Mammalian target of rapamycin signalling modulates amino acid uptake by regulating transporter cell surface abundance in primary human trophoblast cells

Abstract

Abnormal fetal growth increases the risk for perinatal complications and predisposes for the development of obesity, diabetes and cardiovascular disease later in life. Emerging evidence suggests that changes in placental amino acid transport directly contribute to altered fetal growth. However, the molecular mechanisms regulating placental amino acid transport are largely unknown. Here we combined small interfering (si) RNA-mediated silencing approaches with protein expression/localization and functional studies in cultured primary human trophoblast cells to test the hypothesis that mammalian target of rapamycin complex 1 (mTORC1) and 2 (mTORC2) regulate amino acid transporters by post-translational mechanisms. Silencing raptor (inhibits mTORC1) or rictor (inhibits mTORC2) markedly decreased basal System A and System L amino acid transport activity but had no effect on growth factor-stimulated amino acid uptake. Simultaneous inhibition of mTORC1 and 2 completely inhibited both basal and growth factor-stimulated amino acid transport activity. In contrast, mTOR inhibition had no effect on serotonin transport. mTORC1 or mTORC2 silencing markedly decreased the plasma membrane expression of specific System A (SNAT2, SLC38A2) and System L (LAT1, SLC7A5) transporter isoforms without affecting global protein expression. In conclusion, mTORC1 and mTORC2 regulate human trophoblast amino acid transporters by modulating the cell surface abundance of specific transporter isoforms. This is the first report showing regulation of amino acid transport by mTORC2. Because placental mTOR activity and amino acid transport are decreased in human intrauterine growth restriction our data are consistent with the possibility that dysregulation of placental mTOR plays an important role in the development of abnormal fetal growth.

Key points

• Inadequate nutrient supply during fetal life results in intrauterine growth restriction (IUGR), which may lead to obesity, diabetes, and cardiovascular disease later in life.

• A decreased placental amino acid transporter activity has been implicated in the pathophysiology of IUGR; however, the mechanisms regulating placental amino acid transporters in the human are largely unknown.

• We show that inhibition of mammalian target of rapamycin complex 1 or 2 markedly decreases the activity of key placental amino acid transporters in cultured primary human placental cells, mediated by modulating the movement of specific transporter isoforms between the cell interior and the plasma membrane.

• Because mTOR signalling is inhibited in the IUGR placenta, these findings identify one possible mechanism by which fetal nutrient supply is reduced in this pregnancy complication.

• Our data may help us better understand the regulation of amino acid transporters and the molecular mechanisms underlying IUGR.

Introduction

The intrauterine environment is critical for normal fetal development and changes in fetal nutrient availability can programme the fetus for metabolic and cardiovascular disease later in life (Gluckman & Hanson, 2004; Gluckman et al. 2008). Fetal growth is strongly dependent on nutrient supply, which is linked to placental transport capacity. The activity of placental amino acid transporters System L and System A is decreased in intrauterine growth restriction (IUGR) (Mahendran et al. 1993; Glazier et al. 1997; Jansson et al. 1998; Norberg et al. 1998) and has been shown in some reports to be upregulated in fetal overgrowth (Jansson et al. 2002). These data suggest that changes in the activity of placental nutrient transporters may directly contribute to abnormal fetal growth (Sibley et al. 2005; Jansson & Powell, 2006, 2007). The System L amino acid transporter is a sodium-independent exchanger mediating cellular uptake of essential amino acids including leucine (Verrey et al. 2003). This transporter is a heterodimer, consisting of a light chain, typically LAT1 (large neutral amino acid transporter 1) (SLC7A5) or LAT2 (SLC7A8), and a heavy chain, 4F2hc/CD98 (4F2 cell-surface antigen heavy chain/cluster of differentiation 98) (SLC3A2). System A is a sodium-dependent transporter mediating the uptake of non-essential neutral amino acids into the cell (Mackenzie & Erickson, 2004). There are three isoforms of System A: sodium-dependent neutral amino acid transporter 1 (SNAT1) (SLC38A1), SNAT2 (SLC38A2) and SNAT4 (SLC38A4), and all are expressed in the human placenta (Desforges et al. 2006). System A activity establishes the high intracellular concentration of non-essential amino acids, which are used to exchange for extracellular essential amino acids via System L. Thus, System A activity is critical for cellular uptake of both non-essential and essential amino acids.

Because changes in placental amino acid transporter activity have been implicated in altered fetal growth, identification of the factors regulating these transporters may provide insight into the causes underlying the development of important pregnancy complications. However, the molecular mechanisms regulating amino acid transport in human cells are largely unknown. The mammalian target of rapamycin (mTOR) signalling pathway responds to changes in nutrient availability and growth factor signalling to control cell growth (Yang & Guan, 2007; Ma & Blenis, 2009; Foster & Fingar, 2010). mTOR exists in two complexes, mTOR complex 1 (mTORC1) and 2. One of the key differences between these two complexes is that mTOR associates with the protein raptor (regulatory associated protein of mTOR) in mTORC1 and with rictor (rapamycin-insensitive companion of mTOR) in mTORC2 (Yang & Guan, 2007). It is well established that TOR in yeast regulates amino acid permeases (Edinger, 2007) but it is not until more recently that mTOR has emerged as a regulator of amino acid transporters in mammalian cells. In lymphoma cells, the mTOR inhibitor rapamycin selectively downregulated the expression of five genes involved in amino acid transport (Peng et al. 2002). LAT1 mRNA has been shown to be increased in platelet-derived growth factor (PDGF)-treated vascular smooth muscle cells and this induction was dependent on mTOR (Liu et al. 2004). In a murine T-cell line, cell surface expression of 4F2hc was inhibited by 24 h rapamycin incubation (Edinger & Thompson, 2002). System A activity in L6 myotubes has been shown to be upregulated by leucine in a mTOR-dependent manner (Peyrollier et al. 2000).

We recently reported that inhibition of mTOR signalling decreases the activity of human placental amino acid transporters (Roos et al. 2007, 2009). Furthermore, placental mTOR activity is markedly decreased in human IUGR (Roos et al. 2007; Yung et al. 2008). These observations are consistent with a role for placental mTOR signalling in regulating placental amino acid transport and fetal growth. However, the mechanisms involved and the specific role of mTORC1 and mTORC2 signalling in the regulation of amino acid transporters remain to be established. The primary mechanism by which the mTORC1 signalling pathway influences cell function and growth is by controlling protein synthesis (Ma & Blenis, 2009). However, in our previous studies mTOR inhibition using rapamycin markedly inhibited cellular amino acid uptake in human primary trophoblast cells without affecting global protein expression of amino acid transporter isoforms (Roos et al. 2009). These findings are consistent with the possibility that mTOR regulates amino acid transporter activity at the post-translational level. Using gene silencing approaches in cultured primary human trophoblast cells we tested the hypothesis that mTORC1 and mTORC2 regulate placental amino acid transporters mediated by affecting plasma membrane trafficking of specific transporter isoforms.

Methods

Cytotrophoblast isolation and culture

Cytotrophoblast cells were isolated from normal human term placentas and cultured in vitro (Kliman et al. 1986; Lager et al. 2011). Tissue was collected after informed written consent and was approved by the Institutional Review Board of University of Texas Health Science Center at San Antonio. Cells were plated in either 60 mm culture dishes (∼10 × 10⁶ cells/dish for western blot analysis) or 6-well plates (for amino acid uptake experiments; ∼2 × 10⁶ cells/well for rapamycin treatment or ∼3.75 × 10⁶ cells/well for RNAi-mediated gene silencing) and cultured in 5% CO₂, 95% atmosphere air at 37°C for 90 h. Cell culture media (DMEM/Ham's F-12, supplemented with l-glutamine, penicillin, streptomycin, gentamycin and 10% fetal bovine serum) was changed daily.

Assessment of biochemical differentiation and viability

To confirm that trophoblast cells were undergoing biochemical differentiation, and to assess their viability with time in culture, the release of human chorion gonadotropin (hCG) by trophoblast cells into the culture medium after 18, 42, 66 and 90 h was measured using a commercial ELISA kit which detects the β-subunit of hCG (Immuno Biological Labs, Minneapolis, MN, USA). The level of purity of trophoblast cells was determined using vimentin (antibody 20346, Abcam) as a marker for mesenchymally derived cells together with the trophoblast-specific marker cytokeratin 7 (CK7; antibody 9098, Abcam, Cambridge, UK). Syncytin expression was measured as an index of trophoblast cell–cell fusion and differentiation in culture (Langbein et al. 2008). Antibodies directed against caspase-3 (R&D Systems, Minneapolis, MN, USA) and cleaved poly (ADP-ribose) polymerase (Affinity BioReagents, Golden, CA, USA) were used to determine the presence of apoptosis in culture.

RNA interference-mediated silencing of mTORC1 and/or mTORC2

Dharmafect 2 transfection reagent (Thermo Scientific, Rockford, IL, USA) and small interference RNAs (siRNAs) (Sigma-Aldrich, St Louis, MO, USA), targeting raptor (100 nm; sense, 5′ CAGUUCACCGCCAUCUACA) and/or rictor (100 nm; sense, 5′ CGAUCAUGGGCAGGUAUUA), or a non-coding scrambled sequence (100 nm; sense: 5′ GAUCAUACGUGCGAUCAGATT), were added to cultured primary trophoblast cells (∼3.75 × 10⁶ cells per well in 6-well plate; ∼10 × 10⁶ cells in a 60 mm dish) after 18 h in culture, incubated for 24 h, removed, and fresh medium added to the wells (Forbes et al. 2009). At 66 h (data not shown) and 90 h in culture, the efficiency of target silencing was determined at the protein and functional levels using western blot.

Growth factor treatment

Human primary trophoblast cells were incubated in 5.8 ng ml⁻¹ insulin (Sigma-Aldrich) and 300 ng ml⁻¹ insulin-like growth factor 1 (IGF-I; Sigma-Aldrich) from 66 to 90 h in culture. The insulin concentration used in these experiments has previously been shown to stimulate System A in primary villous fragments (Jansson et al. 2003) and cultured human primary trophoblast cells (Jones et al. 2010), and corresponds to normal postprandial insulin levels in pregnant women (Phelps et al. 1981). The IGF-1 concentration used constitutes physiological concentrations of IGF-1 in third trimester maternal serum (Jansson et al. 2008).

System A and System L amino acid uptake assay

Amino acid uptake in cultured primary human trophoblast cells was determined at 90 h in culture. The activity of System A and System L amino acid transporters was assessed by measuring the Na⁺-dependent uptake of [¹⁴C]methyl-aminoisobutyric acid (MeAIB; 20 μm) and the 2-amino-2-norbornane-carboxylic acid (BCH; 64 μmol l⁻¹)-inhibitable uptake of [³H]leucine (0.0125 μm), respectively, as described in detail previously (Roos et al. 2009). In control experiments, System L and A amino acid transporter activity was determined following incubation of cells in 100 μm colchicine, an inhibitor of the cytoskeleton, for 24 h (from 66 to 90 h in culture).

Serotonin uptake

³H-labelled 5-hydroxytryptamine (5-HT, serotonin) uptake was measured in trophoblast cells (control, raptor or/and rictor silenced cells) at 90 h of culture, as described previously (Keating et al. 2009). Transport experiments were performed in a buffer containing 10 mm Hepes (N- 2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid)– Tris and 150 mm NaCl (pH 7.4). Uptake was initiated by the addition of 1 ml of buffer containing 30 nm [³H]5-HT at 37°C. Non-mediated serotonin uptake was determined in the presence of buffer containing 1.5 mmol l⁻¹ unlabelled 5-HT. Incubation was stopped after 6 min by removing the incubation medium and washing the cells three times with ice-cold buffer.

Isolation of microvillous plasma membranes from trophoblast cells

Microvillous plasma membranes (MVMs) were isolated from total homogenates of cultured primary human trophoblast cells using differential centrifugation and Mg²⁺ precipitation as described previously for mouse placental tissue (Jones et al. 2009; Kusinski et al. 2010) and modified for cultured cells. Briefly, centrifugation steps were carried out at 4°C and all other steps were performed on ice. Cells were lysed, scraped off the plate and subsequently homogenized. The cell homogenate was centrifuged at 10,000 g for 15 min, the supernatant was collected and the pellet was resuspended, homogenized in 1 ml of buffer D (10 mm Tris-Hepes, 250 mm sucrose, 1 mm EDTA), and centrifuged again at 10,000 g for 10 min. The two resulting supernatants were combined and centrifuged at 125,000 g for 30 min. The pelleted crude membrane fraction was resuspended in 2 ml of buffer, and 12 mm MgCl₂ was added. The mixture was stirred slowly for 20 min on ice and then centrifuged at 2500 g for 10 min. The supernatant containing MVMs was centrifuged at 125,000 g for 30 min and the final pellet was resuspended in buffer D. Protein concentration was determined using the Bradford assay. MVM enrichment was assessed using the MVM/homogenate ratio of alkaline phosphatase and insulin receptor protein expression as determined using western blot after loading equal amounts of protein of MVM and cell lysates.

Western blotting

For immunoblotting, cells were lysed in buffer containing phosphatase and protease inhibitors. Subsequently, cells were scraped, collected, and sonicated. Proteins in cell lysates and MVMs were separated by electrophoresis. Western blotting was carried out as described (Roos et al. 2009). Protein expression of raptor and rictor and total and phosphorylated p70 S6 kinase (S6K1) (Thr-389), 4E-eukaryotic initiation factor binding protein-1 (4EBP-1) (Thr-37/46) or (Thr-70), S6 ribosomal protein (Ser-235/236), serum and glucocorticoid-regulated kinase 1 (SGK1) (Ser-422), and Akt (Ser-473) was analysed in cell lysates. The SGK1 antibody was obtained from Santa Cruz (Santa Cruz, CA, USA) and the remaining antibodies were purchased from Cell Signaling Technology (Boston, MA, USA). Protein expression of the System A amino acid transporter isoforms (SNAT) 1, 2 and 4 and the System L amino acid transporter isoforms LAT1, LAT2 and 4F2hc was analysed in total cell lysates and MVM preparations. The SNAT1 antibody was produced as described previously (Gu et al. 2001) and was generously provided by Dr Jean Jiang at the University of Texas Health Science Center San Antonio. A polyclonal SNAT2 antibody generated in rabbits (Ling et al. 2001) was received as a generous gift from Dr P. Prasad at the University of Georgia, Augusta. Affinity-purified polyclonal anti-SNAT4 antibodies were produced in rabbits using the epitope YGEVEDELLHAYSKV in human SNAT4 (Eurogentec, Seraing, Belgium). Antibodies targeting the LAT1 and LAT2 were produced in rabbits as described previously (Park et al. 2005). The 4F2hc antibody was purchased from Santa Cruz Biotechnology and anti-β-actin was from Sigma-Aldrich. Analysis of the blots was performed by densitometry using an Alpha Imager (Alpha Innotech Corporation, San Leandro, CA, USA). Target band densities were normalized to loading using β-actin. For each protein target the mean density of the control sample bands was assigned an arbitrary value of 1. All individual densitometry values were expressed relative to this mean.

Immunofluorescence

Trophoblast cells were cultured in 6-well plates containing poly-l-lysine-coated coverslips (1.5 mm thick) and fixed with 4% paraformaldehyde at room temperature for 30 min. After blocking in 2% FBS diluted in PBS, cells were permeabilized using Triton X-100 and incubated with either anti-SNAT2 or anti-4F2hc antibodies for 12 h at 4°C. After several washes, cells were incubated in an Alexa Fluor 546-conjugated rabbit anti-goat IgG or Alexa Fluor 546-conjugated goat anti-rabbit IgG for 1 h at room temperature. After three washes in PBS, coverslips were mounted in a drop of Vectashield (Vector Laboratories, Burlingame, CA, USA). Images were acquired with an Olympus FV1000 confocal laser-scanning microscope.

Data presentation and statistics

The number of experiments (n) represents the number of placentas studied. In the uptake (amino acid/serotonin) experiments, each condition was studied in triplicate, and data were averaged to represent trophoblast cells isolated from one placenta. Data are presented as means ± SEM or +SEM. Statistical significance of differences between control and experimental groups was assessed using repeated-measures (RM) ANOVA or Student's t test. A P value <0.05 was considered significant.

Results

Validation of silencing efficiency

To confirm that the siRNA reduced expression of the proteins encoded by the targeted genes, we performed western blots on siRNA-transfected cells at 90 h of culture (Fig. 1). Raptor or rictor siRNA markedly decreased the protein expression of raptor (−74%; P < 0.001, n= 6) and rictor (−70%; P < 0.001, n= 6), respectively. Importantly, raptor silencing significantly decreased the phosphorylation of S6 kinase (Thr-389) (−61%; P < 0.001, n= 5), 4EBP-1 (Thr-37/46) (−45%; P < 0.05, n= 4) and S6 ribosomal protein (Ser-235/236) (−51%; P < 0.05, n= 4), all functional readouts for mTORC1 signalling (Fig. 1A and B). Similarly, following transfection with rictor siRNA, the expression of phospho-Ser-473-Akt (−55%; P < 0.05, n= 3) (Fig. 1C and D), a read-out of mTORC2 function, was markedly decreased. Following simultaneous silencing of raptor and rictor, their protein expression as well as mTORC1 and 2 function were inhibited to the same extent as in experiments silencing these targets individually (Fig. 2A and B). These results confirm that our RNAi approach efficiently silenced the mTORC1 and mTORC2 signalling pathways.

Figure 1. Silencing efficiency.

A and B, effect of raptor silencing on raptor protein expression and mTORC1 activity. A, representative western blots of raptor (n= 6), phosphorylated S6 kinase (Thr-389) (n= 5), total S6 kinase (n= 6), phosphorylated 4EBP-1 (Thr-37/46) (n= 4), total 4EBP-1 (n= 4), phosphorylated S6 ribosomal protein (Ser-235/236) (n= 4) and total S6 ribosomal protein (n= 4) expression in cell lysates of scramble control, scramble siRNA and raptor silenced cells. Equal loading was performed. B, summary of the western blot data. Values are given as means + SEM. Means without a common letter are statistically different by one-way ANOVA with Tukey–Kramer multiple comparisons post hoc test (P < 0.05). C and D, effect of rictor silencing on rictor protein expression and mTORC2 activity. C, representative western blots of rictor (n= 6), phosphorylated Akt (Ser-473) (n= 3) and total Akt (n= 3) expression in cell lysates of scramble control, scramble siRNA and rictor silenced cells. Equal loading was performed. D, summary of the western blot data. Values are given as means + SEM. Means without a common letter are statistically different by one-way ANOVA with Tukey–Kramer multiple comparisons post hoc test (P < 0.05).

Figure 2. Effect of raptor plus rictor silencing on raptor and rictor protein expression and mTOR activity.

A, representative western blots of raptor (n= 4), rictor (n= 4), phosphorylated S6 ribosomal protein (Ser-235/236) (n= 4) and phosphorylated Akt (Ser-473) (n= 4) expression in cell lysates of scramble siRNA and raptor+rictor silenced cells. Equal loading was performed. B, summary of the western blot data. Values are given as means + SEM. *P < 0.05 versus control; unpaired Student's t test.

Inhibition of mTORC1 and/or mTORC2 does not affect trophoblast cell viability, differentiation or apoptosis

We cultured primary cytotrophoblast cells isolated from human term placenta using a well-established protocol (Kliman et al. 1986). These cells form syncytial islands in culture, and this is used as a model for the syncytiotrophoblast, the transporting epithelium of the human placenta. After 66 h in culture, there was a marked increase in hCG production by trophoblast cells, and the levels remained high until at least 90 h after plating (see online Supplemental material, Fig. S1A). Because hCG is produced predominantly by syncytiotrophoblast, these data provide evidence of cell differentiation and syncytialization. Furthermore, hCG secretion profiles were similar in cells in which both raptor and/or rictor had been silenced as compared to cells incubated in scrambled siRNA, tranfection agent only (scramble control) or in cell culture media only (control) (Supplemental Fig. S1A). Vimentin could not be detected in trophoblast cell cultures at 90 h, suggesting no significant contamination of mesenchyme-derived cells (Supplemental Fig. S1B). However, cultured cells were positive for cytokeratin-7, a well-established marker for trophoblast cells (Supplemental Fig. S1C). We demonstrate further that there was no significant difference in the protein expression of syncytin (a differentiation marker; Supplemental Fig. S2A), or in the apoptosis markers caspase-3 (Supplemental Fig. S2B) and poly (ADP-ribose) polymerase in cultured trophoblast cells transfected with raptor and/or rictor siRNA as compared to control/scramble siRNA. Collectively, these data indicate that key components of the trophoblast mTOR signalling pathway can be silenced without adversely affecting trophoblast cell viability, differentiation or apoptosis.

Cross-talk between mTORC1 and mTORC2 pathways

In many cell lines mTORC1 signalling also appears to regulate mTORC2 signalling and mTORC2 may modulate the mTORC1 signalling pathway (Manning et al. 2005; Dibble et al. 2009). As a result, inhibition of one of the pathways will affect both signalling pathways, posing some challenges in interpreting the data. However, raptor silencing in cultured primary human trophoblast cells did not significantly modulate mTORC2 signalling as assessed by the expression of phospho-Ser-473-Akt (Fig. 3A and B). In addition, rictor silencing did not influence mTORC1 activity (Fig. 3C and D). These data indicate that, in contrast to many cell lines, there may not be an extensive cross-talk between the mTORC1 and mTORC2 pathways in cultured primary human trophoblast cells.

Figure 3. Cross-talk between mTORC1 and mTORC2 signalling pathways.

A and C, effect of raptor silencing on rictor protein expression and mTORC2 activity. A, representative western blots of rictor (n= 4), phosphorylated Akt (Ser-473) (n= 8) and total Akt (n= 4) expression in cell lysates of scramble siRNA and raptor silenced cells. Equal loading was performed. C, summary of the western blot data. Values are given as means + SEM. None of the differences between raptor and scramble siRNA were statistically significant; unpaired Student's t test. B and D, effect of rictor silencing on raptor protein expression and mTORC1 activity. B, representative western blots of raptor (n= 8), phosphorylated S6 kinase (Thr-389) (n= 8), total S6 kinase (n= 4), phosphorylated 4EBP-1 (Thr-37/46) (n= 4), phosphorylated S6 ribosomal protein (Ser-235/236) (n= 8) and total S6 ribosomal protein (n= 4) expression in cell lysates of scramble siRNA and rictor silenced cells. Equal loading was performed. D, summary of the western blot data. Values are given as means + SEM. None of the differences between rictor and scramble siRNA were statistically significant; unpaired Student's t test.

mTOR regulates System A and L amino acid transporter activity

BCH-inhibitable uptake of [³H]leucine (corresponding to System L activity) as well as the Na⁺-dependent uptake of [¹⁴C]MeAIB (a measure of System A-mediated uptake) was linear up to 12 min in cultured primary human trophoblast cells at 90 h in culture (Fig. 4A and B). Based on these time course experiments, an incubation time of 8 min was chosen in subsequent experiments. In order to study the effect of mTOR inhibition on trophoblast amino acid uptake, after 66 h in culture, trophoblast cells were incubated with 100 nm rapamycin or vehicle (0.02% DMSO) for 24 h, and the uptakes of [³H]leucine and [¹⁴C]MeAIB were then measured. As shown in Fig. 4C and D, rapamycin reduced System L uptake by 42% (n= 9; P < 0.001) and System A uptake by 48% (n= 4; P < 0.01) as compared to control.

Figure 4. mTOR regulation of System A and L amino acid transport activity.

A and B, time course of amino acid uptake in trophoblast cells. System L activity (A) was measured as the BCH-inhibitable uptake of [³H]leucine, and System A activity (B) was determined as the Na⁺-dependent uptake of [¹⁴C]MeAIB. Uptakes were linear for at least 12 min (System L: r= 0.91, P < 0.0001, n= 9; System A: r= 0.81, P < 0.0001, n= 6; Spearman's rho). Values are given as means + SEM. C–F, effect of mTOR inhibition on trophoblast basal and growth factor-stimulated amino acid uptake. Inhibition of mTOR by 100 nm rapamycin decreased the basal activity of System L (C) and System A (D) transporters in cultured human primary trophoblast cells. Addition of insulin (5.8 ng ml⁻¹) and IGF-I (300 ng ml⁻¹) to the cell culture media stimulated the System L (E) and System A (F) transporter in control cells. The stimulatory effect of growth factor (GF) (defined as the difference between uptake in growth factor-containing and control media) on System A and System L activity was not significantly affected by the rapamycin. Values are means + SEM; n= 9 for System L and n= 4 for System A. *P < 0.05 versus control; unpaired Student's t test.

Cellular amino acid uptake is stimulated by growth factors such as insulin and IGF-I (Edinger & Thompson, 2002; Mackenzie & Erickson, 2004); however, the mechanisms involved remain to be fully established. We next studied the role of mTOR signalling in mediating the effect of growth factors on System A and L amino acid transporter activity. At 66 h of culture, trophoblast cells were incubated with insulin and IGF-I for 24 h (66–90 h), with or without rapamycin (100 nm). As expected, insulin and IGF-I stimulated System L (+42%, P < 0.001; n= 9, Fig. 4E) and System A activity (+115%, P < 0.01; n= 4, Fig. 4F), in line with previous reports in the literature (Jansson et al. 2003; Roos et al. 2009). However, growth factor-stimulated (defined as the difference between uptake in growth factor-containing and control media) System L (Fig. 4E) and System A activity (Fig. 4F) was unaffected by rapamycin, suggesting that regulation of amino acid uptake by growth factors is not dependent on mTORC1 signalling in human primary trophoblast cells.

mTOR regulation of amino acid transporter activity involves both mTORC1 and mTORC2 pathways

Although rapamycin has been regarded as a specific mTORC1 inhibitor, it has become evident that rapamycin also inhibits mTORC2, in particular when used in higher doses for prolonged incubation times (Sarbassov et al. 2006). Furthermore, rapamycin may not inhibit all mTORC1 functions (Thoreen et al. 2009). To explore the contribution of mTORC1 and/or mTORC2 in regulating System L and A transporter activity we silenced raptor and/or rictor, essential components of mTORC1 and mTORC2, respectively, and then measured the System L and A uptake. Silencing of raptor (System L, –55%, P < 0.001; System A, −44%, P < 0.05; n= 5) or rictor (System L, –49%, P < 0.001; System A, −54%, P < 0.05; n= 5) markedly inhibited basal levels of System L and System A amino acid transporter activity in cultured primary human trophoblast cells (Fig. 5A and B). These data indicate that mTORC1 and mTORC2 independently regulate trophoblast amino acid transport. Next, we examined the effect of simultaneous silencing of raptor and rictor on basal System L and A transport activity. Following knock-down of both raptor and rictor basal System L (P < 0.0001; n= 4) and System A (P < 0.0001; n= 4), amino acid transporter activity was completely inhibited (Fig. 5C and D).

Figure 5. Role of mTORC1 and mTORC2 in regulating System A and System L amino acid uptake in trophoblast cells.

A and B, effect of silencing raptor (mTORC1 inhibition) or rictor (mTORC2 inhibition) on System L (A) and System A (B) transporter activity. Values are means + SEM; n= 5 for System L and System A. Means without a common letter differ P < 0.05 by RMANOVA with Tukey–Kramer multiple comparisons post hoc test. C and D, effect of simultaneous silencing of raptor and rictor on basal System L (C) and System A amino acid uptake (D). Values are means + SEM; n= 4. *P < 0.05 versus control; unpaired Student's t test. E and F, regulation of growth factor-stimulated amino acid uptake by mTORC1 and mTORC2. The stimulatory effect of growth factors (defined as the difference between uptake in growth factor-containing and control media) on System L (E) and System A (F) activity was not significantly influenced by raptor or rictor silencing. In contrast, simultaneous raptor and rictor silencing in trophoblast cells abolished the growth factor-stimulated Systems A and L uptake activity. Values are means + SEM; n= 5 for System L and System A. Means without a common letter differ significantly (P < 0.05) by RMANOVA with Tukey–Kramer multiple comparisons post hoc test.

Either mTORC1 or mTORC2 is sufficient to mediate the stimulatory effect of growth factors on System A and L transport

Silencing of raptor or rictor independently did not attenuate the growth factor stimulation of System L (P > 0.05, n= 5) or System A (P > 0.05, n= 5) amino acid uptake (Fig. 5E and F). In strong contrast, in cells silenced for both raptor and rictor, the growth factor-stimulated System L (P < 0.05, n= 4) and System A (P < 0.05, n= 4) amino acid transporter activity was abolished (Fig. 5E and F).

Activation of mTORC1 and 2 by growth factors

It is well established that mTORC1 is downstream of insulin/IGF-I signalling. mTORC2 phosphorylates SGK1, which has been reported to be activated by growth factors in other cells. In agreement with these observations, we show that growth factors increased the phosphorylation of SGK1 (Ser-422) and S6 ribosomal protein (Ser-235/236) in primary human trophoblast cells (Fig. 6).

Figure 6. Activation of mTORC1 and 2 by growth factors.

A, representative western blot for phospho-SGK1 (readout for mTORC2 signalling) and phospho-ribosomal protein S6 (mTORC1 readout). B, summary of data. Values are means + SEM; n= 3. *P < 0.05 versus control; unpaired Student's t test.

mTOR signalling does not regulate serotonin uptake

We postulated that mTORC1 and mTORC2 regulation of trophoblast transporters is transporter specific. To address this hypothesis, we measured serotonin uptake in cultured primary human trophoblast cells at 90 h of culture in response to raptor or rictor or raptor+rictor silencing. As shown in Supplemental Fig. S3A, [³H]5-HT uptake was linear for 8 min. Based on these time course experiments, an incubation time of 6 min was chosen in subsequent experiments. Silencing of raptor or rictor (P= 0.65, n= 3, Supplemental Fig. S3B) or raptor+rictor (P= 0.93, n= 3, Supplemental Fig. S3C) did not alter trophoblast serotonin uptake as compared to control.

mTORC1 and mTORC2 do not regulate global expression of System L and A transporter isoforms

We sought to identify a molecular mechanism underlying mTORC1 and mTORC2 regulation of System L and A transporter activity. Because mTORC1 signalling is well established to regulate protein synthesis (Thomas & Hall, 1997) we explored the effects of mTOR inhibition on transporter isoform expression. As shown in Fig. 7, the expression of SNAT1, SNAT2, SNAT4, LAT1, LAT2 and 4F2hc in total cell lysates of raptor- or rictor-silenced cells was unaltered as compared to cells treated with scramble siRNA and control. These data indicate that mTOR signalling regulates trophoblast amino acid uptake by post-translational mechanisms, rather than by altering expression levels.

Figure 7. Global protein expression of System L and System A amino acid transporter isoforms in total cell lysates in response to raptor or rictor silencing.

Representative western blots are shown for isoforms of System L (A): L-type amino acid transporter (LAT1 (45 kDa), LAT2 (30 and 50 kDa)), and System A (C): sodium-coupled neutral amino acid transporter (SNAT1 (52 kDa), SNAT2 (52 kDa), SNAT4 (54 kDa)), and 4F2hc (80 kDa), in cell lysates of control, scramble siRNA and raptor or rictor silenced cells. Equal loading was performed. The histograms (B and D) summarize the western blot data from scramble control (n= 5), scramble siRNA (n= 5) and raptor or rictor (n= 5) silenced trophoblast cells. Values are means + SEM. No significant differences between raptor or rictor silenced cells and controls were observed (RMANOVA with Tukey–Kramer multiple comparisons post hoc test).

mTORC1 and mTORC2 signalling regulate trophoblast amino acid transporter trafficking

To study changes in cell surface expression we isolated a microvillous plasma membrane (MVM) fraction from cultured primary trophoblast cells and determined isoform protein expression in response to manipulation of mTORC1 and 2 signalling. The enrichment of alkaline phosphatase, a MVM marker, was determined by the ratio of alkaline phosphatase expression in MVM over total cell lysates. Alkaline phosphatase enrichment in MVM isolated from control cells was 5.7 ± 0.6 (n= 3), which was not different from MVM isolated from raptor (5.5 ± 0.5, n= 3) or rictor silenced cells (5.3 ± 0.2, n= 3). Next, insulin receptor (IR) expression, which is exclusively expressed in the MVM of term in vivo syncytiotrophoblast (Tavare & Holmes, 1989), was determined in the MVM and cell lysates. The ratio of IR expression in MVM over total cell lysates was comparable between control (12.0 ± 1.1, n= 3) and raptor (13.6 ± 2.4, n= 3) or rictor (12.6 ± 2.2, n= 3) silenced cells. These data confirm significant MVM enrichment. We found that inhibition of mTORC1 or mTORC2 caused a marked decrease in the expression of the System A transporter isoform SNAT2 and System L transporter isoform LAT1 in the microvillous plasma membrane fraction, which is in sharp contrast to the unchanged SNAT2 and LAT1 expression in total cell lysates (Fig. 8). However, the plasma membrane abundance of the SNAT1, SNAT4, LAT2 and 4F2hc isoforms was unaltered by mTORC1 or mTORC2 inhibition (Fig. 8). The effect of simultaneous silencing of raptor and rictor on MVM expression of amino acid transporter isoforms was qualitatively similar to that when raptor and rictor were silenced separately, i.e., primary involving a decreases in SNAT 2 and LAT 1 expression (Supplemental Fig. S4). However, the decrease in MVM LAT1 and SNAT2 expression was more extensive in response to the double (raptor+rictor) knockdown (Supplemental Fig. S4), consistent with the transporter activity data (Fig. 5). We then used immunofluorescence to detect amino acid transporter isoforms in fixed cells. We observed that in response to raptor silencing SNAT2 (Fig. 9), but not 4F2hc (data not shown), was relocated from the plasma membrane and the periphery to the cytosol of the syncytial islands. Incubating primary trophoblast cells in colchicine, an agent interfering with microtubuli organization, markedly decreased System L and System A activity as compared to control, confirming the involvement of the cytoskeleton in trafficking amino acid transporters to the plasma membrane (Supplemental Fig. S5). Collectively, these data support the hypothesis that mTORC1 and 2 regulate System L and A amino acid transporter activity by affecting transporter trafficking. Furthermore, the mTORC1 and mTORC2 regulation of trophoblast amino acid transporters is isoform specific, i.e. these signalling pathways regulate only a subset of transporter isoforms.

Figure 8. Protein expression of System L and System A amino acid transporter isoforms in MVMs isolated from control and raptor or rictor silenced trophoblast cells.

A, representative western blots are shown for L-type amino acid transporter (LAT1 (45 kDa), LAT2 (30 and 50 kDa), 4F2hc (80 kDa)) and sodium-coupled neutral amino acid transporter (SNAT1 (52 kDa), SNAT2 (52 kDa), SNAT4 (54 kDa)) in cell lysates and MVMs of control and raptor or rictor silenced cells. Equal loading was performed. The histogram (B) shows the protein expression of transporter isoforms in total cell lysates and MVMs in response to silencing of raptor or rictor (n= 3) as compared to control cells treated with scramble siRNA (n= 3). Expression in control cells for each isoform was arbitrarily assigned a value of one and is indicated by the dotted line in the figure. Values are means + SEM; *P < 0.05 versus control; unpaired Student's t test.

Figure 9. Cellular localization of SNAT2 protein expression in mTORC1 inhibition trophoblast cells.

Trophoblast cells were transfected at 18 h in culture with scramble (A and B) or raptor (C and D) siRNA. At 90 h in culture, cells were fixed and SNAT2 expression (orange) was visualized using immunofluorescence. Nuclei were counterstained using DAPI (4′,6-diamidino-2-phenylindole) (blue). Scale bars represent 10 μm.

Discussion

We utilized gene silencing approaches to acquire functional data in human primary trophoblast cells, which has allowed us to obtain specific mechanistic information of physiological relevance about the role of mTORC1 and mTORC2 signalling pathways in regulating cellular amino acid uptake. We report that mTORC1 and mTORC2 signalling pathways constitute powerful positive regulators of trophoblast System A and System L amino acid transporter activity, which is mediated by modulation of cell surface abundance of a specific subset of transporter isoforms. This is the first report showing regulation of amino acid transport by mTORC2, in any cell type. Because fetal growth is highly dependent on amino acid availability, regulation of placental amino acid transport by trophoblast mTORC1 and 2 signalling may constitute an important molecular link between maternal nutrition and fetal growth. Placental mTOR and amino acid transporter activity has been reported to be inhibited in human intrauterine growth restriction and our data are therefore consistent with the possibility that dysregulation of placental mTOR plays an important role in the development of abnormal fetal growth.

It is well established that when human primary villous trophoblast cells syncytialize in culture they become polarized and develop abundant and regular microvilli on the cell surface facing the culture media (Farmer & Nelson, 1992; Esterman et al. 1997; Morrish et al. 1997). This is the cell surface that is freely accessible in vitro during transport activity measurements and it is highly likely that the measured uptake can be accounted for by transport across this surface. In vivo the System A transport activity is highly polarized to the maternal-facing microvillous plasma membrane of the syncytiotrophoblast (Jansson et al. 2002). System L is believed to be critical for net uptake of essential amino acids, such as leucine, across the microvillous membrane. Although System L activity is present also on the basal plasma membrane (Jansson et al. 2002), net efflux of leucine across the basal plasma membrane is mediated by specific efflux transporters rather than System L (Cleal et al. 2011). Collectively, these data suggest that the amino acid transport studied in cultured human primary trophoblast cells represents the transport across the maternal-facing syncytiotrophoblast microvillous plasma membrane in vivo.

Although the regulation of amino acid permeases by TOR is well documented in yeast (Edinger, 2007), mTOR has only recently been implicated in the regulation of amino acid transporters in mammalian cells (Peyrollier et al. 2000; Edinger & Thompson, 2002; Peng et al. 2002; Liu et al. 2004; Edinger, 2007). Inhibition of mTORC1 and mTORC2 independently decreased System A- and System L-mediated amino acid transport by approximately 50% in cultured human primary trophoblast cells. In contrast, trophoblast serotonin uptake was not affected by mTOR inhibition, suggesting that mTORC1 and 2 regulate a specific subset of transporters, possibly those most critical for cell growth. Strikingly, simultaneous silencing of mTORC1 and 2 completely inhibited basal amino acid uptake by the two transport systems under study. Despite the profound inhibition of amino acid uptake, trophoblast cell viability, differentiation or apoptosis were not affected by simultaneous mTORC1 and 2 inhibition. This is likely to be due to the fact that syncytialized trophoblast cells in culture are non-dividing terminally differentiated cells which have a relatively low basal requirement for amino acids that can be met over the experimental period studied by the high intracellular concentrations present at the start of the experiment.

Although the magnitude of mTORC1 and 2 inhibition through gene silencing in our cell culture model may be more extensive than encountered in vivo, these data suggest that mTOR signalling is a powerful positive regulator of placental amino acid transporters in vivo. Notably, inhibition of mTORC1 or mTORC2 independently did not attenuate growth factor-stimulated amino acid uptake. In strong contrast, in cells in which both raptor and rictor were silenced the growth factor-stimulated System A and System L amino acid transporter activity was abolished. These results indicate that growth factors regulate amino acid transport mediated through both mTORC1 and 2 signalling; however, these two pathways appear complementary because signalling through one of the pathways was sufficient for the full stimulatory effect. These findings are in general agreement with the well-established role of growth factors, such as insulin and IGF-I, as upstream activators of mTORC1 signalling (Thomas & Hall, 1997). Albeit less investigated, evidence in the literature suggests that mTORC2 is also stimulated by growth factors (Kumar et al. 2008). Indeed, our data indicate that growth factors activate mTORC2 signalling in primary human trophoblast cells and that this signalling pathway is involved in the stimulatory effect of growth factors on amino acid transport in the placenta.

Although previously regarded as a specific mTORC1 inhibitor, recent observations suggest that rapamycin may also inhibit mTORC2 (Sarbassov et al. 2006). In the current study, rapamycin inhibited amino acid uptake to the same extent as mTORC1 inhibition after raptor silencing. Furthermore, neither rapamycin nor raptor silencing attenuated growth factor-stimulated amino acid uptake. These results suggest that rapamycin does not affect mTORC2 signalling in primary human trophoblast cells under the conditions used in our study.

The primary mechanism by which the mTORC1 signalling pathway influences cell function and growth is by controlling translation initiation, the rate limiting step of protein synthesis (Ma & Blenis, 2009). Recent high-resolution transcriptome-scale ribosome profiling suggests that mTORC1 regulates mRNA translation largely mediated by 4EBP-1 (Thoreen et al. 2012). Furthermore, some evidence in the literature supports a role for mTORC1 regulation of transcription (Peng et al. 2002). Our data show that mTORC1 and mTORC2 regulate the activity of amino acid transporters in primary human trophoblast cells without affecting global protein expression of transporter isoforms in total cell lysates, suggesting that the two mTOR signalling pathways regulate cell function by post-translational modification. Using plasma membranes isolated from cultured primary trophoblast cells subsequent to raptor or rictor silencing we show that mTORC1 or mTORC2 inhibition markedly influence the abundance of amino acid transporter isoforms at the cell surface indicating that mTOR modulates transporter trafficking to and/or from the plasma membrane. Importantly, the effect of inhibition of the two mTOR signalling pathways was isoform specific, involving LAT1, but not LAT2 or 4F2hc, of the System L isoforms and SNAT2, but not SNAT1 and SNAT4, of the System A amino acid transporter isoforms. Our imaging studies in fixed cells using immunofluorescence are consistent with the findings in isolated plasma membranes. Following simultaneous silencing of raptor and rictor, the decrease in plasma membrane expression of LAT1 and SNAT2 was more pronounced than in response to inhibition of mTORC1 and 2 signalling, separately. Notably, the plasma membrane expression of the other System L (LAT2 and 4F2hc) and System A isoforms (SNAT 1 and 4) remained unchanged following raptor+rictor silencing, suggesting that LAT1 and SNAT2 constitute the main System L and A isoforms in cultured primary human trophoblast cells. In further support of the specific nature of mTOR regulation of plasma membrane transporters, trophoblast cell serotonin uptake was unaffected by mTORC 1 and 2 inhibition.

The heavy chain 4F2hc has been shown to be necessary for LAT1 translocation to the plasma membrane; however, the mechanisms involved have not been studied in detail (Wagner et al. 2001). It is intriguing that MVM 4F2hc protein expression remained unchanged in response to mTOR inhibition, despite a decrease in MVM LAT1 expression. This could be due to the fact that 4F2hc forms heterodimers with five different light chains, in addition to LAT1 (Wagner et al. 2001). Of these, LAT2 is highly expressed in the syncytiotrophoblast MVM, and MVM LAT2 expression was not affected by mTOR inhibition (Fig. 8). Furthermore, activity of the y⁺L transporter (encoded by y⁺LAT1 and y⁺LAT2, which both form heterodimers with 4F2hc) is present in MVM (Ayuk et al. 2000). Finally, ascAT1, another 4F2hc binding partner, is expressed at the mRNA level in human placenta (Fukasawa et al. 2000), although the localization of the protein remains to be established. Thus, it is possible that 4F2hc continues to be targeted to the plasma membranes as heterodimers with light chains other than LAT1 when mTOR signalling is inhibited. This could explain lack of change in MVM 4F2hc expression under these conditions.

Regulation of trafficking of nutrient transporters between intracellular stores and the plasma membrane is a well-established mechanism by which cellular nutrient uptake is modified; however, the role of mTOR signalling in general, and mTORC1 and mTORC2 signalling in particular, in these processes remains largely unknown. The trafficking of glucose transporter 4 (GLUT4) in muscle cells and adipocytes in response to insulin has been studied extensively (Ishiki & Klip, 2005). With regard to amino acid transporters, insulin-stimulated System A amino acid transport in adipocytes and muscle cells has been reported to be due to translocation of the SNAT2 isoform from an intracellular pool to the plasma membrane (Hyde et al. 2002; Hatanaka et al. 2006b). Furthermore, Edinger and coworkers have reported that the plasma membrane trafficking of 4F2hc, the heavy chain of the System L amino acid transporter, is regulated by mTOR in cell lines (Edinger & Thompson, 2002, 2004; Edinger et al. 2003). Interestingly, 4F2hc trafficking was inhibited when cells were transfected by an mTOR mutant with an inactive kinase domain, but not by rapamycin treatment, suggesting that mTORC2 may be involved (Edinger & Thompson, 2004). In contrast to these findings, we found no evidence for mTOR-regulated plasma membrane trafficking of 4F2hc in primary human trophoblast cells. These differences may reflect tissue-specific regulation of amino acid transporters or could be due to distinct mechanisms of regulation in primary human cells as compared to cell lines.

The molecular mechanisms regulating transporter intracellular trafficking have been established in detail for two mammalian transporters: glucose transporter 4 (GLUT4) and the epithelial sodium channel (ENaC). ENaC is regulated by aldosterone through changes in the transporter cell surface expression, mediated by phosphorylation of SGK1 and Nedd4-2 (neuronal precursor cell-expressed, developmentally downregulated gene 4 isoform 2) (Flores et al. 2005). Nedd4-2 is a ubiquitin ligase that catalyses ubiquitination of ENaC localized in the plasma membrane. Because ubiquitination is recognized as a signal to target a protein for internalization, Nedd4-2 controls cell surface expression of ENaC. The molecular mechanisms linking mTORC1 and mTORC2 signalling to amino acid transporter intracellular trafficking in mammalian cells remain to be established. The primary downstream targets of mTORC1 are S6K1 and 4EBP-1, whereas mTORC2 phosphorylates Akt, PKCα and SGK1 and influences the actin skeleton (Alessi et al. 2009). Thus, in analogy to the regulation of ENaC, it is possible that mTORC2 regulates amino acid transporter trafficking by the phosphorylation of SGK1 and Nedd4-2. In support of this possibility, Hatanaka and coworkers recently demonstrated that membrane trafficking of the System A amino acid transporter isoform SNAT2 is regulated by Nedd4-2-dependent ubiquitination in 3T3-L1 adipocytes and pre-adipocytes (Hatanaka et al. 2006a). Alternatively, mTORC2 may influence amino acid transporter trafficking by regulating the actin cytoskeleton. In agreement with this hypothesis, inhibition of actin assembly with lantrunculin in 3T3-L1 adipocytes partly inhibited insulin-stimulated System A uptake (Hatanaka et al. 2006b), suggesting that growth factors regulate System A amino acid transport by both actin-dependent and actin-independent mechanisms.

The activity of key placental amino acid transporters has consistently been shown to be decreased in IUGR (Mahendran et al. 1993; Jansson et al. 1998). In addition, some studies (Jansson et al. 2002), but not all (Kuruvilla et al. 1994), have reported that placental amino acid transporters are upregulated in fetal overgrowth. This suggest that changes in the activity of placental nutrient transporters may directly contribute to abnormal fetal growth (Sibley et al. 2005; Jansson & Powell, 2006). Indeed, animal experiments support the hypothesis that changes in placental amino acid transporter activity mediates, at least in part, the effect of changes in maternal nutrition (Jansson et al. 2006; Jones et al. 2009; Coan et al. 2010; Rosario et al. 2011), or placental blood flow (Jansson & Persson, 1990) on fetal growth. For example, inhibition of placental mTOR and down-regulation of placental amino acid transport precede the development of IUGR in response to maternal protein restriction in the rat (Jansson et al. 2006; Rosario et al. 2011), which is consistent with, but does not prove, a cause-and-effect relationship. Thus, information on the regulation of trophoblast amino acid transporters is critical to the understanding of how important pregnancy complications develop. The findings in the current study suggest that changes in placental mTORC1 and mTORC2 signalling could have marked effects on placental amino acid transport and may be involved in the pathophysiology of abnormal fetal growth. We have proposed that the placenta functions as a nutrient sensor altering placental nutrient transfer and fetal growth in response to changes in the ability of the maternal supply line to deliver nutrients and oxygen to the placenta (Jansson & Powell, 2006). Because placental mTOR signalling is inhibited in IUGR (Roos et al. 2007; Yung et al. 2008) and in light of the findings in this study, it is possible that placental mTOR signalling represents the molecular correlate to a placental nutrient sensor, which links maternal nutrition and fetal growth.

Glossary

BCH

2-amino-2-norbornane-carboxylic acid

4EBP-1

4E-eukaryotic initiation factor binding protein-1

ENaC

epithelial sodium channel

4F2hc

4F2 cell-surface antigen heavy chain

hCG

human chorion gonadotropin

5-HT

5-hydroxytryptamine

IUGR

intrauterine growth restriction

LAT

large neutral amino acid transporter

MeAIB

methyl-aminoisobutyric acid

mTOR

mammalian target of rapamycin

mTORC1 and 2

mammalian target of rapamycin complex 1 and 2

MVM

microvillous plasma membrane

Nedd4-2

neuronal precursor cell-expressed, developmentally downregulated gene 4 isoform 2

raptor

regulatory associated protein of mTOR

rictor

rapamycin-insensitive companion of mTOR

SGK1

serum and glucocorticoid-regulated kinase 1

siRNAs

small interfering RNAs

S6K1

p70 S6 kinase

SNAT

sodium-dependent neutral amino acid transporter

Author contributions

Conception and design of the experiments carried out at University of Texas Health Science Center at San Antonio: F.J.R., T.J. and T.L.P.; collection, analysis and interpretation of data: F.J.R., T.J. and T.L.P.; drafting the article or revising it critically for important intellectual content: F.J.R., Y.K., T.J., and T.L.P. All authors approved the final version of the manuscript.

Supplementary material

Supplemental Fig. S1A

Supplemental Fig. S2A

Supplemental Fig. S3A

Supplemental Fig. S4

Supplemental Fig. S5