The amino acid transporter SLC7A5 confers a poor prognosis in the highly proliferative breast cancer subtypes and is a key therapeutic target in luminal B tumours

Breast cancer (BC) is a heterogeneous disease characterised by variant biology and patient outcome. The amino acid transporter, SLC7A5, plays a role in BC although its impact on patient outcome in different BC subtypes remains to be validated. This study aimed to determine whether the clinicopathological and prognostic value of SLC7A5 is different within the molecular classes of BC. SLC7A5 was assessed at the genomic level, using Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) data (n = 1980), and proteomic level, using immunohistochemical analysis and tissue microarray (TMA) (n = 2664; 1110 training and 1554 validation sets) in well-characterised primary BC cohorts. SLC7A5 expression correlated with clinicopathological and biological parameters, molecular subtypes and patient outcome. SLC7A5 mRNA and protein expression were strongly correlated with larger tumour size and higher grade. High expression was observed in triple negative (TN), human epidermal growth factor receptor 2 (HER2)+, and luminal B subtypes. SLC7A5 mRNA and protein expression was significantly associated with the expression of the key regulator of tumour cell metabolism, c-MYC, specifically in luminal B tumours only (p = 0.001). High expression of SLC7A5 mRNA and protein was associated with poor patient outcome (p < 0.001) but only in the highly proliferative oestrogen receptor (ER)+/ luminal B (p = 0.007) and HER2+ classes of BC (p = 0.03). In multivariate analysis, SLC7A5 protein was an independent risk factor for shorter breast-cancer-specific survival only in ER+ high-proliferation tumours (p = 0.02). SLC7A5 appears to play a role in the aggressive highly proliferative ER+ subtype driven by MYC and could act as a potential therapeutic target. Functional assessment is necessary to reveal the specific role played by this transporter in the ER+ highly proliferative subclass and HER2+ subclass of BC.


Background
Altered metabolic pathways have been readily accepted as part of the revised hallmarks of cancer where cancer cells are able to regulate their metabolism to provide energy and cellular building blocks required for their unremitting proliferation [1]. Many cancer cells are highly reliant on amino acids for their growth, not only because they are a nitrogen donor for the synthesis of nucleotides and other amino acids, but also because they activate mammalian target of rapamycin complex1 (mTORC1) through nutrient signalling pathways which in turn regulates protein translation and cell growth [2,3]. There is also increasing evidence that oncogenes and/or tumour-suppressor genes can reprogramme tumour cell metabolism including the direct regulation of the amino acid transporter, solute carrier family 7 member 5 (SLC7A5), by the oncogene MYC [4,5] and the regulation of expression of the glutamine transporter, SLC1A5, by the tumour suppressor retinoblastoma (Rb) [6].
SLC7A5 is a sodium-independent transporter and acts as an amino acid exchanger by transporting large neutral amino acids such as leucine, phenylalanine and tryptophan by exchange with intracellular glutamine [7]. It therefore functions in supplying amino acids to cancer cells and maintaining intra-cellular leucine, which is considered a master regulator of the mTORC1 signalling pathway [8][9][10]. For functional expression on the plasma membrane, SLC7A5 must heterodimerise with the heavy chain of SLC3A2 [7,11].
It has been reported that SLC7A5 is highly expressed in a variety of cancers including oesophageal carcinoma [12], oral cancer [13] and lung adenocarcinoma [14]. SLC7A5 is co-expressed with the glutamine transporter, SLC1A5, in many cancers suggesting a functional coupling of these transporters in supporting tumour progression [15]. In this study, we aimed to assess SLC7A5 gene copy number and mRNA expression, alongside protein expression in large and well-characterised annotated cohorts of BC to determine its biological, clinicopathological and prognostic value in the different BC molecular classes with particular interest in the highly proliferative aggressive subgroups.

SLC7A5 copy number and gene expression
A cohort of 1980 BC tumours in the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) [16] was used to evaluate SLC7A5 gene copy number aberrations (CNA) and gene expression. DNA/RNA was isolated from fresh frozen samples and genomic and transcriptional profiling was obtained using the Affymetrix SNP 6.0 and Illumina HT-12v3 platforms respectively. CNA were considered at the gene level by segments and the Šidák correction [17] was applied for multiple testing. Gene expression data were pre-processed and normalised as described previously [16]. In this cohort, patients included were oestrogen receptor (ER)-positive (ER+) and/or lymph node (LN)-negative (LN-) and did not receive adjuvant chemotherapy, whereas ER-and LN+ patients received adjuvant treatment. X-tile (version 3.6.1, Yale University, USA) was applied to dichotomise SLC7A5 mRNA expression, based on prediction of breast-cancerspecific survival (BCSS). The association between the SLC7A5 mRNA expression and clinicopathological parameters, molecular subtypes, and patient outcome was investigated. The online dataset, Breast Cancer Gene Expression Miner v4.0 (http://bcgenex.centregauducheau.fr) and breast cancer data from The Cancer Genome Atlas (TCGA) [18] were used for external validation of SLC7A5 copy number/or mRNA expression.

Patients and tumours
This study evaluated well-characterised cohorts of patients with early-stage primary operable invasive BC, who presented aged ≤70 years. Patients in the discovery set (n = 1110) presented at Nottingham City Hospital between 1989 and 1998, while the validation set (n = 1554) includes patients who were presented between 1998 and 2006. Patient management was uniform and based on tumour characteristics by Nottingham Prognostic Index (NPI) and hormone receptor status. Patients within the NPI excellent prognostic group (score ≤3.4) received no adjuvant therapy, but those patients with NPI >3.4 received tamoxifen if ER-positive (± goserelin (Zoladex) in case the patients were premenopausal). Conversely, classical cyclophosphamide, methotrexate and 5-flurouracil (CMF) were used if the patients were ER-negative and fit enough to receive chemotherapy. None of the patients in this study received neoadjuvant therapy. Clinical history, tumour characteristics and information on therapy and outcomes are prospectively maintained. Outcome data included development and time to distant metastasis (DM) and breast-cancerspecific survival (BCSS). There was no difference in the distribution of clinicopathological parameters between the discovery and validation cohorts or the METABRIC series of patients (all correlation coefficients ≥0.91, all p < 0.0001) (Additional file 1: Table S1).

Western blotting
The antibody specificity of anti-SLC7A5 (EPR17573, Abcam, UK) was validated using western blotting in human embryonic kidney (HEK) 293 T over expression lysate (Origene Technologies, Rockville, MD, USA) and MDA-MB-175 (luminal B-like), T47D and MCF7 (luminal A) [19] breast cancer lysate (American Type Culture Collection; Rockville, MD, USA). A dilution of 1:200 of the primary antibody and 1:1000 HRPconjugated (Dako) secondary antibodies were applied: 5% milk /PBS-Tween (0.1%) (Marvel Original Dried Skimmed Milk, Premier Food Groups Ltd., UK) was used for blocking. Mouse monoclonal anti-β-actin primary antibody was used as a marker of endogenously expressed control. SLC7A5 bands were visualised using enhanced chemiluminescence (ECL) showing a single specific band at the correct predicted size (40 kDa) for the SLC7A5 protein.

Tissue arrays and immunohistochemical analysis
The discovery set (n = 1110) were arrayed as previously described using a single 0.6-mm core sampled from the periphery of each invasive tumour [20]. The validation set (n = 1554) were similarly arrayed using a tissue microarray (TMA) GrandMaster (3D Histech). Immunohistochemical (IHC) staining was performed on 4-μm TMA sections using the Novolink polymer detection system (Leica Biosystems, RE7150-K). Briefly, tissue slides were deparaffinised with xylene and rehydrated through three changes of alcohol. Heat-induced antigen epitope retrieval was performed in citrate buffer (pH 6.0) for 20 min using a microwave oven (Whirpool JT359 Jet Chef 1000 W). Endogenous peroxidase activity was blocked by peroxidase block for 5 min. Slides were washed with Tris-buffered saline (TBS, pH 7.6), followed by application of protein block for 5 min. Following another TBS wash, sections were incubated with the primary SLC7A5 antibody diluted at 1:50 in Leica antibody diluent (RE AR9352, Lieca, Biosysytems, UK) overnight at 4°C. Slides were washed with TBS followed by incubation with post primary block for 30 min followed by a TBS wash. Novolink polymer was applied for 30 min: 3,3′-diaminobenzidine (DAB) chromogen was applied for 5 min. Slides were counterstained with Novolink haematoxylin for 6 min, dehydrated and coverslipped.
Stained TMA sections were scored using high resolution digital images (NanoZoomer; Hamamatsu Photonics, Welwyn Garden City, UK), at × 20 magnification. Evaluation of staining for SLC7A5 was based on a semi-quantitative assessment of digital images of the cores using a modified histochemical score (Hscore) which includes an assessment of both the intensity and the percentage of stained cells [21]. Staining intensity was assessed as follows: 0, negative; 1, weak; 2, medium; 3, strong, and the percentage of the positively stained tumour cells was estimated subjectively. The final H-score was calculated by multiplying the percentage of positive cells (0-100) by the intensity (0-3), producing a total range of 0-300. Dichotomisation of protein expression in predicting BCSS was determined using x-tile software.

Statistical analysis
Statistical analysis was performed using SPSS 22.0 statistical software (SPSS Inc., Chicago, IL, USA). Spearman's correlation coefficient was calcualted to examine the association between continuous variables. The chi-square test was performed to analyse relationships between categorical variables. For the continuous variables, differences between three or more groups were assessed using one-way analysis of variance (ANOVA) with the post-hoc Tukey multiple comparison test (for normally distributed data) or Kruskal-Wallis test (for non-normal distribution). Differences between two groups were assessed using the t test (normally distributed data) or Mann-Whitney test (non-normal distribution). Survival curves were analysed by the Kaplan-Meier and log rank test. Cox's proportional hazard method was performed for multivariate analysis to identify the independent prognostic/predictive factors. The p values were adjusted for multiple testing using the Bonferroni correction. A p value ˂0.05 was considered significant. The study endpoints were 10-year BCSS or distant metastasis-free survival (DMFS).

SLC7A5 genomic profiling
SLC7A5 was amplified in 0.3% and 0.6% of BC cases in the METABRIC and TCGA datasets, respectively, while deletion (deep and shallow) was detected in 56% and 68% of cases in the same cohorts respectively. Point mutations in SLC7A5 were extremely rare, where TCGA data reported just one case with a missense mutation (Additional file 2: Figure S4A) [33,34]. SLC7A5 is situated on chromosome 16 (16q24.2); all the annotated genes, which were located on 16q [35] were selected to determine their CNV in relation to SLC7A5 and assess whether these aberrations were locus-specific or involved large chromosomal segments. There was significant positive correlation between SLC7A5 deletion and the deletion of all genes (p < 0.001, Additional file 3: Table S2) in both the METABRIC and TCGA data. However, amplification of three genes (FANCA, CBFA2T3 and CDT1) showed significant association with the amplified SLC7A5 (p ≤ 0.03, data not shown) in the afore-mentioned datasets together.

SLC7A5 expression in breast cancer
SLC7A5 protein expression was observed, predominantly in the membrane of invasive breast cancer cells, with expression levels varying from absent to high ( Fig. 1b and c). The distribution of the SLC7A5 protein expression was unimodal and left-skewed. The SLC7A5 mRNA expression had a normal distribution. Expression of SLC7A5 mRNA and protein were dichotomised using cut points derived from prediction of patient survival using X-tile (https://medicine.yale.edu/lab/rimm/ research/software.aspx; Yale University). Positive  Fig. 2).  PAM50 prediction analysis of microarray, HER2 human epidermal growth factor receptor 2 p value in bold in these tables means statistically significant associations High SLC7A5 mRNA expression was significantly associated with larger tumour size (Fig. 3a, p < 0.001), higher grade (Fig. 3b, p < 0.001), positive nodal metastasis (Fig. 3c, p< 0.001) and poor NPI (Fig. 3d, p < 0.001). Both SLC7A5 mRNA and SLC7A5 protein were associated with medullary-like tumours. Where data were available, these associations were confirmed using the Breast Cancer Gene-Expression Miner v4.0 (Additional file 4: Figure S1A, B) and the TCGA data (Additional file 2: Figure S4B). In addition SLC7A5 copy number loss was significantly associated with good prognostic parameters including, lower grade and good NPI ( Table 2, p < 0.001). There was positive association between SLC7A5 copy number gain and MYC gain (p < 0.001, Table 2).

SLC7A5 expression in molecular BC subtypes
SLC7A5 protein expression was associated with negative hormone receptor status and HER2+ tumours (all p ≤ 0.002, Table 3) and it was highly expressed in triple negative (TN) and basal-like phenotype malignancies compared to non-TN and non-basal-like tumours (p < 0.001, Table 3). Similarly, high expression of SLC7A5 mRNA was significantly associated with hormone receptor negative (ER-and PgR-) and HER2+ tumours (all p < 0.001, Table 4). These results were in concordance with the findings of the Breast Cancer Gene-Expression Miner v4.0 (Additional file 4: Figure S1C-F) and TCGA data analysis (Additional file 2: Figure  S4C-E).
At the gene level, there was a greater copy number gain of SLC7A5 (p < 0.001, Table 2 in the basal-like ER oestrogen receptor, PgR progesterone receptor, HER2 human epidermal growth factor receptor, NA not applicable p value in bold in these tables means statistically significant associations subtype, while SCL7A5 copy number loss was mainly observed in the luminal A subtype (p < 0.001, Table 2).

SLC7A5 expression and other associated markers
Correlation between SLC7A5 mRNA and associated genes was investigated in the METABRIC dataset ( Table 5). The genes were selected based on previous publications, and were either regulatory genes or others that share or support the SLC7A5 biological function, which focused mainly on glutamine transport and glutamine metabolism [2,5,15,[37][38][39][40][41]. There was positive correlation between SLC7A5 mRNA expression and the expression of regulatory genes, several amino acid transporters and genes involved in the glutamine-proline regulatory axis. There was a positive relationship between SLC7A5 and MYC, mTOR and ATF4 (p < 0.001) and the positive relationship between MYC, HIF2A and SLC7A5 was only observed in luminal B tumours (p = 0.01 and p < 0.001, respectively). High SLC7A5 mRNA expression was specifically associated with the enzymes involved with conversion of glutamine (Gln) to proline, where PYCR1 and ALDH18A1 showed a positive relationship with SLC7A5 in almost all subtypes (p < 0.02).
The majority of glutamine transporters were significantly associated with SLC7A5 expression primarily in triple negative tumours and to a lesser extent luminal and HER2+ subtypes. SLC7A5 was significantly correlated with SLC1A5 in all subtypes (p < 0.02).
TP53 mutations were also highly prevalent in breast tumours where there was high SLC7A5 mRNA expression (p < 0.001, Tables 3 and 4). Moreover, high SLC7A5 protein was positively associated with high p53 protein (p < 0.001).
SLC7A5 protein expression was significantly expressed in breast tumours with high Ki67, and the upstream effector MYC (p < 0.001, Table 6). SLC1A5, GLS, PYCR1 and PIK3CA were significantly expressed in breast tumours with high expression of SLC7A5 (p < 0.001), while the low expression of SLC7A5 was associated with high levels of p-mTORC1 (p < 0.001) ( Table 6).

SLC7A5 expression and patient outcome
Both high SLC7A5 mRNA (p < 0.001) (Fig. 4a) and protein (p < 0.001) expression were associated with poor BCSS in the discovery and validation sets (Fig. 5a, b). This association was anticipated as the cutoff was based on the prediction of BCSS.
While SLC7A5 mRNA expression was not predictive of BCSS in any specific molecular class (Fig. 4be), high expression of SLC7A5 protein was only predictive of shorter BCSS in ER+ high proliferation (p = 0.007, Fig. 5d) and HER2+ tumours (p = 0.03, Fig. 5f ).   There was no association between SLC7A5 protein and outcome in ER+ low proliferation (Fig. 5c) or in TNBC (Fig. 5e). In multivariate Cox regression analysis, SLC7A5 mRNA was a predictor of shorter BCSS independent of tumour size, grade or lymph node stage (p = 0.006, Additional file 5: Table S3) but not in any specific subtype. However, SLC7A5 protein was significant only in the ER+ high-proliferation tumours (p = 0.02, Table 7) and not in any other subtypes (data not shown). Likewise, high SLC7A5 protein expression was associated with shorter distant metastases-free survival (DMFS) (p < 0.001; Additional file 6: Figure S2A, B) within the ER+ high-proliferation class (p = 0.03, Additional file 6: Figure S2D) but not in other subtypes (Additional file 6: Figure S2C, E, F). This association was identified in the discovery set and validated in the validation set. The relationship between high SLC7A5 mRNA expression and poor patient outcome in ER+ disease, but not ER-disease, was confirmed using Breast Cancer Gene-Expression Miner (Additional file 7: Figure S3A, B, C).

Discussion
Breast cancer is a heterogeneous disease with various subtypes [42] differing in terms of morphology, molecular and biological profiles, response to therapy and clinical behaviour. In addition, different subtypes exhibit disparity in their metabolic pathways and their nutritional needs. The most common form of BC (~55-80%) is the ER+/luminal tumour [43,44], and tumours that belong to this class are also variable in terms of recurrence, mortality rates and disease prognosis [43]. Therefore, understanding the biology of BC and exploring the metabolic pathways could help to identify potential novel therapeutic targets.
Cancer cells must alter their metabolism in order to satisfy the demands of necessary energy and cellular building blocks. It is widely known that amino acid transport systems play a principal role in sustaining the proliferation of cancer cells by supplying the required amino acids for protein synthesis and by activation of nutrient signalling through the mTORC1 complex. This study has revealed for the first time that SLC7A5 is a key amino acid transporter in the more aggressive and highly proliferative ER+ tumours.
SLC7A5 is located in 16q24.2. This study showed that SLC7A5 deletion, but not amplification, was significantly correlated with all the annotated genes located in the same chromosomal region, indicating that the deletion was not locus-specific. Interestingly, Ecadherin (CDH1), which was located in 16q22.1, was also implicated. It is widely known that most lobular tumours harbour loss of heterozygosity (LOH) at chromosome 16 and are missing the wild type CDH1 allele [45]. In this study, approximately 40% of METABRIC cases with SLC7A5 loss were associated with invasive lobular histology. In addition, SLC7A5 protein expression in lobular carcinoma has a relatively lower mean rank value compared to the other histological subtypes, confirming that deletions involve large segments of q16, which can reflect the BC phenotype.  p value in bold in these tables means statistically significant associations SLC7A5 is widely expressed in many human cancers and various cancer cell lines [46]. The current study included two large discovery and validation cohorts to confirm the significant association between the high SLC7A5 protein expression and the poor prognostic clinico-pathological parameters, including larger tumour size, higher grade and poor NPI. Furthermore, high SLC7A5 expression was significantly associated with higher expression of the proliferative marker (Ki67). This supports the results of previous studies which demonstrated that these two biomarkers are significantly correlated in tongue cancer [47], neuroendocrine carcinoma of the lung [48], thymic carcinoma [49] and breast cancer [50], indicating that SLC7A5 is critical for proliferation in cancer cells.
With respect to the breast cancer ER+ subtypes, SLC7A5 expression was lower in ER+ tumours that have low proliferation (luminal A subtype) compared with the highly proliferative ER+ (luminal B) malignancies, and it was primarily associated with poor patient outcome and shorter DMFS in the latter class only. This is most likely due to their heavier energy and nutrient requirements for cell survival, proliferation and metastasis. This was anticipated, as it has been shown that over expression of SLC7A5 is actively involved in the proliferation of vascular smooth muscle cells [51] and it is co-expressed with vascular endothelial growth factor (VEGF) in the primary and metastatic sites of many cancers [37], which may be implicated for the metastatic process. In this study the most significant positive correlation between mRNA expression of SLC7A5 and VEGFA was identified in the luminal B subtype. In this regard, Bartlett  SLC7A5 mRNA and protein was also highly expressed in TNBC and HER2+ BC, in concordance with Furuya et al. [50]. However, in these subtypes the significant association between SLC7A5 protein expression and patient outcome was only observed in the HER2+ tumours. Among all BC subtypes, SLC7A5 protein expression was an independent predictor of short BCSS in ER+ high-proliferation tumours only. In this regard, the larger sample of ER+ high-proliferation cases might reflect the stronger significance compared with the smaller sample of HER2+ and TNBC tumours. We therefore suggest that further confirmation in larger cohorts of HER2+ and TN tumours is warranted.
Previous studies have shown regulation of SLC7A5 by other proteins including the tumour oncogene Myc, which induces SLC7A5 [4,5]. In the current study, the relationship between SLC7A5 and other regulatory proteins in both mRNA and protein expression was investigated. A positive relationship was observed between SLC7A5 and Myc in both protein and mRNA levels, and this correlation remained significant only in luminal B subtype, when different subtypes were investigated. ATF4-dependent transcripts also encode for SLC7A5 upon amino acid deprivation [39] and in this study there was positive correlation between ATF4 and SLC7A5 gene expression, in line with expectations. A previous study showed that activation of the HIF2α pathway increases mTORC1 activity by upregulating expression of the amino acid carrier SLC7A5 [38] and the current study confirmed the positive correlation between HIF2α and SLC7A5, which was only observed in luminal B tumours. SLC7A5 functions by importing essential amino acids to cancer cells and research has detailed the role of amino acids, particularly leucine, in activating mTORC1, which in turn controls protein translation and cell proliferation, and prevents apoptosis in malignant cells [2,3]. This study showed positive correlation between SLC7A5 and mTOR at the mRNA level. However, there were conflicting results in the analysis of protein levels of SLC7A5, whereby high SLC7A5 expression was associated with lower expression of the mTORC1 phosphorylated at ser (2448), which was included in this study.
This was unsurprising as Cheng et al. confirmed that phosphorylation of mTORC1 at ser (2448), which is stimulated by growth factors, was mutually exclusive with mTORC1 phosphorylated at thr (2446), which is regulated by amino acids [53]. These observations may explain why SLC7A5 over expression is primarily associated with poor outcome only in the high proliferation ER+ tumours.
This study further investigated the association of SLC7A5 expression with other glutamine transporters, in which some variability in the expression of amino acid transporters across molecular subtypes was observed. For example, the TN subtype was the main class associated with the transporters required for glutamine influx, perhaps because it depends on delivery of glutamine instead of synthesis. In contrast, positive correlation between SLC7A5 and the glutamine synthase enzyme GLUL was observed in luminal A tumours, suggesting that this subtype might rely on glutamine neosynthesis rather than uptake. SLC1A5 functionally couples with SLC7A5 to allow the cellular influx and efflux of glutamine, as SLC1A5 mediates uptake of glutamine, while SLC7A5 uses intracellular glutamine concentrations to adjust the essential amino acid cytoplasmic pool for metabolic demands and signalling to mTORC1 [15]. Here we observed that SLC7A5 and SLC1A5 are positively correlated in all the BC subtypes.
Previous studies have raised awareness and revealed the importance of the proline-glutamine (Pro-Gln) regulatory axis in BC. SLC7A5 appears to have a pivotal role in this regulatory axis, as its expression was highly associated with the enzymes that mediate glutamate degradation to form the amino acid proline, which has been shown to play a role in assisting tumour growth by different mechanisms [54].
Blocking of SLC7A5 using its inhibitor, 2aminobicyclo-(2,2,1)-heptane-2-carboxylic acid (BCH), efficiently decreased colony formation of MDA-MB-231 TNBC cells [55]. Even though the consequences of blocking SLC7A5 in the highly proliferative ER+ tumours remain undetermined, this study suggests that SLC7A5 can potentially be used as a therapeutic target for luminal B BC.