Deletion of Amino Acid Transporter ASCT2 (SLC1A5) Reveals an Essential Role for Transporters SNAT1 (SLC38A1) and SNAT2 (SLC38A2) to Sustain Glutaminolysis in Cancer Cells*

Many cancer cells depend on glutamine as they use the glutaminolysis pathway to generate building blocks and energy for anabolic purposes. As a result, glutamine transporters are essential for cancer growth and are potential targets for cancer chemotherapy with ASCT2 (SLC1A5) being investigated most intensively. Here we show that HeLa epithelial cervical cancer cells and 143B osteosarcoma cells express a set of glutamine transporters including SNAT1 (SLC38A1), SNAT2 (SLC38A2), SNAT4 (SLC38A4), LAT1 (SLC7A5), and ASCT2 (SLC1A5). Net glutamine uptake did not depend on ASCT2 but required expression of SNAT1 and SNAT2. Deletion of ASCT2 did not reduce cell growth but caused an amino acid starvation response and up-regulation of SNAT1 to replace ASCT2 functionally. Silencing of GCN2 in the ASCT2(−/−) background reduced cell growth, showing that a combined targeted approach would inhibit growth of glutamine-dependent cancer cells.

since depletion of TCA cycle intermediates will render the cycle non-functional, and compromise the capacity to generate energy (2). Linear pathways, by contrast, increase the flow-through to adapt to high metabolic demand. Converting the TCA cycle into a linear pathway is of considerable advantage to rapidly dividing cells, such as cancer cells, activated lymphocytes and stem cells. This linearized version of the TCA cycle has been termed glutaminolysis, which allows to generate energy and metabolic building blocks (2, 3). Glutamine, the starting substrate of this pathway, is imported by the cell and deaminated to glutamate. Glutamate is converted into 2-oxoglutarate by glutamate dehydrogenase or by a transaminase reaction. After going through several steps of the TCA cycle, malate or oxaloacetate are being produced. Oxaloacetate can be converted into aspartate, which is a major precursor for nucleotide biosynthesis. Both oxaloacetate and malate can also be converted into pyruvate, which can be used to generate alanine, lactate or acetyl-CoA. As a result, many cancer cells depend on glutamine, unless they express glutamine synthetase (4). It is worth noting that glutaminolysis through the conversion of 2-oxoglutarate to oxaloacetate produces 2 NADH, 1 FADH, and 1 GTP, and therefore provides the cell with significant energy. Activation of the glutaminolysis pathway in cancer cells is accompanied by a reduced entry of pyruvate into the TCA cycle (5). The adaptation of metabolism to cellular growth is associated with transcription factors myc and HIF, both of which are up-regulated in many cancer cells (6,7). In contrast to differentiated cells, expression of glutamine transporters is essential for fast growing cells. Overexpression of myc is associated with the induction of ASCT2 (SLC1A5) (8), a glutamine transporter highly expressed in many cancer cells (9). As a result, ASCT2 is generally considered to mediate the entry of glutamine to the glutaminolysis pathway (e.g. (10,11)). In addition, ASCT2 is often expressed together with 4F2hc/LAT1 (SLC3A2/SLC7A5), a heteromeric antiporter that exchanges large neutral amino acids. Both transporters have been implicated in cancer growth and mTOR signaling in many studies (e.g. (12,13)). It has been proposed that ASCT2 takes up glutamine, which then acts as an exchange substrate to accumulate leucine via 4F2hc/LAT1 (10). This proposal is problematic, as ASCT2 is also an obligatory amino acid exchanger for small neutral amino acids and does not mediate net uptake of glutamine unless other amino acids are available for release (14). Moreover, glutamine is not a good intracellular exchange substrate for 4F2hc/LAT1 (15). Thus expression of a net transporter for neutral amino acids is likely to be important for cell growth. Net neutral amino acid transporters are found in the SLC38 family of Sodium Neutral Amino acid Transporters (SNAT) (16). The family is subdivided into two groups, namely system A amino acid transporters and system N amino acid transporters. System A amino acid transporters [SNAT1 (SLC38A1), SNAT2 (SLC38A2), and SNAT4 (SLC38A4)] are Na + -neutral amino acid cotransporters transporting a wide variety of small neutral amino acids, whereas system N transporters [SNAT3 (SLC38A1), SNAT5 (SLC38A5), and SNAT8 (SLC38A8)] are more substrate specific, preferring glutamine, asparagine, and histidine (16). Functionally, system N transporters are characterized by their tolerance to Na + replacement by Li + , while system A transporters are sensitive to inhibition by the amino acid analogue N-methylaminoisobutyric acid (MeAIB). SNAT5 has been shown to be activated by myc, but its contribution to glutamine uptake in cancer cells has not been studied (17). Here we demonstrate that system A (SNAT1 and SNAT2) activity is essential to mediate net glutamine uptake and glutaminolysis in cancer cell lines, while ASCT2 and LAT1 harmonize intracellular amino acid pools. We thus propose a unified model of amino acid homeostasis in cancer cells.

EXPERIMENTAL PROCEDURES
Cell lines and cell culture-Human cervical adenocarcinoma epithelial cells (HeLa)  and human thymidine-kinase-negative osteosarcoma cells, 143B (TK-) [gifted by Dr. David Tscharke (Research School of Biology, ANU)] were cultured in DMEM/Ham's F12 (Sigma D8437) containing 10% foetal bovine serum (FBS, Life Technologies) at 37°C in a humidified atmosphere of 5% CO 2 in air. For subculturing, cells were detached by trypsinization (0.25% trypsin-EDTA, GIBCO). Cell counting was performed using a Scepter cell counter (Millipore, USA) or a haemocytometer. All complete media for maintaining cells were supplemented with 2mM L-glutamine (GIBCO). Cell viability after trypsinization was generally ≥95% as evaluated by trypan-blue exclusion.
Growth assays-To eliminate effects of serum-derived trace glutamine, dialyzed FBS was used (Sigma). For growth assays, HeLa and 143B cells were seeded into 96-well culture plates at a density of 4,000-5,000 cells per well and kept initially in DMEM/Ham's F12/10% dialyzed FBS without glutamine. After 4h, media were replaced with the desired test solutions and cell proliferation was measured using an IncuCyte system (Essen BioScience). IncuCyte data are shown as cell confluence (mean ± SEM) at set intervals.
Metabolomics-Labelling of metabolite pools derived from [ 13 C5]glutamine was carried out as described (18), with some modifications. 143B cells were labelled in the presence of 4mM [ 13 C5]glutamine for 1h or 24h in otherwise glutamine-free DMEM/Ham's F12 supplemented with 10% dialyzed FBS. After the labelling period, medium was aspirated and 1 ml of deionized water was added to the cells. After swirling briefly, the water was removed and cells covered with liquid nitrogen. Once the liquid nitrogen had evaporated, 600µl of ice-cold CH3OH:CHCl3 (9:1) (supplemented with 1mM ribitol) was added and incubated for 10min. The cells were scraped together and transferred into microcentrifuge tubes and incubated for an additional 5min at 4°C. The extracts were centrifuged at 4°C for 5min at 16,100 x g and the supernate transferred into a fresh 1.5ml tube. Samples were sent on dry ice to Metabolomics Australia (Parkville, Victoria) for analysis.
High-performance liquid chromatography (HPLC)-The HPLC system consisted of a Dionex UltiMate™ 3000 UHPLC in line with a fluorescence FLD-3400 detector. To measure amino acid consumption over time, 143B cells were grown initially in 35-or 60-mm dishes and maintained in complete culture media in the tissue-culture incubator until grown to full confluence. Before starting the experiment, the cells were washed thrice in Hanks' balanced salt solution pH7.4 containing 5mM D-glucose and 0.024mM NaHCO3. Cells were then incubated for 5h in the same buffer containing 0.2mM glutamine. Supernates were sampled at 0, 2, and 4 or 5h, cleared by centrifugation and freeze-dried overnight. Dried samples were resuspended in 100μl 50mM NaHCO3 (pH9.0). Benzylserine (3μl 1mM) was added as internal standard and 10μl of 2,4,6-trinitrobenzenesulfonic acid (TNBS, 10mM in CH3OH) was added for amino acid derivatization. Subsequently, samples were incubated for 2h at 37°C, and 10μl was injected for analysis. Samples were eluted off a Kinetex 1.7 μ C18 (2.1mm × 100mm) column at 35°C. The mobile phase consisted of (solvent A) 100mM ammoniumacetate (pH 7.0) and (solvent B) acetonitrile. The gradient was A/B = 90%/10% at 0 min increasing to 50%/50% after 10 min and to 100% B after 12 min. Eluents were detected at λ = 335 nm. Standard curves were generated for amino acids of interest by varying the injection volume and integrating the corresponding peak areas.
Reverse-transcription PCR-NucleoSpin RNAII kit (Macherey-Nagel) was used to isolate total RNA from cells according to manufacturer's instructions. RNA concentration was quantified by a NanoDrop spectrophotometer (Thermo Fisher Scientific). RNA was reverse-transcribed to cDNA using a SuperScript-II reversetranscriptase kit (Invitrogen). The cDNA of interest was amplified using Taq polymerase (Qiagen) using serial dilutions of the template to optimize semi-quantitative analysis. PCR primer sequences are available on request.
RNA silencing-Low passages (<20) of 143B cells were grown in DMEM/Ham's F12 (Sigma D8347) medium supplemented with 10% FBS and 2mM glutamine (total concentration 4mM). On the day before transfection, cells were split and seeded out in 35 mm cell culture dishes at 150,000-300,000 cells. Immediately before transfection the medium was renewed. For transfection (all volumes per dish) 4μl of Lipofectamine RNAimax (Life Technologies) was combined with 250μl OPTI-MEM (Life Technologies) and separately 30pmol of RNAi construct (Ambion Silencer Select Pre-designed siRNAs as listed in Table 1) was combined with 250μl of OPTI-MEM. Both solutions were combined after 5min and were incubated for a further 20-30min at room temperature before adding the transfection complexes dropwise to the cells. All transfections were performed in triplicates. Transfected cells were incubated at 37°C and 5%CO2 for 4-6h after which the medium was replaced with fresh DMEM/Ham's F12/FBS10%/2mM glutamine. Transport or western blot analyses were performed after 48h unless stated otherwise.
Genomic mutation of the ASCT2 (Slc1a5) gene-A commercial CRISPR/Cas9 system was used (Sigma). The construct U6gRNA-pCMV-Cas9-2A-GFP contains a 22bp guide RNA (cctcgaagcagtcaacctcccg) resulting in cleavage/repair of the Slc1a5 gene in exon 7. An endotoxin-free preparation (Macherey and Nagel) of the plasmid was used for transfection of 143B cells maintained in DMEM/Ham's F12/10%FBS/2mM glutamine. Cells were split and seeded out in a 60-mm dish to reach 40% confluence on the day before transfection. Immediately before transfection, the cells were replenished with fresh DMEM/HamF12/10%FCS/2mM glutamine. Plasmid DNA (4μg) and 10μl Lipofectamine 2000 (Invitrogen) were separately incubated in 500μl of Opti-MEM (Invitrogen) for 5min at room temperature, before combining them and incubating for a further 20-30min at room temperature to form complexes. The complexes were then added drop-wise to the cells and placed in an incubator at 37°C/5 % CO 2, followed by a media change after 4-6h. After 48h of expression, cells were trypsinized (0.25 %Trypsin/EDTA (Invitrogen)) and collected by centrifugation (500 x g) followed by three washes in Dulbecco's phosphate-buffered saline supplemented with 5mM glucose and 1% dialyzed FBS (Sigma). Cells were then passed through a cell strainer (70μm, Corning), centrifuged and suspended in PBS supplemented with 5mM glucose and 1% dialyzed FBS. Single cell sorting was performed using an ARIA II FACS machine (FACS Facility Australian National University) with GFP intensity set at two different levels. Single cells were collected in 96-well plates containing DMEM/Ham's F12/10%FBS/2mM glutamine and incubated for up to 3 weeks at 37°C/5%CO 2. Established clones were trypsinized and seeded into 25cm 2 cell culture flasks (Corning) for further propagation. A compound heterozygote mutation of human ASCT2 at the target site was identified by sequencing.
Membrane protein preparation and surface biotinylation-Cells were grown on 100 mm dishes and washed thrice in 10ml ice-cold PBS. Cells were collected and spun down for 5min at 500 x g. The supernate was replaced by 5ml hypoosmotic buffer (15mM KCl, 2mM MgCl2, 0.1mM EDTA, 10mM HEPES, pH 8.0) and the cells resuspended by mixing. After incubation on ice for 5min, membranes were ruptured by 35 strokes in a Potter homogenizer. Debris and the nuclei were removed by centrifugation at 2000 x g for 10min. Membranes were isolated from the supernatant by centrifugation at 180,000 x g at 4°C for 60min. Pellets were resuspended in 200μl 5mM glycine. For surface biotinylation, cells were grown on 100mm dishes and washed thrice in 10ml modified PBS (supplemented with 1mM CaCl2 and 1mM MgCl2, pH 8.0). Cells were then covered with 0.5 mg/ml EZ-link Sulfo-NHS-lc-Biotin (Thermo Fisher Scientific) in modified PBS (pH8.0) and incubated for 45min at room temperature on a rotary shaker at low speed. Biotinylation was terminated by washing thrice in modified PBS supplemented with 100mM glycine, pH8.0. Cells were scraped together, transferred to a 1.5ml reaction tube and lysed by addition of 1ml 150 mM NaCl, 1% Triton X-100, 20mM Tris.HCl pH7.5. The homogenate was incubated on ice for 1.5h to complete lysis. Subsequently, the homogenate was centrifuged at 13500 x g in a table-top centrifuge for 10min, the supernate was transferred to a new tube. After protein determination, equal amounts of protein homogenate were added to 150μl high capacity streptavidin agarose beads. The beads were incubated overnight at 4°C on a rotary shaker before washing four times in lysis buffer. The streptavidin-agarose slurry was mixed with protein sample buffer and sample reducing reagent. After boiling for 5min, a 40μl sample was loaded onto polyacrylamide gels.
SDS-PAGE and western blotting-For protein extraction, cells were lysed in RIPA protein-extraction buffer (Sigma), supplemented with Protease Inhibitor Cocktail (EDTA-free, Roche). After homogenization samples were centrifuged at 16,000 × g for 10min. Total soluble proteins from the supernates were measured using the Bradford reagent, using bovine serum albumin as standard.
To prepare protein samples for SDS-PAGE, 50 -100μg total protein was mixed with 5μl 4 x LDS sample buffer (Invitrogen), 2μl reducing agent (Invitrogen) and made up to 20μl using MilliQ water. Samples were then incubated at 70°C for 10min before loading onto the gel. Electrophoresis was performed using 4-12% Bis-Tris polyacrylamide NuPAGE® gels (Invitrogen), electrophoresed in an XCell SureLock® Mini-Cell (Invitrogen) under reducing conditions according to standard manufacturer procedures. The SeeBlue Plus 2 pre-stained protein ladder (Invitrogen), was used to estimate the apparent molecular weight of fractionated proteins. Following SDS-PAGE, fractionated proteins were transferred onto nitrocellulose membranes (GE Healthcare) using the Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad) according to standard protocols. Blots were blocked for 2h at room temperature (or overnight at 4°C) in 50ml 10% (w/v) skim milk in PBS with 0.15% TWEEN 20 (PBS-T). After washing thrice in PBS-T for 10min each, the blots were incubated with the first antibody for 2 h or overnight in 5ml skim milk (2%, w/v) in PBS-T at dilutions listed in Table 2. Excess primary antibody was removed by washing three times with PBS-T. Blots were incubated with 5ml of diluted secondary antibody for 1h. After washing thrice in PBS-T and a final rinse in PBS, reactive bands were detected by enhanced chemiluminescence, using Luminata Crescendo or Forte western HRP Substrate (Millipore Merck). For re-probing, the same blots were incubated for 45min at 70°C in 50ml stripping buffer (62.5mM Tris-HCl (pH6.8), 2% SDS and 100mM 2-mercaptoethanol). Membranes were then washed thrice with PBS-T and blocked for 3h using 10% (w/v) skim milk in PBS-T before re-probing with next antibody as described above.
Oocyte expression systems and flux experiments-Xenopus laevis (Nasco/USA) oocytes were isolated and maintained as described previously (19). Selected oocytes were injected with 10ng of human ASCT2, human SNAT1, human SNAT2 or human SNAT5 cRNA and were used as described previously (20).
Growth and statistical analysis-Growth curves were analyzed using the Gompertz equation for cell growth y=a × exp(-exp(k (x-xc))) or the logistic equation y=a/(1+ exp(-k (x-xc))). For all experiments, the number of technical repeats is shown as (n) and the number of biological/experimental repeats shown as (e). Unpaired T-tests were performed to compare means of the combined biological repeats.

RESULTS
Glutamine is an essential amino acid for cells that carry out the glutaminolysis pathway (3). In agreement with this metabolic phenotype we found that growth of HeLa cells and 143B cells highly depended on glutamine. Both cell lines completely ceased growth in the absence of glutamine (data not shown). At the same time, both cell lines produced alanine, aspartate and glutamate in a glutamine-dependent manner. For instance, 143B cells produced 4 ± 0.2μM aspartate, 23 ± 5μM glutamate and 80 ± 10μM alanine from 200μM glutamine over 5h. Moreover, metabolic labelling with [ 13 C5]glutamine resulted in strong 13 C enrichment of cellular glutamate, succinate, fumarate, malate and aspartate pools (results presented further below). These results confirmed the use of the glutaminolysis pathway. Accordingly, myc was detected in the nuclei of both cell lines (data not shown). Having established functional glutaminolysis in both cell lines, we wondered how import of glutamine is mediated in these cells. Analysis of amino acid transporter expression by RT-PCR revealed the presence of a variety of glutamine transporters, namely ASCT2, LAT1, LAT2, SNAT1 and SNAT2. SNAT5 was detected in HeLa but not in 143B cells (Fig. 1A,B). A separate analysis of SNAT isoforms in 143B cells showed high expression levels of SNAT1 and SNAT2 (Fig. 1C). SNAT4, SNAT7 and SNAT9 were weakly expressed, and may serve roles in addition to glutamine transport. For instance, SNAT9 is a lysosomal arginine transporter involved in mTOR signaling (21,22). To correlate the expression data with functional data, we analyzed glutamine uptake into HeLa and 143B cells using competition experiments. The competing amino acids and analogues were chosen to identify system A activity (MeAIB) and ASCT2 activity (Thr, which is not an inhibitor of system A) and system L (Leu) in an additive manner as published previously (23). Despite the dominant mRNA expression of system A isoforms SNAT1 and SNAT2, the system A inhibitor MeAIB decreased glutamine uptake only by a small fraction (Fig. 1D,E). Glutamine uptake was predominantly Na + -dependent, particularly in 143B cells. With the exception of system N (molecular isoforms SNAT3 and SNAT5), Na + -dependent transporters cannot use Li + instead of Na + ; thus, lithium tolerance is a good indicator of the presence of SNAT3 and SNAT5. However, replacement of NaCl by LiCl reduced glutamine uptake only slightly less than replacement by NMDG, suggesting that Li +tolerant transporters have a limited role in glutamine uptake. As a result glutamine transport appeared to have two broad components: a small Na + -independent and a large Na + -dependent (Li +sensitive) component. Glutamine uptake was almost completely inhibited by 10mM threonine, a typical ASCT2 substrate. Combining 10mM threonine and 10mM leucine (a typical substrate of Na + -independent system L transporters LAT1 and LAT2) reduced uptake to background levels. This profile is consistent with a contribution to glutamine transport by the Na + -independent transport system L (isoforms LAT1 and LAT2), and by the Na + -dependent transport system ASC (isoform ASCT2). Although LAT1 activity has been described in many cancer cells (24), its contribution to glutamine uptake is small, because this amino acid is not a good substrate for this transporter (15). The analysis of glutamine transport can be explained by the action of three transport systems, namely system ASC (ASCT2), and smaller contributions by system L (LAT1, LAT2) and system A (SNAT1, SNAT2). The bulk of glutamine transport is thus mediated by obligatory antiporters, similar to transport activities described before in cancer cells (25). To confirm that ASCT2 is indeed the dominant transport activity in both cell lines, we attempted to use known pharmacological ASCT2 inhibitors, such as benzylserine (BS) and L-γ-glutamyl-pnitroanilide (GPNA) (26,27) and the SNAT5 inhibitor L-γ-glutamylhydroxamate (GHX) (28).
Despite the absence of SNAT5 expression in 143B cells L-γ-glutamylhydroxamate was effective in suppressing growth of both cell lines at concentrations > 1mM; γ-glutamyl-pnitroanilide (Fig. 1F) and benzylserine (not shown) were less effective. This prompted us to investigate the specificity of these inhibitors using Xenopus laevis oocytes expressing human cRNAs of ASCT2, SNAT1, 2, 4, and 5 (Fig. 1G). Surprisingly, addition of 3mM benzylserine failed to inhibit uptake of glutamine in oocytes expressing human ASCT2. By contrast, L-γglutamyl-p-nitroanilide inhibited glutamine uptake via ASCT2, but also inhibited SNAT1, 2, 4, and 5. The compound L-γglutamylhydroxamate inhibited ASCT2, SNAT1, 4 and 5. Overall, the data suggest that these published glutamine transporter inhibitors are either ineffective or non-specific. As a result, we decided to suppress expression of ASCT2, SNAT1, SNAT2, and SNAT4 by RNAi (Fig. 2). Silencing of ASCT2 mRNA reduced the uptake activity for glutamine by about 50% (Fig. 2A) and also selectively depleted ASCT2 mRNA (Fig.  2B) and protein (Fig. 2C) (78% reduction, p=0.0004). Silencing of ASCT2 did not change the activity of the mTOR pathway in 143B cells as ascertained by phosphorylation of ribosomal protein S6 (p>0.05). It did, however, increase the abundance of phospho-eIF2α 26-fold (p=0.001), consistent with eliciting an amino acid starvation response (Fig. 2C). When cultured in the presence of 4mM glutamine, treatment of 143B cells with SNAT1, SNAT2 and SNAT4 siRNA slightly reduced transport activity without reaching significance (Fig. 2D, black bars). Silencing of SNAT1 and SNAT4 mRNA was almost complete, while silencing of SNAT2 mRNA was incomplete (about 50%, data not shown). However, the SNAT2 protein was barely detectable in the plasma membrane of cells maintained in complete media (Fig. 2E). This is consistent with the well-known induction of SNAT2 mRNA (Fig. 2F) and protein under amino acid depletion conditions (29) (Fig. 2E), restricting its expression in replete media. SNAT1 showed a smaller response under amino acid depletion, while ASCT2 expression remained constant (Fig. 2E). Consistent with an amino acid dependent regulation, the contribution of SNAT1 and SNAT2 to glutamine transport became more obvious and significant when cells were cultured in media supplemented with 2mM glutamine instead of 4mM glutamine (Fig. 2D, grey bars). Accordingly, surface expression of SNAT1 and 2 increased when glutamine supplementation was reduced (Fig. 2G). Overall, these experiments confirmed that Na +dependent uptake of glutamine (at a concentration of 100µM) is dominated by ASCT2, with smaller contributions by SNAT1 and SNAT2, depending on glutamine levels. Since silencing did not completely abolish ASCT2 protein expression in the plasma membrane, we generated an ASCT2 (-/-) line through CRISPR/Cas-9 methodology (Fig. 3A). Glutamine uptake was reduced by more than 60% consistent with the silencing experiments (Fig.  3B). The remaining glutamine uptake was sensitive to inhibition by MeAIB and was abolished by a combination of MeAIB and threonine. The amount of MeAIB-sensitive glutamine uptake increased significantly from 5.5 ± 0.5nmol/(6min x mg protein) in the parental 143B cells to 12 ± 1nmol/(6min x mg protein) in the ASCT2 (-/-) cells (p = 0.002). Silencing experiments (Fig. 3C) showed that this increase was mainly mediated by SNAT1, the transport activity of which was barely detectable in the parental cell line in complete media containing 4mM glutamine. A similar contribution of SNAT1 was observed in media supplemented with 2mM glutamine (Fig. 3C, grey bars). In agreement with a higher contribution of SNAT1, a slight increase of its protein was observed in ASCT2 (-/-) cells (Fig. 3A). Despite the strongly reduced glutamine uptake, we found that the growth rate of ASCT2 (-/-) cells was essentially unaltered compared to the parental cells in the presence of 1mM or 3mM glutamine (Fig. 3D). Consistently, net glutamine consumption was almost the same in the parental cells and in the ASCT2 (-/-) cells (Fig. 3E). This suggests that ASCT2 has an important role in avoiding an unloaded tRNA response, but is not involved in net glutamine uptake to sustain glutaminolysis. This notion was confirmed by metabolic labelling with [ 13 C5]glutamine. In both, 143B parental cells and in ASCT2 (-/-) cells, metabolites of the glutaminolysis pathway were highly enriched (Fig. 3F). A second pathway that was labelled equally strongly in both cell lines was proline biosynthesis. This pathway has recently been recognized as a critical pathway for cancer cells and embryonic stem cells (30,31) (Fig. 3F). In agreement with a significant role of SNAT1 in net glutamine uptake, silencing of SNAT1 significantly decreased net glutamine consumption in 143B cells, while silencing of ASCT2 had little effect. No synergism was observed when ASCT2 and SNAT1 were silenced together (Fig. 3G). Consistent with the silencing experiments, ASCT2 (-/-) cells showed unaltered mTOR signaling (Fig. 4A) as evidenced by phosphorylation of p70S6 kinase, ribosomal protein S6 and 4E-BP-1 (all p>0.05). Moreover, they showed a 20-fold increase (p=0.002) in eIF2α phosphorylation indicating an unloaded tRNA response (Fig. 4A). Reduced expression of ASCT2 in heterozygous cell clones was consistently accompanied by an increased signal of p-eIF2α (p=0.04). The unloaded tRNA response is known to increase SNAT2 expression (32), thereby compensating for the loss of ASCT2. We reasoned that blocking the unloaded tRNA response through silencing of GCN2 should block cell growth in the background of ASCT2 (-/-) cells. In agreement with this notion, the growth rate of GCN2-silenced cells was reduced from k = 0.035 to k=0.0018 (Fig. 4B). Parental 143B cells, by contrast, showed no difference in the growth curve when GCN2 mRNA was silenced (Fig. 4C), despite significant reduction of its mRNA (Fig. 4D). Silencing of SNAT1 alone reduced cell growth only slightly (Fig. 4E). This suggested a compensatory upregulation of SNAT2, which we observed at the protein level (Fig. 4F). As mentioned above, silencing of SNAT2 mRNA was always incomplete (about 50%). To block SNAT1 and SNAT2 activity completely, we combined silencing of SNAT1 with the system A inhibitor MeAIB, which caused selective withdrawal of SNAT2, but not SNAT1, from the plasma membrane after its expression was induced by amino acid depletion (Fig. 4G) (33). The combination of both reagents drastically reduced the growth rate of 143B cells from k=0.05 to k=0.003 (Fig. 4E). Together, these results demonstrate a crucial role of SNAT1 in supplying glutamine for glutaminolysis, with SNAT2 acting as a "backup" under conditions of amino acid depletion. Analysis of 947 cell lines (34) using Oncomine demonstrated a consistently strong expression of SNAT1, SNAT2, and ASCT2 in almost all of the analyzed cell lines (data not shown). Other glutamine transporters, such as SNAT3, SNAT4, and SNAT5, by contrast, were inconsistently expressed (data not shown). This supports a general role for SNAT1 in sustaining glutaminolysis.

DISCUSSION
Glutamine addiction has been noted in many cancer cells and has recently been recognized as a specific metabolic adaptation to the demands of a growing cell (2,3). Conversion of the canonical TCA cycle into the linear glutaminolysis pathway creates the demand for glutamine transporters as an essential element for glutamine-dependent cancer cells. It was recognized 10 years ago that many cancer cells up-regulate ASCT2 and LAT1 (13). As shown in this and in other studies, ASCT2 and LAT1 are dominantly active transporters in many cancer cell lines (25,(35)(36)(37). Accordingly, both transporters are currently targeted to reduce cancer growth (38)(39)(40). Unfortunately, many studies assume the ability of ASCT2 to mediate net glutamine uptake (see for example Fig. 7 in (10) or Fig. 1 in (11)); moreover, inhibitors, such as benzylserine, and γglutamyl-p-nitroanilide, that are often used to test involvement of ASCT2, are non-specific or ineffective. We demonstrate here that ASCT2 is important for amino acid homeostasis, but that SNAT1 and SNAT2 mediate net glutamine uptake for the glutaminolysis pathway. Whether SNAT1 and SNAT2 have additional effects as transceptors remains to be shown (41). We propose a novel view of amino acid homeostasis in cancer cells that requires three different types of transporters (Fig. 5). Type 1 are amino acid harmonizers. These are rapid exchangers for a group of amino acids such as large amino acids (LAT1, LAT2) or small amino acids (ASCT1, ASCT2). In order to maintain a harmonized mix of all 20 proteinogenic amino acids, these transporters should be faster than net uptake. Type 2 are amino acid loaders that mediate net uptake of a group of amino acids including glutamine. Loaders must have at least one overlapping amino acid substrate with the harmonizers. Glutamine and alanine are ideal substrates as they are highly abundant in blood plasma and are major substrates for SNAT1. Type 3 are rescue transporters. As shown here, SNAT2 mRNA is highly abundant, but under homeostatic amino acid levels, it is not translated into protein.
Non-harmonized amino acid mixtures will result in uncharged tRNAs, activating GCN2 (42), which in turn will cause translation of the abundant SNAT2 mRNA by a CAP-independent mechanism (32). Independently, it appears that SNAT2 protein is degraded after ubiquitination (43) and this may occur under nutrient-replete conditions, but is halted when amino acids are depleted.
ASCT2 silencing has been shown to induce rapid apoptosis in hepatoma cells (44) and to reduce growth of melanoma and pancreatic cancer cells (39,40). As shown here, an ASCT2-deficient cancer cell line can display normal growth, due to upregulation of SNAT1 and SNAT2, which have overlapping substrate specificity with ASCT2. One of the differences may be that cells that are sensitive to ASCT2 inhibition, show reduced mTOR signaling when ASCT2 is silenced (12). This has been proposed to involve the amino acid exchanger LAT1, exchanging intracellular glutamine imported by ASCT2 against branchedchain amino acids, which in turn would activate mTOR (10). One of the problems with this proposal is that ASCT2 also requires an exchange substrate. In addition, LAT1 has limited capacity to use intracellular glutamine as an exchange substrate (15). Activation of mTOR is mediated by cytosolic and lysosomal leucine levels and lysosomal amino acid transporters (21,22). SNAT1 and SNAT2 proficiently provide exchange substrates to LAT1, thereby explaining the normal mTOR signaling in ASCT2 (-/-) cells. Targeting SNAT1 could have potential for the treatment of clear cell renal cell carcinoma (CCRCC). Analysis of four different sporadic CCRCC sets (45)(46)(47)(48) using Oncomine reveals 3-5-fold higher expression of SNAT1 in tumor tissue compared to matched normal tissue (Fig.  6A). The vast majority of CCRCC are caused by biallelic mutation or inactivation of the von Hippel-Lindau tumor suppressor (VHL) (49). The VHL protein is an adapter protein that interacts with a number of proteins, most notably targeting HIF1α for degradation by the proteasome. Inactivation of VHL thus increases HIF1α expression. Analysis of HIF1α target genes in CCRCC using the same datasets but also including hereditary CCRCC shows SNAT1 (SLC38A1) among well-known targets such as the monocarboxylate transporter 4 (SLC16A3), lactate dehydrogenase A (LDH A), and hexokinase 2 (HK2) (Fig. 6B). SNAT1 expression is not only driven by HIF1α, but also by other transcription factors. The ENCODE database (50) shows a strong signature of myc binding and RNA polymerase 2 (Pol 2) around the promoter of the SNAT1 gene Slc38a1 in in K562 and HeLa cells (data not shown). Moreover, hepatocyte growth factor receptor is overexpressed in many cancers and also appears to up-regulate SNAT1 gene expression through its MAP-kinase signaling pathway (51) (Fig. 6C). This suggests that SNAT1 transport activity is an important function for many fast growing cancer cells.  Amino acid symporters and exchangers work in combination to ensure a homeostatic intracellular mixture of all neutral amino acids. SNAT1 is a major net importer (loader) of neutral amino acids including glutamine as a substrate for glutaminolysis. Amino acid exchangers (harmonizers) quickly harmonize amino acid composition in times of imbalanced amino acid usage. An imbalance of the amino acid composition results in an amino acid starvation response, which increases SNAT2 expression (rescue).