15522 Corrections Proc. Natl. Acad. Sci. USA 93 (1996)

Biochemistry. In the article ‘‘The putative actin-binding role of Ecology. In the article ‘‘A meta-analysis of the freshwater hydrophobic residues Trp546 and Phe547 in chicken gizzard cascade’’ by Michael T. Brett and Charles R. Goldman, which heavy meromyosin’’ by Hirofumi Onishi, Manuel F. Morales, appeared in number 15, July 23, 1996, of Proc. Natl. Acad. Sci. Kazuo Katoh, and Keigi Fujiwara, which appeared in number USA (93, 7723–7726), the following correction should be 26, December 19, 1995 of Proc. Natl. Acad. Sci. USA (92, noted. Due to a typesetter’s error that occurred after the page 11965–11969), the authors request that the following be noted. proofs were corrected, the names of the first authors for nine In the Results section under ‘‘Enzymatic Properties,’’ 104 and references were omitted. The first authors for the appropriate 19 ␮MϪ1 should be 1.04 ϫ 104 and 1.9 ϫ 103 MϪ1, respectively. references are (reference number and author name): 17, In the last column of Table 1, 22, 104, 27, and 19 ␮MϪ1 should Andersson, G.; 22, Langeland, A.; 24, Ranta, E.; 28, Ham- be 2.2 ϫ 103, 1.04 ϫ 104, 2.7 ϫ 103, and 1.9 ϫ 103 MϪ1, bright, K. D.; 33, Drenner, R. W.; 35, Mazumder, A.; 36, respectively. The correct table is shown below. Meijer, M. L.; 38, Lazzaro, X.; and 49, Christoffersen, K.

Table 1. ATPase activities of wild-type and mutant HMMs Immunology. The title of the article ‘‘An essential role for MgATPase, tyrosine kinase in the regulation of Bruton’s B-cell ’’ nmol of Pi by J. Simon Anderson, Mark Teutsch, Zengjun Dong, and per min per Henry H. Wortis, which appeared in number 20, October 1, mg of 1996, of Proc. Natl. Acad. Sci. USA (93, 10966–10971), ap- Actin-activated MgATPase peared incorrectly due to a printer’s error. The correct title is ‘‘An essential role for Bruton’s tyrosine kinase in the regula- High Low V , nmol of P per K , max i a tion of B-cell apoptosis.’’ HMM salt salt min per mg MϪ1 ϫ 10Ϫ3 Wild-type Medical Sciences. Ϫkinase 3.7 1.3 76 2.2 In the article ‘‘Expression of the ϩkinase — 3.2 637 10.4 transporter GLUT5 in human breast cancer’’ by S. Pilar Mutant Zamora-Leo´n, David W. Golde, Ilona I. Concha, Coralia I. Ϫkinase 4.4 2.7 49 2.7 Rivas, Fernando Delgado-Lo´pez, Jose´Baselga, Francisco Nu- ϩkinase — 5.8 65 1.9 alart, and Juan Carlos Vera, which appeared in number 5, March 5, 1996, of Proc. Natl. Acad. Sci. USA (93, 1847–1852), Vmax is the maximum actin-activated ATPase activity of HMM and the authors request that the following be noted. ‘‘Our state- Ka is the apparent binding constant for HMM to actin, which is defined ment regarding the absence of GLUT5 immunoreactivity in to be the reciprocal of the apparent Km from the double reciprocal normal breast tissue needs to be revised. Subsequent analysis plots (Fig. 4). To phosphorylate the regulatory light chain of HMM, myosin light chain kinase, calmodulin, and Ca2ϩ were added to the done with more sensitive methodology revealed GLUT5 im- ATPase assay medium. —, Not measured. munoreactivity in some normal breast ductal epithelium. Our conclusion regarding the expression of GLUT5 in human Biochemistry. In the article ‘‘Cloning and characterization of breast as a specific manifestation of the neoplastic state is four murine homeobox ’’ by Alessandra Cecilia Roves- therefore premature.’’ calli, Sadamitsu Asoh, and Marshall Nirenberg, which ap- peared in number 20, October 1, 1996, of Proc. Natl. Acad. Sci. Medical Sciences. In the article ‘‘Core binding factor ␤-smooth USA (93, 10691–10696), the following should be noted: The muscle myosin heavy chain chimeric protein involved in acute Genbank accession numbers for Uncx-4.1, OG-2, OG-9, OG- myeloid leukemia forms unusual nuclear rod-like structures in 12, OG-12a, and OG-12b were incorrect in the footnotes to the transformed NIH 3T3 cells’’ by Cisca Wijmenga, Paula E. article. The correct order is as follows: Uncx-4.1: U65069, Gregory, Amitav Hajra, Evelin Schro¨ck, Thomas Ried, Roland U65070; OG-2: U65067; OG-9: U65068; OG-12: U65071, Eils, P. Paul Liu, and Francis S. Collins, which appeared in U65072; OG-12a: U66918; and OG-12b: U67055. number 4, February 20, 1996, of Proc. Natl. Acad. Sci. USA (93, 1630–1635), the authors request that the following be noted. ‘‘The basic conclusion of this paper, that the CBF␤-SMMHC chimeric protein forms nuclear rod-like structures in NIH 3T3 cells in which the protein is overexpressed, appears correct. However, the data shown in Figs. 3, 5, and 6, involving deletions of the chimeric protein, should be disregarded.’’ Downloaded by guest on September 30, 2021 Proc. Natl. Acad. Sci. USA Vol. 93, pp. 1847-1852, March 1996 Medical Sciences

Expression of the fructose transporter GLUT5 in human breast cancer S. PILAR ZAMORA-LE6N*, DAVID W. GOLDE*t, ILONA I. CONCHAt, CORALIA I. RIVAS*, FERNANDO DELGADO-L6PEZ*, Jost BASELGAt, FRANCISCO NUALART§, AND JUAN CARLOS VERA*¶ *Program in Molecular Pharmacology and Therapeutics and tDepartment of Medicine, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021; TInstituto de Bioquimica, Facultad de Ciencias, Universidad Austral de Chile, Campus Isla Teja, Casilla 567, Valdivia, Chile; and §Departamento de Histologia y Embriologia, Facultad de Ciencias, Universidad de Concepci6n, Concepcion, Chile Communlicated by Paul A. Marks, Memorial Sloan-Kettering Cancer Center, New York, NY, October 9, 1995 (received for review August 30, 1995)

ABSTRACT The primary metabolic characteristic of ma- normal breast tissue (19). Immunohistochemical evidence has lignant cells is an increased uptake of and its anaer- been obtained indicating that GLUTI is highly expressed in obic metabolism. We studied the expression and function of breast cancer cells compared to their normal counterparts, but the glucose transporters in human breast cancer cell lines and no apparent changes in the expression of GLUT2 were de- analyzed their expression in normal and neoplastic primary tected (19). We analyzed the expression and function of human breast tissue. Hexose uptake assays and immunoblot- glucose transporters in the human breast cancer cell lines ting experiments revealed that the breast carcinoma cell lines MCF-7 and MDA-468 and in normal and neoplastic human MCF-7 and MDA-468 express the glucose transporters breast tissue and found that in addition to overexpressing GLUT1 and GLUT2, isoforms expressed in both normal and GLUT1, human breast cancer tissue selectively expresses the neoplastic breast tissue. We also found that the breast cancer high-affinity fructose transporter GLUT5. cell lines transport fructose and express the fructose trans- porter GLUT5. Immunolocalization studies revealed that GLUT5 is highly expressed in vivo in human breast cancer but MATERIALS AND METHODS is absent in normal human breast tissue. These findings The human breast cancer cell lines MCF-7 and MDA-468 were indicate that human breast cancer cells have a specialized obtained from American Type Culture Collection and grown capacity to transport fructose, a metabolic substrate believed in a mixture of Dulbecco's modified Eagle's medium contain- to be used by few human tissues. Identification of a high- ing high glucose (17.5 mM) and F-12 medium (1:1; vol/vol) affinity fructose transporter on human breast cancer cells supplemented with 10% fetal bovine serum and 1% penicillin/ opens opportunities to develop novel strategies for early streptomycin. For uptake assays, the cells were grown as diagnosis and treatment of breast cancer. monolayer cultures in six-well plates to a density of 1 x 106 cells per well. Cultures were carefully selected under the Breast cancer is the second leading cause of cancer death in microscope to ensure that only plates showing uniformly women in the United States and it is estimated that -12% of growing cells were used. Two wells in each plate were used to women in the United States will develop breast cancer during determine the number of cells, and the four companion wells their lifetime (1). Breast cancer cells have a high level of were used for the uptake assays. The cells were washed with and metabolism, a circumstance common to incubation buffer (22) (15 mM Hepes/135 mM NaCl/5 mM most cancer cells (2). The high rate of glucose uptake in cancer KCl/1.8 mM CaCl2/0.8 mM MgCl2) and incubated in the same cells is used in the clinic to localize tumors in patients and to medium for 30 min at 37°C. Uptake assays were performed at assess tumor metabolism and response to therapy by positron room temperature in 1 ml of incubation buffer containing 0.2 emission tomography (PET) scanning with [18F]fluorodeoxy- mM deoxyglucose and 2-4 ,tCi (1 Ci = 37 GBq) of 2-deoxy- glucose (3-6). D-[1,2-(N)3H]glucose (30.6 Ci per mmol; DuPont/NEN). Up- Two systems for the transport of glucose are available in take was stopped by washing the cells with ice-cold phosphate- mammalian cells: the Na+/glucose (7) ex- buffered saline (PBS). Cells were dissolved in 0.5 ml of lysis pressed primarily in and kidney, and the buffer (10 mM Tris HCl, pH 8.0/0.2% SDS), and the incor- facilitative family (8), products of distinct porated radioactivity was assayed by liquid scintillation spec- genes that are expressed in all cells in a finely controlled and trometry. Where appropriate, competitors and inhibitors were tissue-specific manner. Six different facilitative glucose trans- added to the uptake assays or preincubated with the cells. porter isoforms have been molecularly cloned: GLUTI, ex- Fructose uptake assays were performed in incubation buffer pressed in all tissues and especially abundant in erythrocytes containing 1 mM fructose and 0.8 ,uCi of D-[U-14C]fructose and brain (9); GLUT2, present in , pancreatic islet ,B cells, per ml (285 mCi/mmol; Amersham). Samples were processed kidney, and at the basolateral surface of the adsorptive cells of as indicated for deoxyglucose uptake. Data represent means + the small intestine (10); GLUT3, abundant in brain (11); SD of four samples. GLUT4, restricted to adipose and skeletal tissues (12); Immunoblotting was performed as described (23) using GLUT5, expressed in small intestine and sperm cells (13, 14); anti-GLUT antibodies (East Acres Biologicals, Southbridge, and GLUT7, restricted to microsomes of liver cells (15). MA) and horseradish peroxidase goat anti-rabbit IgG and Available evidence indicates that the mechanism by which enhanced chemiluminescence (Amersham). For immunocyto- cancer cells increase their ability to take up glucose involves chemistry, cells were grown on eight-well microscope slides, the selective overexpression of GLUTI (16-19). fixed with buffered formaldehyde/acetone, washed with PBS, It is known that GLUTI is responsible for glucose transport and incubated in PBS containing 5% bovine serum albumin in breast tissue (19-21) and GLUT2 has also been detected in (BSA) followed by incubation for 1 h at room temperature in the same buffer containing 1% BSA, 0.3% Triton X-100, and The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in Abbreviation: PET, positron emission tomography. accordance with 18 U.S.C. §1734 solely to indicate this fact. STo whom reprint requests should be addressed. 1847 1848 Medical Sciences: Zamora-Leon et al. Proc. Natl. Acad. Sci. USA 93 (1996) anti-GLUT antibodies (1:100) or rabbit preimmune serum. uptake data in terms of intracellular concentrations. After 1 h Cells were then incubated with fluorescein isothiocyanate goat of incubation, the MDA-468 cells accumulated an intracellular anti-rabbit IgG (Life Technologies; 1:40) for 1 h, mounted, and concentration of deoxyglucose in excess of 140-fold the exter- analyzed by fluorescence microscopy. Breast tissue expression nal concentration, whereas the MCF-7 cells accumulated of GLUT5 was determined by immunohistochemical analysis intracellularly "15-fold the external glucose concentration. of a set of thin sections prepared from archived paraffin tissue Data from Lineweaver-Burk plots (Fig. 1 B and C) or from blocks. Paraffin was removed by incubating the sections in Eadie-Hofstee plots (data not shown) revealed the presence of xylene followed by absolute alcohol and then the sections were two functional components with separate affinities for uptake hydrated by immersion in graded alcohol solutions. Sections of deoxyglucose in both cell lines-a high-affinity component were incubated in PBS containing 5% skim milk, washed with with an apparent Km for transport of 2 mM and a second PBS, and incubated with the anti-GLUT antibodies (1:100) for component of lower affinity with an apparent Km for transport 2 h. After extensive washing with PBS, sections were incubated of 10 mM. When we measured uptake of deoxyglucose at very for 1 h with alkaline phosphatase goat anti-rabbit IgG (1:500) short intervals, from 5 sec to 2 min, the rate of uptake of and color developed with 4-nitroblue tetrazolium chloride and deoxyglucose by MDA-468 cells was 3-fold higher than that of 5-bromo-4-chloro-3-indolyl phosphate. MCF-7 cells (Fig. 1D). Therefore, the 5-fold difference in uptake observed in long-term uptake experiments (Fig. 1A) RESULTS reflects a step secondary to transport, most likely the intra- cellular trapping of deoxyglucose as deoxyglucose 6-phosphate We measured the uptake ofdeoxyglucose, an analog ofglucose (8). Kinetic analysis using Lineweaver-Burk (Fig. 1 E and F) transported only by the facilitative glucose transporters (8), in or Eadie-Hofstee (data not shown) plots confirmed the pres- MCF-7 and MDA-468 cells. These cell lines are widely used to ence of two functionally distinct glucose transporters in the characterize the behavior of human breast cancer in vitro and breast cancer cells. Using the Michaelis expression for a in vivo (24). Both cell lines showed a notable capacity to take single-substrate reaction under conditions of initial velocity up deoxyglucose (Fig. 1A). Uptake was approximately linear (25), we estimated that at a glucose concentration of 5.5 mM, for the first 20 min of incubation, with the MDA-468 cells the high-affinity system comprises three-quarters of the ca- taking up at least 4-fold more deoxyglucose than the MCF-7 pacity of the breast carcinoma cell lines to take up glucose (Fig. cells. Methylglucose, an analog of glucose that enters the cells 1 G and H). but is not metabolized, was used to estimate the intracellular The apparent Km values for transport of deoxyglucose in volume available for exchange with the external medium (8). breast cancer cells were similar to those described for the MDA-468 cells had an approximate intracellular exchange transport of deoxyglucose mediated by GLUT1 and GLUT2 in volume of 3 ,ul per 106 cells, while in MCF-7 cells the value was other cellular systems. We further tested the expression of 5 ,ul per 106 cells. The volume estimates were used to express GLUT2 in breast cancer cells by measuring the transport of

100 A 0.08 0.3B C 60 ME 0.06 / Q,0 > 0.2 - 0.04 a _ 40 E 20 0.1 0.02 -_ 0.0ir MCF-7 0.00 MDA-468 0 2 4 6 0 2 4 6 1/ DOG (1/mM) 1/ DOG (1/mM) 8 FD7 1.8 -E . (50 cosq 6 MDA-468 > 1.2

0 0 4 0.6 * a E 2 CF-i - 0.04 f MCF-7 0 30 60 90 120 0 1 2 3 4 0 1 2 3 4 Time (sec) 1/ DOG (1/mM) 1/ DOG (1/mM)

0 5 10 15 20 DOG (mM)

FIG. 1. Human breast cancer cell lines MCF-7 and MDA-468 express two functionally distinct facilitative glucose transporters. (A) Time course of the uptake of deoxyglucose by MCF-7 (0) and MDA-468 (-) cells. (B) Double reciprocal plot of substrate dependence for the uptake of deoxyglucose by MCF-7 cells using 10-min assays. (C) Double reciprocal plot of substrate dependence for the uptake of deoxyglucose by MDA-468 cells using 10-min assays. (D) Time course of transport of deoxyglucose by MCF-7 (0) and MDA-468 (0) cells. (E) Double reciprocal plot of substrate dependence for the transport of deoxyglucose by MCF-7 cells using 40-sec assays. (F) Double reciprocal plot of substrate dependence for the transport of deoxyglucose by MDA-468 cells using 40-sec assays. (G) Uptake of deoxyglucose mediated by the high-affinity (-) and the low-affinity (0) transporter in MCF-7 cells. (H) Uptake of deoxyglucose mediated by the high-affinity (0) and the low-affinity (0) transporter in MDA-468 cells. Data represent means ± SD of four samples. DOG, deoxyglucose. Medical Sciences: Zamora-Le6n et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1849

a- 12 A 0.4 C 0.8 0.3 i - 6 MDA-468 -0.2 0C.)u) 0.4 / E 36 ° - 0.1 MCF-7 MDA-468 0 0.0 0.0 0 10 20 30 40 0.0 0.4 0.8 0.0 0.2 0.4 Time (min) 1/ Fructose (1/mM) 1/ Fructose (1/mM)

_ 100 DD 100 t E 100 Z 0 0D S80 280 Cyt E 8 60 t60 CL 860

0 40 40 -0 4 Cyt B 0. co 20,2 , 0 MCF-7 'c 20 MDA-468 a?20 F 0 5 10 15 20 25 0 5 10 1520 25 ~~~0 0 0.1 10 1000 Fructose (mM) Fructose (mM) Cytochalasin (uM)

1 8°0 H CytE 100 CDo .~~~ 2~~~ Cyt E ~~~~~-gl'uco'seCD$s CSL2"80. a, ~80 D o 60 . , 0, 60 ructose. 4040 CytB 0 DOG 202 0 0.1 10 1000 0 0.1 10 1000 0 0.1 10 1000 Cytochalasin (uM) Hexose (mM) Hexose (mM) FIG. 2. Human breast cancer cell lines MCF-7 and MDA-468 express a high-affinity fructose transporter distinct from GLUT2. (A) Time course of the uptake of fructose in MCF-7 (0) and MDA-468 (0) cells. (B) Double reciprocal plot of substrate dependence for the transport of fructose by MCF-7 cells using 50-sec assays. (C) Double reciprocal plot of substrate dependence for the transport of fructose by MDA-468 cells using 50-sec assays. (D) Uptake of fructose mediated by the high-affinity (-) and the low-affinity (0) pathway in MCF-7 cells. (E) Uptake of fructose mediated by the high-affinity (0) and the low-affinity (0) pathway in MDA-468 cells. Data represent means ± SD of four samples. (F) Effect of cytochalasin B (-) and cytochalasin E (0) on uptake of deoxyglucose by MCF-7 cells. (G) Effect of cytochalasin B (0) and cytochalasin E (0) on uptake of fructose by MCF-7 cells. (H) Effect of fructose (0), deoxyglucose (0), and L-glucose (v) on uptake of deoxyglucose by MCF-7 cells. (I) Effect of deoxyglucose (-) and fructose (v) on uptake of fructose by MCF-7 cells. DOG, deoxyglucose. fructose because GLUT2 is able to transport fructose in GLUT2 (26). At 10 mM fructose, the transport of deoxyglu- addition to glucose (26). Both cell lines were able to take up cose was not substantially affected, and at 100 mM fructose fructose (Fig. 2A), and uptake was linear for approximately there was only -30% inhibition of deoxyglucose uptake (Fig. the first 60 sec. In long-term uptake studies, we observed that 2H). These results suggest that GLUT2 likely corresponded to fructose uptake by MDA-468 cells was -5-fold greater than in the low-affinity pathway. As controls we used L-glucose (8), a MCF-7 cells. At short uptake times the difference was only sugar that is not transported by the facilitative glucose trans- 2-fold, similar to that observed for transport of deoxyglucose. porters and did not inhibit the transport of deoxyglucose by the Lineweaver-Burk analysis of the transport of fructose revealed MCF-7 cells, and deoxyglucose, which competed for two fructose transport activities-a high-affinity component with an apparent Km of 10 mM for fructose transport in both MCF-7 and MDA-468 cells (Fig. 2 B and C) and a low-affinity 12 1 2 1 2 1 2 1 2 pathway that showed no saturation at concentrations of fruc- tose as high as 50 mM (data not shown). The value of 10 mM is 1 order of magnitude lower than the values previously described for transport of fructose by GLUT2 (26). At con- 977- centrations of fructose lower than 10 mM, the high-affinity pathway contributed >90% of the capacity of the MCF-7 (Fig. 2D) and MDA-468 (Fig. 2E) cells to transport fructose. 69 - These data suggested the presence of the fructose trans- porter GLUT5 in MCF-7 cells. GLUT5 has a Km for the transport of fructose of 6 mM, is not inhibited by cytochalasin B, a specific inhibitor of the facilitative hexose transporters, 46- and does not transport deoxyglucose (14). In MCF-7 cells, transport of deoxyglucose was completely inhibited by cy- tochalasin B but not by cytochalasin E, an analog of cytocha- lasin B that does not interact with the glucose transporters (8) (Fig. 2F). Fifty percent inhibition was observed at 0.4 ,tM cytochalasin B, a value that falls between the value of the K1 GLUTTI GLUT2 Gl.UT3 GILUT4 GLUT.S for and the for GLUT2 GLUT1 (-0.2 ,uM) Ki (-2 ,uM) (27). FIG. 3. Hexose transporters GLUT1, GLUT2, and GLUT5 are On the other hand, 100 ,uM cytochalasin B inhibited <40% of expressed in human breast cancer cell lines MCF-7 (lanes 1) and fructose uptake, and <20% inhibition was observed with 1 ,uM MDA-468 (lanes 2). Results of the immunoblots with anti-GLUTI, cytochalasin B (Fig. 2G). Cytochalasin E did not interfere with anti-GLUT2, anti-GLUT3, anti-GLUT4, and anti-GLUT5 antibodies the transport of fructose. It is known that fructose can are shown. Sizes on left are kDa. Arrow indicates the migration of the completely inhibit the uptake of deoxyglucose mediated by human erythrocyte glucose transporter. 1850 Medical Sciences: Zamora-Leon et al. Proc. Natl. Acad. Sci. USA 93 (1996) completely the uptake of fructose mediated by GLUT2 but does not affect fructose uptake mediated by GLUT5 (14). Deoxyglucose (100 mM) inhibited <20% of fructose uptake (Fig. 21), contrasted with the effect of 100 mM fructose, which inhibited [3H]fructose uptake completely (Fig. 21). The presence of GLUT1, GLUT2, and GLUTS in the human breast cancer cell lines was confirmed by immuno- blotting and immunolocalization with anti-glucose transporter antibodies. Immunoblotting experiments revealed the pres- ence of several overlapping anti-GLUT1 immunoreactive bands in total cell homogenates from both cell lines (Fig. 3). Bands of 45-80 kDa were labeled in the MCF-7 cells, com- pared to bands of 60-90 kDa in the MDA-468 cells. The anti-GLUT1 antibody reacted with a band of -45 kDa in an immunoblot of human erythrocyte (data not shown). GLUT2 was also expressed in both breast cancer cell lines (Fig. 3). Anti-GLUT2 antibodies reacted with several bands of 46-90 kDa and differences in the intensities of the various bands were observed depending on the cell line. The anti- GLUT2 antibody reacted with a unique band of -50 kDa in an immunoblot of total liver proteins (data not shown). GLUT5 was also expressed in the MCF-7 and MDA-468 cell lines (Fig. 3). Bands of 50-70 kDa were labeled in the MCF-7 cells, compared to bands of 50-85 kDa in the MDA-468 cells. Proteins from human testis showed a broad band of 40-60 kDa when immunoblotted with the anti-GLUT5 antibody (data not shown). No reactivity was observed when the blots were incubated with an anti-GLUT3 antibody, whereas GLUT4 was present in very small amounts in the two breast cancer cell lines (Fig. 3). GLUT3 reacted with a unique 50-kDa band present in human sperm proteins and GLUT4 reacted with a 40- to 50-kDa band from human (data not shown). Immunolocalization studies using immunofluorescence con- firmed the presence of the transporters GLUT1, GLUT2, and GLUT5 in the breast cancer cell lines (Fig. 4). Most of the fluorescence was associated with the adjacent to the nucleus and the plasma membrane, and the staining was most FIG. 4. Immunolocalization of hexose transporters expressed in hu- man breast cancer cell lines MCF-7 and MDA-468. For immunohisto- intense in cells probed with the anti-GLUT1 and anti-GLUT5 chemistry, cells were incubated with the different anti-GLUT antibodies antibodies. Cellular staining with anti-GLUT2 antibodies was followed by incubation with a secondary antibody coupled to fluorescein. weaker than with anti-GLUT1 or anti-GLUT5 antibodies and was clearly stronger in MDA-468 than in MCF-7 cells (Fig. 4). [3H]deoxyglucose uptake completely at 100 mM (Fig. 2H). A low level of fluorescence was observed when both cell lines Previous data indicated that deoxyglucose is able to inhibit were probed with anti-GLUT4 antibodies, and no fluorescence

FIG. 5. Immunolocalization of fructose transporter GLUT5 in human breast tissue. (A) Immunohistochemistry of normal mammary tissue. Mammary epithelial cells do not express GLUT5 as revealed by basal anti-GLUT5 reactivity. (B and C) Immunohistochemical localization of GLUT5 in human breast cancer tissue. High-expression of GLUT5 is evidenced by strong anti-GLUT5 immunoreactivity. (x25.) Medical Sciences: Zamora-Leo'n et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1851 was detected in cells reacted with anti-GLUT3 antibodies or does not transport glucose (14) and is expressed in small preimmune serum (Fig. 4). intestine, sperm cells, and brain (31, 32), with very low level The results of immunoblotting and immunofluorescence expression in adipose tissue and muscle (13, 14, 28). GLUT5 experiments were concordant with the results of the transport is also expressed in Caco-2 cells, a human colon cancer cell line studies and indicate that the breast cancer cell lines MCF-7 and that differentiates in culture into cells with the properties of MDA-468 express high levels of the transporters GLUT1, small intestine (33). GLUT5 transports fructose GLUT2, and GLUT5. The presence of the high-affinity fruc- with high affinity (14), whereas GLUT2 transports this sugar tose transporter GLUT5 in breast cancer cell lines was sur- with very low affinity (26). At the low concentrations of prising since this transporter is believed to have a restricted cell fructose present in vivo, GLUT5 is likely to mediate a high and tissue distribution (13, 14, 28). The finding suggested that fraction (>90%) of the uptake of fructose due to the order of the expression of GLUT5 in these cells could be related to the magnitude difference in the respective Km for transport of neoplastic state. We therefore tested for expression of GLUT5 fructose by GLUT2 and GLUT5. These findings suggest an in normal and neoplastic primary human breast tissue. All 20 important role for GLUT5 in cellular uptake of fructose by primary breast cancer tissues tested were positive for expres- breast cancer cells compared to normal breast tissue in which sion of GLUT5 (Fig. 5 B and C; data not shown). Strong transport of fructose is mediated by GLUT2. Our results staining was seen in the perinuclear region, cytoplasm, and cell indicate that neoplastic transformation of breast epithelial membrane of tumor cells. Staining was also seen in malignant cells leads to expression of a high-affinity fructose transporter cells invading the fibroadipose tissue. There was no staining of permitting enhanced uptake offructose, a substrate apparently normal mammary tissue, indicating that normal mammary used by few human tissues. Based on the Warburg theory, we epithelium does not express GLUT5 (Fig. 5A; data not shown). can speculate that fructose may be a good substrate for energy No staining was observed in breast cancer tissue probed with generation in malignant cells that prefer the glycolytic pathway preimmune serum. In control experiments, anti-GLUT1 stain- since lactic acid generation through may not be ing was stronger in breast cancer tissue compared to normal subjected to the regulatory steps that control glucolysis. The breast tissue, consistent with overexpression of GLUT1 in the fructolysis pathway may provide the neoplastic breast cancer neoplastic cells (data not shown). Thus, while GLUT1 is cells with a metabolic advantage. These results suggest that present in normal breast and is overexpressed in breast cancer, fructose uptake could represent a useful target for PET GLUT5 is absent in normal breast tissue and is expressed at imaging and possibly the development of novel therapeutic high levels in human breast cancer. agents in breast cancer. This work was supported by grants from the National Institutes of DISCUSSION Health (CA30388, RO1 HL42107, and P30 CA08748), Memorial Our data indicate that the breast carcinoma cell lines MCF-7 Sloan-Kettering Institutional funds, and Grant S-95-24 from the and MDA-468 express three members of the facilitative hexose Direcci6n de Investigaci6n, Universidad Austral de Chile. transporter family, the glucose transporters GLUT1 and GLUT2, and the high-affinity fructose transporter GLUT5. 1. Harris, J. R., Lippman, M. E., Veronesi, U. & Willet, W. (1992) Deoxyglucose uptake and competition experiments in the cell N. Engl. J. Med. 327, 390-398. 2. Warburg, 0. (1956) Science 123, 309-314. lines indicated the presence of two functionally distinct glucose 3. Minn, H. & Soini, I. (1989) Eur. J. Nuclear Med. 15, 61-66. transport systems with the characteristics expected for GLUT1 4. Tse, N. Y., Hoh, C. K, Hawkins, R. A., Zinner, M. J., Dahlbom, and GLUT2, with apparent Km values of 2 and 10 mM for the M., Choi, Y., Maddahi, J., Brunicardi, F. C., Phelps, M. E. & transport of deoxyglucose, respectively. Fructose uptake and Glaspy, J. A. (1992) Ann. Surg. 216, 27-34. competition experiments indicated the presence in breast 5. Nieweg, 0. E., Kim, E. E., Wong, W. H., Broussard, W. F., carcinoma cell lines of a high-affinity transporter of fructose Singletary, S. E., Hortobagyi, G. N. & Tilbury, R. S. (1993) functionally similar to GLUT5 (apparent Km, 8 mM) and a Cancer 71, 3920-3925. low-affinity pathway that failed to saturate at fructose con- 6. Wahl, R. L., Zasadny, K., Helvie, M., Hutchins, G. D., Weber, B. centrations as high as 50 mM functionally similar to GLUT2 & Cody, R. (1993) J. Clin. Oncol. 11, 2101-2111. > 50 The of 7. Hediger, M. A. & Rhoads, D. B. (1994) Physiol. Rev. 74,993-1026. (Km mM). presence GLUT1, GLUT2, and GLUT5 8. Carruthers, A. (1990) Physiol. Rev. 70, 1135-1176. in the breast carcinoma cell lines was confirmed by immuno- 9. Mueckler, M., Caruso, C., Baldwin, S. A., Panico, M., Blench, I., blotting and immunolocalization experiments, which also re- Morris, H. R., Allard, W. J., Lienhard, G. E. & Lodish, H. F. vealed a low level of expression of GLUT4 and no expression (1985) Science 229, 941-945. of GLUT3. 10. Thorens, B., Sarkar, H. K, Kaback, H. R. & Lodish, H. F. (1988) We found that GLUT1 and GLUT2 were present in normal Cell 55, 281-290. and neoplastic breast tissue but that GLUT5 was expressed in 11. Kayano, T., Fukumoto, H., Eddy, R. L., Fan, Y. S., Byers, M. G., human breast cancer and was absent from normal human Shows, T. B. & Bell, G. I. (1988)J. Bio. Chem. 263,15245-15248. breast tissue. Enhanced uptake and metabolism of glucose is 12. Birnbaum, M. J. (1989) Cell 57, 305-315. a common characteristic of cancer cells and involves increased 13. Kayano, T., Burant, C. F., Fukumoto, H., Gould, G. W., Fan, Y. S., Eddy, R. L., Byers, M. G., Shows, T. B., Seino, S. & Bell, expression of GLUT1 in neoplasia (2, 16-19). Enhanced G. I. (1990) J. Biol. Chem. 265, 13276-13282. glucose uptake in tumor cells is used to detect tumors and to 14. Burant, C. F., Takeda, J., Brot-Laroche, E., Bell, G. I. & Dav- follow their response to treatment in a noninvasive manner by idson, N. 0. (1992) J. Biol. Chem. 267, 14523-14526. PET imaging with [18F]fluorodeoxyglucose (3-6), a substrate 15. Waddell, I. D., Zomerschoe, A. G., Voice, M. W. & Burchell, A. that enters cells through the glucose transporters. GLUT2 is (1992) Biochem. J. 286, 173-177. present in normal and neoplastic human breast tissue, al- 16. Birnbaum, M. J., Haspel, H. C. & Rosen, 0. M. (1987) Science though no changes in expression in the neoplastic state have 235, 1495-1498. been detected (19). 17. Flier, J. S., Mueckler, M. M., Usher, P. & Lodish, H. F. (1987) Our finding that GLUT5 is expressed in human breast Science 235, 1492-1495. cancer and is in 18. Yamamoto, T., Seino, Y., Fukumoto, H., Koh, G., Yano, H., absent normal breast tissue may have clinical Inagaki, N., Yamada, Y., Inoue, K., Manabe, T. & Imura, H. application. GLUT2 and GLUT5 can mediate the transport of (1990) Biochem. Biophys. Res. Commun. 170, 223-230. fructose (14, 26) and GLUT2 is also capable of transporting 19. Brown, R. S. & Wahl, R. L. (1993) Cancer 72, 2979-2985. glucose (26, 27, 29). GLUT2 is expressed in liver, pancreatic 3 20. Threadgold, L. C. & Kuhn, N. J. (1984) Biochem. J. 218,213-219. cells, the basolateral surface of kidney and small intestine 21. Burnol, A. F., Leturque, A., Loizeau, M., Postic, C. & Girard, J. epithelia (10, 30), and in normal breast tissue (19). GLUT5 (1990) Biochem. J. 270, 277-279. 1852 Medical Sciences: Zamora-Le6n et al. Proc. Natl. Acad. Sci. USA 93 (1996)

22. Vera, J. C., Rivas, C. I., Fischbarg, J. & Golde, D. W. (1993) 28. Rand, E. B., Depaoli, A. M., Davidson, N. O., Bell, G. I. & Nature (London) 364, 79-82. Burant, C. F. (1993) Am. J. Physiol. 264, G1169-G1 176. 23. Vera, J. C., Rivas, C. I., Velasquez, F. V., Zhang, R. H., Concha, 29. Vera, J. C. & Rosen, 0. M. (1989) Mol. Cell. Biol. 9, 4187-4195. I. I. & Golde, D. W. (1995) J. Biol. Chem. 270, 23706-23712. 30. Thorens, B., Cheng, Z. Q., Brown, D. & Lodish, H. F. (1990)Am. 24. Baselga, J., Norton, L., Masui, H., Coplan, K., Miller, W. H. & J. Physiol. 259, C279-C285. Mendelsohn, J. (1993) J. Natl. Cancer Inst. 85, 1327-1333. 31. Shepherd, P. R., Gibbs, E. M., Wesslau, C., Gould, G. W. & 25. Dixon, M. & Webb, E. C. (1979) Enzymes (Academic, New Kahn, B. B. (1992) 41, 1360-1365. York), pp. 55-78. 32. Mantych, G. J., James, D. E. & Devaskar, S. U. (1993) Endocri- 26. Colville, C. A., Seatter, M. J., Jess, T. J., Gould, G. W. & nology 132, 35-40. Thomas, H. M. (1993) Biochem. J. 290, 701-706. 33. Mahraoui, L., Takeda, J., Mesonero, J., Chantret, I., Dussaulx, E., 27. Axelrod, J. D. & Pilch, P. F. (1983) Biochemistry 22, 2222-2227. Bell, G. I. & Brot-Laroche, E. (1994) Biochem. J. 301, 169-175.