GLUT1 is the most abundantly expressed hexose transporter in MMTV-c-ErbB2 mammary tumors and numerous mouse mammary carcinoma cell lines
A qPCR screen evaluating the expression of eleven of the twelve mouse GLUT family members was performed with the following samples: tumors derived from MMTV-c-ErbB2 mice 
, two mammary tumor cell lines derived from such tumors (78617 and 85815 
), a mammary carcinoma cell line derived from MMTV-PyVMT mice (Met1 
), a BALB/c mouse mammary carcinoma cell line (4T1 
), and immortalized mouse mammary epithelial cell line (EPH4). This initial screen demonstrated that GLUT1 was the most abundantly expressed GLUT family member in ErbB2 overexpressing mouse mammary tumors and all five cell lines examined (). Other GLUT family members which were expressed included GLUT6, GLUT8 and GLUT9 (, ). We further analyzed the expression of GLUT1, GLUT6, GLUT8 and GLUT9 in multiple samples of the ErbB2-overexpressing tumors and cell lines by performing qPCR against a standard curve for each transporter to determine the number of copies of each transporter, independent of any reference gene (). This analysis demonstrated that GLUT1 was the most highly expressed transporter in all samples and that all samples also expressed GLUT8, 78617 cells expressed GLUT6 and the ErbB2 tumors expressed GLUT9 (). These samples all lack expression of GLUT7 (data not shown, ), the one GLUT family member not examined in our initial screen ().
GLUT1 is the most abundantly expressed GLUT family member in MMTV-c-ErbB2 tumors and a number of mouse mammary carcinoma cell lines.
Reduced expression of GLUT1 in 78617GL cells decreases glucose usage, lipid synthesis and proliferation in vitro.
Immunoblot analysis utilizing lysates from a primary MMTV-c-ErbB2 mouse mammary tumor and five cell lines derived from tumors taken from these mice (78617, 78622, 78717, 85815 and 85819) using HeLa cells as a control, revealed that all express GLUT1 protein (). Numerous immunoreactive bands are present in GLUT1 immunoblot analysis due to glycosylation of GLUT1 
and later experiments show all bands are reduced by RNAi or Cre recombinase and conversely all bands are present when GLUT1 is overexpressed. 85815 cells express very low levels of the GLUT1 protein demonstrating that GLUT1 protein levels do not necessarily correlate with the amount of RNA detected by qPCR (). Expression of keratin 18 (K18) demonstrates that the cell lines are of epithelial origin and all lanes are loaded similarly ().
Reduced expression of GLUT1 in 78617GL cells decreases glucose usage, lipid synthesis and proliferation in vitro
We had previously noted increased lactate concentration, decreased glucose concentration and increased GLUT1 expression in tumors from MMTV-c-ErbB2 mice compared to mammary epithelial cells 
and wanted to test whether manipulation of GLUT1 expression levels in ErbB2-overexpressing mammary carcinoma cells could modulate aspects of glucose usage, bioenergetics and biosynthesis in vitro
. The 78617 cell line was transduced with a retrovirus encoding GFP-luciferase to create “78617GL” cells. These cells were transduced with lentiviruses expressing shRNA molecules targeting GLUT1 or a non-silencing control shRNA to generate pools of cells stably expressing each shRNA. Immunoblot analysis revealed that the GLUT1 shRNA effectively reduced GLUT1 protein expression as compared to cells expressing the control shRNA (). 78617GL cells with undetectable amounts of GLUT1 protein were viable and survived in culture for numerous passages and demonstrated no apparent increase in caspase 3 activation (data not shown). There was also no difference in the expression of numerous Bcl2 family members, including PUMA (data not shown), which has been shown to be increased under conditions of low glucose to mediate p53-dependent cell death 
We sought to determine whether these cells with reduced GLUT1 protein have altered glucose usage. There is a 35% decrease in the uptake of 3H-2-deoxyglucose (3H -2-DOG) in 78617GL cells expressing GLUT1 shRNA compared to the cells expressing control shRNA (). Similarly, 78617GL cells expressing control shRNA consume more glucose and secrete more lactate than 78617GL cells expressing GLUT1 shRNA (). These data demonstrate that reduction of GLUT1 expression leads to a reduction of glucose transport, glucose consumption and lactate secretion suggestive of reduced glycolysis and glucose metabolism.
We wanted to determine if decreasing GLUT1 expression would reduce proliferation since cell division requires energy (ATP) and carbon for macromolecule biosynthesis, both of which can be supplied by glucose. 78617GL cells expressing GLUT1 shRNA had a reduced rate of proliferation as compared to the cells expressing the control shRNA, which was apparent on days 1, 2, and 3 (). Similarly, 78617 cells expressing GLUT1 shRNA formed fewer colonies when grown in soft agar (, Figure S1
). Colonies from cells expressing control shRNA and GLUT1 shRNA had similar patterns of cleaved caspase 3 staining: larger colonies tended to have staining at the core of the colony while medium and small colonies tended to have very little cleaved caspase 3 staining (Figure S1
), though the colonies from shGLUT1 cells did appear to have more frequent central apoptosis. Nuclear BrdU incorporation tended to occur at the perimeter of the colonies (Figure S1
) and 78617GL cells expressing control shRNA had a higher percentage of BrdU positive nuclei at their edge than colonies derived from cells expressing GLUT1 shRNA ( and Figure S1
) suggesting cells with reduced GLUT1 have reduced proliferation in soft agar. 78617 cells expressing GLUT1 shRNA or control shRNA (grown on plastic) had identical concentrations of ATP () suggesting that impaired glucose usage and GLUT1 expression does not deplete ATP under the conditions of this study. However, 78617GL cells expressing GLUT1 shRNA cultured with radiolabeled glucose convert less of the aqueous label to the non-aqueous lipid containing fraction than cells expressing control shRNA (), suggesting reduced lipid synthesis in cells expressing GLUT1 shRNA.
The shRNA mediated reduction of GLUT1 appears to eliminate detectable GLUT1 protein by immunoblot analysis (), but 78617GL cells expressing GLUT1 shRNA still transport and consume glucose (), albeit at a lower rate than the control cells. qPCR analysis reveals that GLUT1 is still the most highly expressed GLUT family member in cells expressing GLUT1 shRNA and that no other GLUT family member or SGLT1 is being upregulated (at the RNA level) in cells lacking detectable GLUT1 protein (). Thus, glucose consumption by cells expressing GLUT1 shRNA may still occur via remaining GLUT1 (or other transporters) or by non-specific means such as pinocytosis.
Reduced expression of GLUT1 decreases tumor growth in nude mice
To address the role of GLUT1 in tumor growth in vivo, 78617GL cells expressing control shRNA or GLUT1 shRNA were transplanted into contralateral #4 mammary glands of five athymic nude mice and tumor growth was monitored by bioluminescence on days 2, 4, 6, and 8 post tumor cell transplant () which is quantitated and averaged for all five mice (). In each mouse, the tumor derived from cells expressing GLUT1 shRNA was smaller than the contralateral tumor derived from cells expressing control shRNA, and this difference was most evident on days 6 and/or 8 ( and data not shown), and this is also true of the average across all five mice (). Four of the five tumors derived from cells expressing control shRNA outweighed the contralateral tumor expressing GLUT1 shRNA (with the fifth tumor pair weighing the same) (data not shown), which corroborates the bioluminescence data by suggesting decreased growth of cells expressing GLUT1 shRNA in vivo. A separate, similar experiment utilizing five mice corroborated these results: 78617GL cells expressing GLUT1 shRNA are smaller than control tumors in their initial ten days of growth (data not shown).
Reduced expression of GLUT1 in 78617GL cells decreases tumor growth.
Immunoblot analysis of tumor lysates and immunohistochemical staining of tumor sections reveals that tumors derived from 78617GL cells expressing GLUT1 shRNA maintain low levels of GLUT1 protein (); however this GLUT1 protein may reflect either endothelial cells or infiltrating immune cells present in the tumors analyzed. qPCR analysis revealed that there was no change in the levels of other GLUT family members between the two tumor types suggesting that the low level of GLUT1 was not compensated for by an increase in the expression of these other GLUTs (data not shown), similar to what was observed with these cell lines in vitro prior to transplantation into mice ().
Surprisingly, tumors derived from cells expressing control shRNA and GLUT1 shRNA both revealed very little staining for cleaved caspase 3 (). Quantification of cleaved caspase 3 staining revealed no difference in apoptosis between the two tumor types (data not shown). Quantification of the percentage of BrdU positive nuclei in 3 high power images from four tumors of each type revealed that tumors derived from cells expressing GLUT1 shRNA had a lower rate of proliferation than the control tumors (). This suggests that the differences in tumor size more likely reflect differences in proliferation than differences in apoptosis. Since these tumors were harvested at an early stage when they weighed ~200 mg, there was no central necrosis observed.
Eliminating GLUT1 expression in G1fP cells decreases glucose usage, lipid synthesis and proliferation in vitro
Use of RNAi reduced, but did not eliminate GLUT1 expression in 78617GL cells, so we generated a system in which we could eliminate GLUT1 expression. Harvested mammary epithelial cells from GLUT1fl/fl
mice (Figures S2
) were transformed in culture with the polyomavirus middle T antigen, an oncogene that shares many similarities to ErbB2 and is frequently used in mouse models of breast cancer 
, to establish a stably proliferating cell line named “G1fP” (for G
yVMT. G1fP cells were infected at an MOI of 100 with adenovirus expressing GFP or Cre recombinase which resulted in the expression of GFP and the elimination of detectable GLUT1, respectively (). Expression of CK18 in these cells demonstrates their epithelial origin and equal loading (). G1fP cells exposed to Cre recombinase had a 60% reduction in 3
H -2-DOG transport, a 60% reduction in glucose consumption and an approximate 80% reduction in lactate secretion as compared to cells exposed to GFP adenovirus (). G1fP cells lacking GLUT1 also had a reduced rate of proliferation (). While there was no difference in the ATP content of G1fP cells expressing GLUT1 or lacking GLUT1 expression (), the cells lacking GLUT1 expression had a greater than 60% decrease in the flux of glucose into lipid (). The expression of the GLUT family of transporters and SGLT1 evaluated by qPCR revealed that this cell line preferentially expressed GLUT1 with low level expression of GLUT3, GLUT6, and GLUT8, but removal of GLUT1 by Cre recombinase did not increase the expression of any of the other remaining hexose transporters ().
Eliminating expression of GLUT1 in G1fP cells decreases glucose usage, lipid synthesis and proliferation in vitro.
Eliminating expression of GLUT1 decreases tumor growth in nude mice
To test whether G1fP cells, which were transformed in culture, are capable of tumor formation, G1fP cells expressing luciferase exposed to Ad-GFP or Ad-Cre were injected into contralateral #4 mammary glands of three athymic nude mice. In all three mice, the cells which had been exposed to GFP formed tumors with latencies ranging from 55–120 days while the mammary glands injected with cells exposed to Cre recombinase had no palpable tumor at the time of sacrifice or by later histologic examination (data not shown). A tumor cell line, “G1fPt” (for G1fP tumor), was generated from the first tumor arising from injected Ad-GFP infected G1fP cells. G1fPt cells were expanded in culture and then exposed to Ad-GFP or Ad-Cre and lysates evaluated by immunoblot analysis as described to confirm GLUT1 expression and elimination (data not shown). G1fPt cells exposed to Ad-GFP or Ad-Cre were injected into contralateral mammary glands of eight athymic nude mice and tumor development was monitored by bioluminescence on days 12, 15 and 20 post-injection (). The average bioluminescence from the two tumor types was similar on day 12, but by day 20, the tumors from the cells exposed to Cre were more than 35% smaller than the tumors with GLUT1 intact (). Similarly, in six of the eight harvested tumor pairs, the tumor from cells exposed to GFP weighed more than the tumor from cells exposed to Cre; and only one Cre tumor weighed more than the contralateral GFP tumor (data not shown). Examination of GLUT1 expression in the tumors by immunoblot analysis and IHC demonstrates that the tumors from Ad-Cre infected cells maintain reduced GLUT1 expression as compared to the control GFP tumors (). Both tumor types had similar, low levels of cleaved caspase 3 staining (). Tumor sections were stained for Ki67 (), a proliferation marker, and tumors from cells exposed to Cre recombinase had a 25% lower proliferative index than the tumors exposed to GFP (). These results corroborate the data from 78617GL cells: cells with reduced GLUT1 have reduced proliferation and form tumors more poorly than the control cells.
Elimination of GLUT1 expression in G1fPt cells decreases tumor growth.
Overexpression of GLUT1 in 85815GL cells increases glucose transport without increasing proliferation
85815 cells contain low levels of GLUT1 protein compared to the other mouse mammary tumor cell lines examined () and reduction of GLUT1 in 85815 cells does not reduce proliferation of these cells in vitro or tumor growth initiated by these cells in vivo (data not shown). To ask whether increasing the level of GLUT1 in these cells would enhance proliferation and tumor growth in vivo, GLUT1 was overexpressed in 85815 cells expressing GFP-luciferase (85815GL cells).
85815GL cells were infected with a retroviral vector encoding human GLUT1 cDNA followed by selection in puromycin-containing media. Immunoblot analysis of lysates from these cells demonstrates greatly elevated levels of GLUT1 as compared to cells transduced with an empty vector control (). Overexpression of GLUT1 increased 3H-2-DOG transport greater than five-fold as compared to the empty vector control cells (), demonstrating that the overexpressed GLUT1 was functional. The overexpression of GLUT1 in 85815GL cells did not increase proliferation and in fact proliferation slowed slightly by day 3 ().
Overexpression of GLUT1 in 85815GL cells increases glucose transport without increasing proliferation.
Overexpression of GLUT1 accelerates tumor formation
To examine whether overexpression of GLUT1 stimulates tumor growth, 85815GL cells overexpressing GLUT1 or empty vector were injected into contralateral mammary glands of five athymic nude mice and tumor growth monitored by bioluminescence on days 3, 6, 9, 12 and 14 post tumor cell transplant (). Each tumor derived from cells overexpressing GLUT1 was bigger than the contralateral control tumor by day 14, and the average bioluminescence shows that as a group the tumors overexpressing GLUT1 were bigger on days 12 and 14 (). Additionally, weights of tumors harvested on day 14 demonstrate that all five tumors derived from tumors overexpressing GLUT1 are larger (by 6–73%) than the contralateral tumor (data not shown).
Overexpression of GLUT1 in 85815GL cells accelerates tumor formation.
Immunoblot and immunohistochemistry analysis demonstrated that tumors derived from 85815GL cells overexpressing GLUT1 maintain robust expression of GLUT1 while the control tumors have very little GLUT1 expression (). Low power photomicrographs of cleaved caspase 3 stained sections demonstrate that tumors derived from 85815GL cells overexpressing GLUT1 appear to have less apoptosis than the control tumors ( left), which is in contrast to the low levels of cleaved caspase 3 observed in the previous two models ( and ). Indeed, tumors overexpressing GLUT1 have five times less cleaved caspase 3 staining than the control tumors (). Tumors derived from 85815GL cells overexpressing GLUT1 also have increased proliferation as determined by BrdU nuclear incorporation (). These data suggest that the increased tumor formation by 85815GL cells overexpressing GLUT1 results from both a decrease in apoptosis and an increase in proliferation.