 deWciency have been studied for the presenceof these early and potentially reversible glomerular ele-ments that are the harbinger of diabetic nephropathy. His-torically, only approximately one-third of patients withdiabetes mellitus develop diabetic nephropathy, a criticalfact that underscores the importance of epigenetic phenom-ena in its pathogenesis.

We report here for the Wrst time that 3-DF, a detoxiWca-tion product of the Maillard reaction intermediate, 3-DG,whose formation is dependent on the concentration of glu-cose, was massively elevated in the urine of a patient withGLUT2 deWciency. The accumulation of glucose insiderenal tubular cells is a likely source of increased glycationof protein and formation of 3-DG, which is then detoxiWedto 3-DF and released into urine, either by alternate trans-porters or as a consequence of cell injury. The high concen-tration of 3-DF in urine compared to the plasma indicatesthat most of the urinary 3-DF was formed in the kidney.Since the plasma 3-DF concentration was only slightlyincreased, the data suggest that the urinary 3-DF resultslargely from increased renal production and detoxiWcationof 3-DG. The reason why the patient’s plasma 3-DF level isslightly above the upper conWdence limit for normals ismost likely due to the leak of 3-DF from the renal tubularcells into the peritubular Xuid. These observations arguethat, even in the absence of hyperglycemia, 3-DG accumu-lates in renal tubular cells lacking GLUT2 because of glu-cose since GLUT2 is located on the baselateral membraneand its absence limits glucose eZux.

As a potential precursor of a number of AGEs, 3-DG isthought to play a role in the development of renal and vas-cular damage in diabetes [7], and as an agent possiblyresponsible for diabetic embryopathy [14]. As in these

Fig. 1. Urine and plasma 3-DF/3-DG levels in a patient with Fanconi–Bickel syndrome compared to results in a cohort of control subjects and a cohort ofpatients with type 1 and type 2 diabetes mellitus. (A) Urinary 3-DF is signiWcantly increased in the non-diabetic patient with a genetic deWciency in theGLUT2 transporter. Data on normal control subjects (n D 12) and diabetic patients (n D 14) are reproduced from [5]. Results of analysis of three separatemorning urine samples collected over a period of 12 months are shown for the patient. (B) Plasma 3-DF is slightly increased in the patient. Data on normalcontrol subjects (n D 18) and diabetic patients (n D 19) are reproduced from [5]. (C) Plasma 3-DG is normal in the patient. Data on normal control subjects(n D 18) and diabetic patients (n D 19) are reproduced from [5].

Uri

ne

3-D

F (

mm

ol /

mo

l cre

atin

ine)

0

10

20

30

40

50

Control DM Glut-2 (-/-)

Pla

sma

3-D

G (

µM)

0.00

0.05

0.10

0.15

Control DM Glut-2 (-/-)

Pla

sma

3-D

F (µ

M)

0.0

0.5

1.0

1.5

Control DM Glut-2 (-/-)

A B C

476 G.T. Berry et al. / Molecular Genetics and Metabolism 86 (2005) 473–477

diseases, the source of the 3-DF, and its precursor 3-DG, isnot certain. The latter may be formed by decomposition ofthe Amadori adduct, fructoselysine, on intracellular pro-teins [15], possibly via the fructosamine-3-kinase reaction[16]. It may also be formed by degradation of fructose-3-phosphate, formed from fructose that may be increased inrenal tissue via the sorbitol pathway [17,18]. Finally, other,as yet unidentiWed, Maillard reaction intermediates mayalso be increased in cells with GLUT2 deWciency. Similarly,as a result of G6Pase deWciency, glucose-6-phosphate or itsmetabolites may contribute to the development of kidneydisease in GSD, type 1. Our observations suggest that intra-cellular rather than plasma precursors of toxic glucosemetabolites are likely sources of the renal pathophysiologyin GLUT2 and G6Pase deWciency, and raise questionsabout the relative role of extracellular versus intracellularglucose in development of diabetic nephropathy. In poorlycontrolled diabetes mellitus, the explanation for wastage ofmultiple Wltered solutes had always been the “solvent drageVect” but this may not be the case. Experimental workfrom as far back as 1964 had indicated that only readilymetabolizable hexoses such as glucose infused into normalhuman subjects could lead to an impairment in renal tubu-lar solute reabsorption [19]. The impairment in reabsorp-tion of Wltered solutes in both diabetes mellitus andGLUT2 deWciency may be secondary to renal tubular intra-cellular glucose accumulation. Glucose poisoning withinrenal tubular cells may independently contribute to thedevelopment of renal glomerular disease in both diabetesand GLUT2 deWciency. This is not to say that it is the solecause of diabetic nephropathy but rather that some of theearly and potentially reversible elements, potentially com-mon to both diseases, may be due to disturbances in glu-cose metabolism within tubular cells. Supporting theargument that intracellular glucose is a pathogenic factor inthe development of renal disease in these disorders is thefact that patients with massive renal glycosuria due to deW-cient brush-border membrane Na+/glucose cotransporter 2activity from SGLT2 gene deWciency [20] do not developthe renal Fanconi syndrome or a glomerulopathy. In thesepatients, the luminal glucose is not absorbed into the tubu-lar cells.

Renal glomerular disease was not thought to be a fre-quent complication of the Fanconi–Bickel syndrome [21].However, it will be important to continue to characterizethe renal disease in patients with this extremely rare syn-drome, including the diVerentiation between glomerularand tubular proteinuria, and the determination of the realprevalence of renal insuYciency and glomerulosclerosis. Aprevious review [2] indicated that 4 out of 7 patients had areduced GFR. These 4 patients were in the 2nd or 3rddecade of life. The data of Manz et al., as well as our ownthat implies an early and potentially reversible hyperWltra-tion phase of renal disease, are not incompatible with thetimeline of progression of renal insuYciency seen inpatients with GSD, type 1 [12,13]. Analysis of renal andskin collagen from patients with Fanconi–Bickel syndrome

and GSD, type 1 to determine the tissue speciWcity of Mail-lard-type damage may also clarify the role of intracellularvs extracellular Maillard reaction intermediates in the path-ogenesis of renal disease in Fanconi–Bickel syndrome,GSD, type 1 and diabetes mellitus. Thus, these studies on arare disease resulting from a genetic defect in glucose trans-port may lead to a better understanding of the pathogenesisand multifactorial nature of diabetic nephropathy.

Acknowledgment

The clinical studies were preformed in the General Clini-cal Research Center (NIH Grant MO1RR00240).

References

[1] G. Fanconi, H. Bickel, Die chronische Aminoacidurie (Aminosäuredi-abetes oder nephrotisch-glukosurischer Zwergwuchs) bei der Glykog-enose und der Cystinkrankheit, Helv. Paediatr. Acta. 5 (1949) 359–396.

[2] F. Manz, H. Bickel, J. Brodehl, D. Feist, K. Gellissen, B. Geschöll-Bauer, G. Gilli, E. Harms, H. Helwig, W. Nützenadel, Fanconi–Bickelsyndrome, Pediatr. Nephrol. 1 (1987) 509–518.

[3] R. Santer, R. Schneppenheim, A. Dombrowski, H. Gotze, B. Stein-mann, J. Schaub, Mutations in GLUT2, the gene for the liver-typeglucose transporter, in patients with Fanconi–Bickel syndrome, Nat.Genet. 17 (1997) 324–326.

[4] H. Fukumoto, S. Seino, H. Imura, Y. Seino, R.L. Eddy, Y. Fuku-shima, M.G. Byers, T.B. Shows, G.I. Bell, Sequence, tissue distribu-tion, and chromosomal localization of mRNA encoding a humanglucose transporter-like protein, Proc. Natl. Acad. Sci. USA 85 (1988)5434–5438.

[5] K.J. Wells-Knecht, T.J. Lyons, D.R. McCance, S.R. Thorpe, M.S.Feather, J.W. Baynes, 3-Deoxyfructose concentrations are increasedin human plasma and urine in diabetes, Diabetes 43 (1994) 1152–1156.

[6] T. Niwa, 3-Deoxyglucosone: metabolism, analysis, biological activity,and clinical implication, J. Chromatogr. 731 (1999) 23–36.

[7] K.J. Knecht, M.S. Feather, S.R. Thorpe, J.W. Baynes, Detection ofMaillard reaction intermediates in vivo: 3-deoxyglucosone in humanplasma and 3-deoxyfructose in plasma and urine, Arch. Biochem. Bio-phys. 294 (1992) 130–137.

[8] G.T. Berry, L. Baker, F.S. Kaplan, C.L. Witzleben, Diabetes-like renalglomerular disease in Fanconi–Bickel syndrome, Pediatr. Nephrol. 9(1995) 287–291.

[9] Y.K. Park, E.A. Yetley, Intakes and food sources of fructose in theUnited States, Am. J. Clin. Nutr. 58 (1993) 737S–747S.

[10] J. Takeda, I. Kayano, H. Fukomoto, G.I. Bell, Organization of thehuman GLUT2 (pancreatic �-cell and hepatocyte) glucose transportergene, Diabetes 42 (1993) 773–777.

[11] R. Santer, S. Groth, M. Kinner, A. Dombrowski, G.T. Berry, J.Brodehl, J.V. Leonard, S. Moses, S. Norgren, F. Skovby, R. Sch-neppenheim, B. Steinmann, J. Schaub, The mutation spectrum ofthe facilitative glucose transporter gene SLC2A2 (GLUT2) inpatients with Fanconi–Bickel syndrome, Hum. Genet. 110 (2002)21–29.

[12] Y.T. Chen, R.A. Coleman, J.I. Scheinman, P.C. Kolbeck, J.B. Sidbury,Renal disease in type I glycogen storage disease, N. Engl. J. Med. 318(1988) 7–11.

[13] L. Baker, S. Dahlem, S. Goldfarb, E.F. Kern, C.A. Stanley, J. Egler,J.S. Olshan, S. Heyman, HyperWltration and renal disease in glycogenstorage disease, type I, Kidney Int. 35 (1989) 1345–1350.

[14] U.J. Eriksson, P. Wentzel, H.S. Minhas, P.J. Thornalley, Teratogenic-ity of 3-deoxyglucosone and diabetic embryopathy, Diabetes 47(1998) 1960–1966.

G.T. Berry et al. / Molecular Genetics and Metabolism 86 (2005) 473–477 477

[15] M.S. Feather, Amine-assisted dehydration reactions, Prog. FoodNutr. Sci. 5 (1981) 37–45.

[16] G. Delpierre, M.H. Rider, F. Collard, V. Stroobant, H. Santos, E. VanSchaftingen, IdentiWcation, cloning, and heterologous expression of amammalian fructosamine-3-kinase, Diabetes 49 (2000) 1627–1634.

[17] S. Lal, B.S. Szwergold, A.H. Taylor, W.C. Randall, F. Kappler, K. Wells-Knecht, J.W. Baynes, T.R. Brown, Metabolism of fructose-3-phosphatein the diabetic rat lens, Arch. Biochem. Biophys. 318 (1995) 191–199.

[18] S. Tsukushi, T. Katsuzaki, I. Aoyama, F. Takayama, T. Miyazaki, K.Shimokata, T. Niwa, Increased erythrocyte 3-DG and AGEs in dia-betic hemodialysis patients: role of the polyol pathway, Kidney Int. 55(1999) 1970–1976.

[19] M. Fox, S. Thier, L. Rosenberg, S. Segal, Impaired renal tubular func-tion induced by sugar infusion in man, J. Clin. Endocrinol. Metab. 24(1964) 1318–1327.

[20] R. Santer, M. Kinner, C.L. Lassen, R. Schneppenheim, P. Eggert, M.Bald, J. Brodehl, M. Daschner, J.H. Ehrich, M. Kemper, S. Li Bolti, T.Neuhaus, F. Skovby, P.G. Swift, J. Schaub, D. Klaeke, Molecularanalysis of the SGLT2 gene in patients with renal glucosuria, J. Am.Soc. Nephrol. 14 (2003) 2873–2882.

[21] R. Santer, R. Schneppenheim, D. Suter, J. Schaub, B. Steinmann, Fan-coni–Bickel syndrome—the original patient and his natural history,historical steps leading to the primary defect, and a review of the liter-ature, Eur. J. Pediatr. 157 (1998) 783–797.