Induction of CTH expression in response to amino acid starvation confers resistance to anti-LAT1 therapy in MDA-MB-231 cells

L type amino acid transporter 1 (LAT1) is an attractive molecular target for cancer therapy because of its overexpression in many cancer cells. JPH203, a selective LAT1 inhibitor, causes amino acid deprivation and suppresses cancer cell proliferation. However, several cancer cells showed resistance to amino acid deprivation. In this study, we aimed to elucidate the molecular mechanism of different sensitivity between 2 breast cancer cells to anti-LAT1 therapy. MDA-MB-231 cells were more resistant to growth suppression effect of JPH203 than T-47D cells (IC50 was 200 ± 12.5 μM for MDA-MB-231, and 5 ± 1.1 μM for T-47D cells; p < 0.05). Transcriptome and biochemical analysis were done in these cells in the presence/absence of JPH203. JPH203 induced intracellular amino acid deprivation stress in both cells, but it upregulated cystathionine γ lyase (CTH), an enzyme for synthesis of antioxidants, only in MDA-MB-231 cells. Moreover, siRNA-mediated CTH knockdown induced oxidative stress in response to JPH203 leading to decreased cell viability in MDA-MB-231 cells. These results suggest that activation of anti-oxidation pathways in response to amino acid deprivation confers resistance to anti-LAT1 therapy.

L-type amino acid transporter (LAT) family are membrane proteins that transport large neutral amino acids across the cell membrane, which consist of four members (LAT1, LAT2, LAT3 and LAT4) [1][2][3][4][5] . Among them, LAT1 which forms a functional heterodimer with 4F2 heavy chain (4F2hc) has been considered to be a promising target for cancer therapy 5,6 . Its expression levels are high in the fetal tissue and restricted to the barrier such as blood brain barrier or cells with rapid proliferation such as bone marrow in adults 7 . Many tumor cells express LAT1 and its levels of expression are correlated with the aggressiveness of the malignancy 8,9 . In adult non-tumor cells, large neutral amino acids are transported by LAT2 10,11 . This tumor selective expression pattern together with novel mode of action, i.e. restricting amino acids delivery specifically for cancer cells, attracted the attention of many investigators. Endou et al. developed JPH203, a specific competitive inhibitor for LAT1, which is now in clinical trial in Japan 12 .
Restricting essential amino acids uptake into the tumor cell suppresses its proliferation via inhibition of mammalian target of rapamycin (mTOR)-p70S6 Kinase (p70S6K) pathway [13][14][15] . If we assume that the sensitivity of mTOR to essential amino acids are similar across the cells, the potency of JPH203 for a particular cell would be predicted by the expression levels of LAT1. However, the potency of JPH203 on growth suppression was reported to be different among various cancer cell lines that express similar levels of LAT1 16 .
In addition to suppressing mTOR pathway, genetic inhibition or JPH203 treatment reported to activate amino acid stress response pathway, which consists of general control nonderepressible 2 (GCN2)-eukaryotic Initiation Factor 2 α (eIF2α)-activating transcription factor 4 (ATF4) 17 . When cells are depleted of amino acids, GCN2 is activated by phosphorylation and subsequently phosphorylates eIF2α, which leads to inhibition in general protein translation. Phosphorylated eIF2α promotes ATF4 translation in the face of inhibition of general protein translation 18 . Upregulation of ATF4 may facilitate apoptosis of cancer cells in some circumstances 19 , while it may also promote cancer cell survival and tumorigenesis 20 . One of the downstream molecules regulated by ATF4  www.nature.com/scientificreports/ cells (Fig. 1G). In addition, a molecule in the amino acid nutritional stress pathway CHAC1 22 , which degrades antioxidant GSH, was upregulated in response to JPH203 in both cells ( Supplementary Fig. 1), suggesting that some of the reactions to amino acids starvation were commonly observed in both of these cells.

JPH203 treatment induced CTH in MDA-MB-231 cells but not in T-47D cells. Consistent with
the relative resistance to JPH203 treatment, the transcriptome of MDA-MB-231 cells was not much affected by JPH203 ( Fig. 2A); only 6 genes were differentially expressed in the presence of JPH203 (Table 1). On the other hand, 516 genes were differentially expressed in T-47D cells ( Fig. 2A). Of these 6 genes upregulated by JPH203 treatment in MDA-MB-231 cells, 2 genes, cystathionine γ lyase (CTH) and nicotinamide N-methyltransferase (NNMT) were upregulated only in MDA-MB-231 cells (Table 1). Interestingly, both of these enzymes act in an anti-oxidative pathway that generates cysteine from methionine [23][24][25] . After confirming that CTH mRNA was upregulated in response to JPH203 only in MDA-MB-231 cells by qPCR (Fig. 2B), we decided to focus on CTH, a more downstream enzyme in the process.  www.nature.com/scientificreports/ CTH was likely to be induced through GCN2-ATF4 pathway in MDA-MB-231 cells. In order to elucidate the pathway leading to CTH upregulation, differentially expressed genes in T-47D cells in response to JPH203 treatment were analyzed. Amino acid restriction was known to induce cellular stress response known as GCN2-eIF2α-ATF4 pathway and genes related to this pathway were overrepresented among upregulated genes in T-47D cells in response to JPH203 ( Table 2) 17,26,27 . Although ATF4 was not differentially expressed in MDA-MB-231 cells in microarray experiments, it did show upregulation in response to JPH203 treatment in these cells by qPCR (Fig. 3A). Consistent with the result of transcriptome analysis, ATF4 gene was also upregulated in T-47D cells in response to JPH203 by qPCR (Fig. 3A). Because activities of GCN2 and eIF2α were regulated by phosphorylation rather than transcription, their protein level expression and phosphorylation levels were examined. Both in MDA-MB-231 cells and T-47D cells, JPH203 treatment increased phosphorylation of GCN2 and eIF2α without apparent changes in total protein expression levels (Fig. 3B). As expected from qPCR data, ATF4 protein expression was increased in both of these cells by JPH203, but CTH protein expression was increased only in MDA-MB-231 cells in response to JPH203 (Fig. 3B). This increased expression of CTH was abolished by knocking down ATF4 (Fig. 3C), suggesting that ATF4 is at least one of the regulators for CTH expression.
To confirm that CTH is induced by amino acid depletion but not by unknown action of JPH203 per se, both cells were incubated in the medium depleted of LAT1 substrate amino acids, which are essential amino acids and tyrosine. As shown in Fig. 3D, CTH expression was higher at baseline and increased by the treatment much stronger in MDA-MB-231 cells than in T-47D cells. Along the same line, more than 100 μM of JPH203 did not increase CTH expression in T-47D cells while those doses of JPH203 clearly upregulate CTH expression in MDA-MB-231 cells (Fig. 3E). If there were unknown side effect of JPH203 that stimulates CTH expression, T-47D cells would have increased CTH in response to higher doses of JPH203. Although this is not a direct proof, it is very difficult to argue that the observed upregulation of CTH was due to some unknown side effect (i.e. not mediated by amino acid depletion) of JPH203.

JPH203-mediated CTH induction mitigated reactive oxygen species (ROS) release caused by amino acid starvation stress.
We next sought to examine whether JPH203-mediated CTH induction attenuated ROS levels derived from amino acid starvation stress. To that end, we employed cell-permeable DCFH-DA, which is a fluorogenic dye to indicate cellular ROS levels, in JPH203-treated cells. ROS production was significantly increased by 2.5-fold in JPH203-treated T-47D cells, while it was not altered in JPH203-treated MDA-MB-231 cells (Fig. 5A,B). Exogenous H 2 O 2 led to increased ROS levels by 4.5-fold and by 4.2-fold compared with control in both MDA-MB-231 and T-47D cells, respectively.

CTH overexpression supported cell survival under JPH203-induced amino acid starvation in T-47D cells.
To investigate whether CTH overexpression could attenuate an anti-proliferative effect by JPH203, T-47D cells were transiently transfected with pcDNA3.1-CTH plasmid (Fig. 6A). In its usual culture   and T-47D cells were incubated for 72 h in amino acid restriction media, the cell lysates were subjected to western blot analysis and probed for CTH and β-actin (loading control). LAT1 substrate amino acids, which are eight essential amino acids (EAA; l-isoleucine, l-phenylalanine, l-tryptophan, l-valine, l-histidine, l-leucine and l-methionine) and l-tyrosine, were removed from normal RPMI (1.0) to 50% (0.5), or free (0). (E) MDA-MB-231 and T-47D cells were treated with JPH203 at different concentrations (0, 0.05, 5, 100 or 200 µM) for 48 h. The cell lysates were subjected to western blot analysis for determining CTH protein levels. www.nature.com/scientificreports/ CTH overexpressed cells by 25% or by 66% respectively (Fig. 6B,C). CTH overexpression significantly increased T-47D cell viability by 13% in JPH203 and by 8% in H 2 O 2 , respectively (Fig. 6D). Collectively, these findings suggested that increased levels of CTH could reduce ROS levels caused by amino acid deprivation stress and give the cell an ability to survive such a stress.

MDA-MB-231 cells were primed for anti-oxidant production at baseline.
Noting that expression levels of CTH was higher in MDA-MB-231 cells than in T-47D cells at baseline (Fig. 3B,D), we looked at microarray data for anti-oxidant related molecules. Not only CTH, but also cystine/glutamic acid transporter (xCT) was expressed more abundantly in MDA-MB-231 cells compared with T-47D cells (Fig. 7). These results suggest that machinery for producing antioxidant GSH and cysteine was more active even at baseline in MDA-MB-231 cells.

Discussion
In this study, we compared 2 breast cancer cell lines, MDA-MB-231 and T-47D cells. Leucine uptake was almost exclusively mediated by LAT1 in both cells and JPH203 inhibited leucine uptake of these cells at IC50 uptake of 0.06 ± 0.02 μM for MDA-MB-231 cells and 0.03 ± 0.01 μM for T-47D cells, respectively (p = 0.07). However, MDA-MB-231 cells exhibited "resistance" to anti-proliferative effect of JPH203; IC50 proliferation of 200 ± 12.5 μM for MDA-MB-231 cells and 5 ± 1.1 μM for T-47D cells, respectively (p < 0.05). The discrepancy in IC50 uptake and IC50 proliferation for JPH203 has been reported previously, where JPH203's IC50 uptake = 0.06 μM and IC50 proliferation = 4.1 μM in colon cancer cell line HT29 cells. They reasoned that there exist much more amino acids in the proliferation assay than in the uptake assay 12 . In addition, the time course for leucine uptake was 2 min while the cell proliferation assay took 4 days in our study. Because JPH203 was not an irreversible inhibitor, it is impossible to completely block substrates uptake all through a long incubation period, especially in the media rich in amino acids and in the presence of other leucine transporters such as LAT3 albeit small amount comparing to LAT1. In line with this, Cormerais et al. found that 90% reduction of LAT1 expression by itself does not lead to growth suppression, suggesting that more than IC90 uptake of JPH203 (in our case more than 1 μM in both cells; Fig. 1E) would require to attain significant growth suppression 17 . The proliferation of T-47D cells  www.nature.com/scientificreports/ was effectively suppressed by 5 μM of JPH203, but that of MDA-MB-231 cells was not (Fig. 1F). Upon obtaining these data, we decided to clarify the reason why the proliferation of MDA-MB-231 cells was not suppressed to the similar degree as T-47D cells in the presence of more than IC50 uptake concentrations (i.e. more than 1 μM). The difference in the anti-proliferative effect of JPH203 among cancer cell lines has been reported elsewhere, but no underlying mechanism has been identified to date to the best of our knowledge 16,[28][29][30] . The fact that mTOR-p70S6K pathway was equally suppressed and that a key cellular stress related transcription factor, ATF4, was upregulated in both cells indicates that JPH203 efficiently inhibited essential amino acids uptake into the cells and it induced cellular nutritional stress response to a similar degree between MDA-MB-231 and T-47D cells. Following ATF4 upregulation, these 2 cells upregulated different sets of downstream molecules; anti-oxidant molecules were increased only in MDA-MB-231 cells (Fig. 8). CTH is an enzyme which transforms cystathionine derived from methionine into cysteine and contributes to produce antioxidants such as glutathione and taurine 31 . Another gene upregulated in response to JPH203 was NNMT, which catalyzes transmethylation from S-adenosylmethionine to S-adenosylhomocysteine, which can be converted to homocysteine, then to cystathionine, a substrate for CTH 23,32 . When MDA-MB-231 cells were treated with 100 μM of JPH203, their viability was down from ~ 60% of vehicle treatment in control siRNA transfected cells to ~ 30% in CTH knockdown cells (Fig. 4F), which were almost the same levels of growth suppression observed with 100 μM of JPH203 treatment in wild type T-47D cells (Fig. 1F). Thus, it is likely that JPH203 resistance in MDA-MB-231 cells are largely mediated by upregulation of CTH in these cells.
Moreover, at baseline, a molecule that increases intracellular cysteine, cystine/glutamic acid transporter (xCT) was more abundantly expressed in MDA-MB-231 cells than T-47D cells (Fig. 7) 33 . Taken together, both www.nature.com/scientificreports/ at baseline and in response to JPH203, MDA-MB-231 cells appeared to be more prepared to handle ROS load, thereby better survive metabolic stress caused by amino acid depletion than T-47D cells. Given the fact that increased ROS load is the mechanism of inducing cell death in cytotoxic cancer chemotherapy or radiation therapy 34 , it is tempting to speculate that increased anti-oxidative capacity in MDA-MB-231 cells may explain, at least partially, their resistance to classic cancer therapies in addition to JPH203. Whether these findings are generally applicable to triple negative breast cancers is an interesting subject for future research. Regardless, as stated above, upregulation of CTH is likely to be the main mechanism of resistance to JPH203 treatment at least in MDA-MB-231 cells.
To address the question whether increased antioxidant capacity observed in MDA-MB-231 cells in response to JPH203 is due to amino acid depletion or unknown side effect of JPH203, the cells were cultured in the medium depleted of LAT1 substrate amino acids, essential amino acids plus tyrosine (Fig. 3D). This treatment Even T-47D cells, which did not increase CTH in the presence of JPH203, slightly increased CTH when substrate amino acids were totally depleted. However, complete depletion of essential amino acids plus tyrosine was impossible with even more than 100 μM of JPH203, because JPH203 is a reversible competitive inhibitor in a culture condition rich in substrate amino acids for much longer periods (4 days) than uptake assay (2 min). In addition, both of these cells expressed small amount of essential amino acid transporters that were not blocked by JPH203 (Fig. 1A).  www.nature.com/scientificreports/ Another result showing that treatment with more than 5 μM (up to 200 μM) of JPH203 did not increase CTH in T-47D cells (Fig. 3E) speaks against direct (i.e. not mediated by amino acid depletion) upregulation of CTH by JPH203. Taken together, it is likely that increased anti-oxidant capacity in MDA-MB-231 cells in response to JPH203 treatment was due to amino acid depletion, not a side effect of JPH203. In this regard, Cormerais et al. 17 reported that amino acid stress response of ATF4 upregulation and CGN2 phosphorylation is observed in LAT1 knockout cells. Both T-47D and MDA-MB-231 cells in our study showed the same stress response by JPH203 treatment (Fig. 3B). These results further support our interpretation that JPH203 induced cellular stress response in MDA-MB-231 cells by restricting amino acid uptake through LAT1.
Investigators including us often regard that inhibition of LAT1 causes tumor cell growth suppression through mTORC1 pathway inhibition or simply lack of essential amino acids, but this study implied that cellular antioxidant pathways in response to amino acid deprivation stress may be at least as important for tumor cell viability.

Materials and methods
RNA extraction and real-time quantitative PCR analysis. Cells were plated on 60 mm dishes at a density of 4 × 10 5 cells/dish and cultured for 2 days. Cells were subsequently treated with 100 μM JPH203 for 12 h before harvesting total RNA. Total RNA was isolated using Isogen (Nippon Gene, Tokyo, Japan) according to the manufacturer's instruction. The first-strand complementary DNA (cDNA) were synthesized from 1 μg of total RNA using MuLV Reverse Transcriptase (Life Technologies, Carlsbad, CA, USA) with oligo dT primer. Real-time PCR was performed with Premix Ex Taq (Takara Bio Inc., Shiga, Japan) or SYBR Select Master Mix (Thermo Fisher Scientific, MA, USA) using 7300 Real-Time PCR system (Thermo Fisher Scientific, MA, USA). Designed primers and probes were shown in Table 3. [ 14 C] l-leucine uptake assay. Cells were seeded on 24-well plates at a density of 1 × 10 5 cells/well for 2 days before experiment. Cells were subsequently incubated with Na + containing buffer (125 mM NaCl, 4 siRNA-mediated knockdown. Cells were seeded on 24-well plates at a density of 5 × 10 4 cells/well, and precultured in RPMI1640 medium supplemented with 10% FBS without antibiotics for 24 h before siRNA Table 3. Primer pairs and probes for real-time PCR.
Original blot data. As per the policy of the journal, original blot images of composite figures were presented in Supplementary Fig. 2.