Macropinocytosis drives T cell growth by sustaining the activation of mTORC1

Macropinocytosis is an evolutionarily-conserved, large-scale, fluid-phase form of endocytosis that has been ascribed different functions including antigen presentation in macrophages and dendritic cells, regulation of receptor density in neurons, and regulation of tumor growth under nutrient-limiting conditions. However, whether macropinocytosis regulates the expansion of non-transformed mammalian cells is unknown. Here we show that primary mouse and human T cells engage in macropinocytosis that increases in magnitude upon T cell activation to support T cell growth even under amino acid (AA) replete conditions. Mechanistically, macropinocytosis in T cells provides access of extracellular AA to an endolysosomal compartment to sustain activation of the mechanistic target of rapamycin complex 1 (mTORC1) that promotes T cell growth. Our results thus implicate a function of macropinocytosis in mammalian cell growth beyond Ras-transformed tumor cells via sustained mTORC1 activation.

M acropinocytosis is an evolutionarily conserved, largescale, fluid-phase form of endocytosis observed in single-celled organisms, such as Dictyostelium, as well as select cell types in multicellular organisms [1][2][3] . In Dictyostelium, macropinocytosis provides nutrients for cell growth. In mammals, macropinocytosis has been ascribed different functions including antigen presentation in macrophages and dendritic cells and regulation of receptor density in neurons [4][5][6] . More recently, macropinocytosis has been shown to regulate the growth of Ras-transformed tumor cells under conditions where the availability of extracellular amino acids (AA) is limiting [7][8][9] . Under such conditions, proteins taken up by macropinocytosis are degraded into AA, which fuel cell growth and survival by incorporation into central carbon metabolism and by activation of the mechanistic target of rapamycin complex 1 (mTORC1). However, whether macropinocytosis regulates the growth of nontransformed mammalian cells is unknown.
To determine if macropinocytosis is required for the growth of non-transformed cells, we ask if the growth of primary T lymphocytes is dependent upon macropinocytosis. T cells are an appropriate cell type with which to address this question since upon activation they enter the cell cycle rapidly, show substantial increases in cell size during the G1 phase of the cell cycle, and have high proliferative capacity. Here we report that macropinocytosis is necessary for T cell antigen receptor (TCR)/CD28induced growth of T cells even when AA are available in the extracellular space. Macropinocytosis in activated T cells provides a means of access of extracellular AA to an endolysosomal compartment necessary for sustained activation of mTORC1 that drives T cell growth. Thus, a role for macropinocytosis in mammalian cell growth is not limited to Ras-transformed tumor cells but may be a more general phenomenon involved in the growth of non-transformed cells.

T cells engage in macropinocytosis.
Macropinocytosis has not been reported previously in T cells. To determine if T cells engage in macropinocytosis, we tested the ability of murine splenic T cells to endocytose 70 kDa fluorescein-dextran (Fdex) that is used commonly as a macropinocytosis probe 3,10,11 . The size of 70 kDa Fdex limits its uptake into cells by small-scale endocytic processes such as clathrin-mediated endocytosis 11,12 . As assessed by flow cytometry, murine CD4+ and CD8+ T cells readily took up 70 kDa Fdex (Fig. 1a-c and Supplementary Fig. 1a, b). The extent of probe uptake increased upon activation of T cells with monoclonal antibodies (mAb) against the CD3 component of the TCR complex and the T cell co-stimulatory receptor, CD28, coincident with an increase in cell size. T cell fluorescence upon probe incubation was essentially abolished at 4°C (Fig. 1a and Supplementary Fig. 1a). Thus, acquisition of T cell fluorescence is a consequence of an energy-dependent endocytic process and not binding to a cell surface receptor.
In order to visualize putative T cell macropinosomes, we used fluorochrome-labeled bovine serum albumin (BSA) as an alternative probe. Similar to 70 kDa Fdex, BSA is taken up into cells primarily by macropinocytosis 7,9 . However, in contrast to 70 Fdex, BSA is retained in cells after fixation. Flow cytometric analyses confirmed uptake of fluorochrome-labeled BSA by murine CD4+ and CD8+ T cells that increased upon CD3/28 mAb stimulation coincident with an increase in T cell size ( Fig. 1d-f and Supplementary Fig. 1c, d). As shown by confocal microscopy of fixed cells, labeled BSA was identified within distinct vesicles of CD3/28 mAb-stimulated T cells that ranged in number from 1 to 10 or more per cell (Fig. 1g, h and Supplementary Fig. 1e). The size of vesicles is consistent with their identity as macropinosomes that range in size from 200 nm to 1 μm in diameter 1-3 . To determine if T cell endocytosis of BSA increased upon TCR recognition of cognate peptide-MHC ligands, we used T cells from OTII TCR transgenic (Tg) mice that are specific for ovalbumin (Ova) peptide 323-29 in complex with the I-A b , MHC class II molecule 13 . Upon adoptive transfer to wild-type mice, OTII TCR Tg T cells showed increased uptake of BSA in response to immunization of recipients with Ova protein (Fig. 1i-k). Human peripheral blood CD4+ and CD8+ T cells also took up BSA in a temperature-dependent fashion and the extent of uptake was increased upon stimulation with CD3/28 mAb (Fig. 1l-n) or the T cell mitogen phytohemagglutinin (PHA) ( Supplementary  Fig. 1f, g). Thus, endocytosis of macropinocytosis probes is not unique to murine T cells.
We used scanning electron microscopy (SEM) to identify forming macropinosomes in murine T cells. Macropinosomes at different stages of development were readily identified in both unstimulated and CD3/28-stimulated T cells, although were more frequently observed in the latter, coincident with increased cell surface ruffling (Fig. 2a, b). Macropinocytic structures in T cells strongly resembled those observed in CSF1-stimulated murine bone marrow (BM) macrophages (Fig. 2c). Macropinocytic cup formation is driven by actin polymerization, in contrast to most small-scale forms of endocytosisis [1][2][3]11 . Consistent with the identity of T cell surface structures identified by SEM as developing macropinocytic cups, staining of CD3/28-stimulated T cells with phalloidin revealed polymerized actin in loops that extended 1-2 μm from the cell surface (Fig. 2d).
To establish that T cell uptake of large molecular weight probes is mediated by macropinocytosis, we examined the effect of 5-(N-Ethyl-N-isopropyl) amiloride (EIPA), which is a highly specific inhibitor of macropinocytosis [1][2][3] . EIPA blocks the function of a plasma membrane Na+/H+ exchanger that is required for macropinocytosis 14 . We also examined the influence of the inhibitors of actin cytoskeletal dynamics, jasplakinolide in combination with blebbistatin (J/B) 15 . EIPA and J/B inhibited uptake of BSA by CD3/28 mAb-stimulated murine and human CD4+ and CD8+ T cells at concentrations previously reported to block macropinocytosis in other cell types ( Fig. 3a-h and Supplementary Fig. 2g). The same results were obtained with purified murine T cells and when 70 kDa Fdex was used as a probe ( Supplementary Fig. 2a, b). In contrast, PitStop, an inhibitor of clathrin-mediated endocytosis 16 , had no influence upon probe uptake ( Fig. 3b and Supplementary Fig. 2a, b). Uptake of BSA by unstimulated murine T cells was also blocked by EIPA and J/B but not PitStop showing that constitutive probe uptake is also mediated by macropinocytosis ( Supplementary  Fig. 2f).
Mechanism of T cell macropinocytosis. Macropinocytosis is dependent upon activation of the Ras small GTPase in most cell types 15,17,18 . In classical macropinocytosis, active Ras triggers activation of phosphatidylinositol 3-kinase (PI3K) leading to activation of Rac1 and Cdc42 Rho-family GTPases that orchestrate actin polymerization and macropinocytic cup formation 1-3 . The serine/threonine protein tyrosine kinase Pak-1 has also been implicated in cup formation and closure [1][2][3] . To further characterize T cell macropinocytosis, we tested the effect of inhibitors of these signaling molecules. Farnesyl thiosalicylic acid (FTS), which rapidly displaces H-, N-, and K-Ras isoforms from membranes resulting in their rapid degradation 19 , had no effect upon probe uptake by CD3/28 mAb-stimulated T cells (Fig. 3b). In contrast, LY294002 20 , EHT 1864 21 and IPA-3 22 , which inhibit PI3K, Rac1 and Pak1, respectively, partially blocked probe uptake ( Supplementary Fig. 2c). We also examined macropinocytosis in T cells from mice that lack the RasGRP1 guanine nucleotide exchange factor, which is required for Ras activation in this cell type 23 . Loss of RasGRP1 in T cells did not affect T cell uptake of BSA ( Supplementary Fig. 2d, e). Thus, T cell macropinocytosis differs from classical macropinocytosis in that it is independent of Ras, although shows partial dependency upon PI3K, Rac1, and Pak1. Similar results were obtained with unstimulated murine T cells with the exception that LY294002 had no inhibitory effect ( Supplementary Fig. 2f).
T cell macropinocytosis is necessary for G1 growth. T cells activated with CD3/28 mAb show major increases in cell size between 12 and 20 h post-stimulation, which corresponds to the G1 phase of the cell cycle 24 (Supplementary Fig. 3a). Therefore, to determine if macropinocytosis is necessary for G1 growth, we tested    Supplementary Fig. 3b). Overall, we observed a strong correlation between the influence of an inhibitor upon macropinocytosis and T cell growth ( Fig. 4b and Supplementary Fig. 3c). Importantly, at the concentrations used, inhibitors showed minimal toxicity during the 12-20 h incubation period. These findings support the contention that macropinocytosis is required for T cell growth even under conditions of optimal nutrient availability.

Macropinocytosis-dependent mTORC1 activation in T cells.
In AA-starved Ras-transformed tumor cells, macropinocytosis has been linked to activation of mTORC1 8,9 . Endolysosomal AA derived from macropinocytosed and subsequently digested Murine splenocytes (a-f) were unstimulated or stimulated with CD3/28 mAb for 24 h (a-c) or 20 h (d-f). 70 kDa Fdex (a-c) or BSA-Alexa 488 (d-f) probes were added to cells for the last 4 h or 8 h of culture, respectively, at the indicated temperatures. a, b, d, e Representative flow cytometry histogram plots of CD4+ T cell probe uptake and contour plots of probe uptake vs. FSC-A. c, f Mean ± 1 SEM of the ratio of Fdex (c) or BSA (f) probe uptake in stimulated vs. unstimulated CD4+ and CD8+ T cells at 37°C (a-c, n = 10 independent experiments; d-f, n = 15 and 12 independent experiments for CD4+ and CD8+ T cells, respectively). ****P < 0.0001 by Student's 1-sample 2-sided t-test. g Images show temperature-dependent uptake of BSA-Alexa 488 by CD3/28 mAb-stimulated purified murine pan T cells (probe incubation from 12 to 20 h) into structures that resemble macropinosomes. Representative images of six repeat experiments are shown. h Quantitation of the number of macropinosomes per CD4 T cell (n = 169 and 196 cells at 37°C and 4°C, respectively). i Method used to assess uptake of BSA-Alexa 647 by OTII TCR Tg T cells in vivo following immunization with whole Ova protein. j, k Representative flow cytometry contour plot of probe uptake vs. FSC-A (j) and mean ± 1 SEM of median fluorescence intensity (MFI) of Alexa 647 fluorescence of OTII TCR Tg T cells from unimmunized and immunized mice (n = 4 mice for each condition) (k). ***P < 0.001 by Student's 2-sample 2-sided t-test. l-n Human PBMC were unstimulated or stimulated with CD3/28 mAb for 20 h. BSA-Alexa 488 was added to cells for the last 8 h of culture at the indicated temperatures. l, m Representative flow cytometry histogram plots of CD4+ T cell probe uptake and contour plots of probe uptake vs. FSC-A. n Mean ± 1 SEM of the ratio of BSA uptake in stimulated vs. unstimulated CD4+ and CD8+ T cells at 37°C (n = 3 independent experiments). *P < 0.05, **P < 0.01, by Student's 1-sample 2-sided t-test. Source data are provided as Source Data file. proteins in these cells are thought to provide one of two signals required for activation of mTORC1 at the lysosomal surface 8,9,25,26 . Although in AA-starved Ras-transformed tumor cells mTORC1 activation restricts growth, under AA replete conditions, mTORC1, through phosphorylation of ribosomal protein p70 S6 kinase and other targets, promotes anabolic processes that drive growth 25,26 . Therefore, we examined the possibility that macropinocytosis is required for T cell growth because it is necessary for delivery of AA to an endolysosomal compartment at which site they promote mTORC1 activation. With the use of DQ Red BSA that exhibits fluorescence only upon digestion in lysosomes 27 , we first confirmed that macropinocytosed material in activated T cells is targeted to lysosomes (Fig. 4c, d and Supplementary Fig. 3d, e). Subsequently, by flow cytometric staining for phospho-S6, we established that mTORC1 remains active in CD4+ and CD8+ T cells at 12 and 20 h after CD3/28 mAb stimulation ( Fig. 4e and Supplementary Fig. 3g). Addition of Torin1, a potent inhibitor of mTORC1 28 Fig. 4e-h). Furthermore, EIPA inhibited activation of mTORC1 in T cells that were stimulated acutely with CD3/28 mAb but did not inhibit activation of ERK mitogen-activated protein kinases (MAPK) or NFκB in the same short time course stimulation experiments ( Supplementary  Fig. 5a-e). Importantly, Torin 1 inhibited growth of CD4+ and CD8+ T cells when added to cultures at 12 h, albeit less potently than EIPA and J/B ( Fig. 4a and Supplementary Fig. 3b). Thus, a role for macropinocytosis in T cell growth can be attributed, at least in part, to a function in sustained activation of mTORC1.  Fig. 3h) and more potently than they inhibited macropinocytosis ( Supplementary Fig. 2c), which could be explained by a role of the respective targets not only in macropinocytosis but also signaling pathways that regulate the Rheb small GTPase (Supplementary Fig. 6g) 25,26 .
mTORC1 activation by AA in T cells requires macropinocytosis. To determine if macropinocytosis-acquired AA from the extracellular space directly for mTORC1 activation or indirectly as a consequence of protein uptake, we tested the effect of NH 4 Cl upon sustained mTORC1 activation. NH 4 Cl raises intra-lysosomal pH and thereby inhibits protein degradation in lysosomes ( Fig. 5a and Supplementary Fig. 6a, b). However, NH 4 Cl had no effect upon sustained mTORC1 activation in CD4+ or CD8+ T cells when added to cultures at 12 h, thus showing that protein degradation is not necessary for this event (Fig. 5b and Supplementary Fig. 6c, d). To assess whether extracellular AA were sufficient for sustained mTORC1 activation, T cells were stimulated with CD3/28 mAb for 12 h in complete medium containing fetal calf serum (FCS), washed extensively and recultured for 2 h in minimal medium with or without AA. Reculture of T cells in media containing all 20 AA was sufficient to sustain activation of mTORC1 in CD4+ and CD8+ T cells when compared to T cells recultured in the absence of AA (Fig. 5c, d and   a b CD3/28-stim CD8+ T cells CD3/28-stim pan T cells Supplementary Fig. 6e, f). Furthermore, re-culture of T cells in media containing only L, Q, R, and S AA, which have been implicated in lysosomal activation of mTORC1 previously 26 , sustained mTORC1 activation compared to media that did not contain AA. Among these four AA, L and R are likely most important with regards mTORC1 activation. Thus, media that lack either L or R alone are unable to sustain mTORC1 activation whereas media that lack S or Q alone are able to sustain mTORC1 activation ( Supplementary Fig. 7a, b), confirming and extending previous observations 29 . Significantly, sustained activation of mTORC1 by all AA or L, Q, R, and S was blocked by EIPA and J/ B, thus indicating a requirement for macropinocytosis for AAsustained mTORC1 activation (Fig. 5c, d and Supplementary  Fig. 6e, f). Together, these data support a model in which macropinocytosis drives T cell growth through provision of free AA to an endolysosomal compartment necessary for sustained activation of mTORC1 (Supplementary Fig. 6g).

Discussion
The findings reported herein constitute the first demonstration of a role for macropinocytosis in the growth of non-transformed mammalian cells and suggest that this may be a more general phenomenon applicable to the growth of other primary cell types. Why T cells are dependent upon macropinocytosis for sustained mTORC1 activation and growth even under conditions where AA are freely available in the extracellular space remains to be determined. T cells express abundant AA transporters that could theoretically transport AA from the extracellular space to the cytosol and subsequently to lysosomes where they would promote mTORC1 activation. Two AA transporters, slc7a5 and slc1a5, involved in the transport of large neutral AA and glutamine, respectively, are required for mTORC1 activation in T cells 29,30 . However, it is unknown if these transporters are present on lysosomal membranes in T cells and if they are able to transport AA into this organelle. Instead, their role in mTORC1 activation could be consequent to their ability to transport AA from the extracellular space to the cytosol exclusively, at which site the imported AA would be detected by cytosolic AA sensors that are also known to control activation of mTORC1 26 . Potentially, activation of mTORC1 in T cells could be strictly dependent upon inputs from both cytosolic and lysosomal AA sensors. In this regard, a requirement for macropinocytosis for sustained mTORC1 activation in T cells could be explained by an  Supplementary Fig. 6b, d). c Splenocytes were stimulated with CD3/28 mAb in complete medium (RPMI plus FCS) for 12 h, washed and recultured in the indicated media for 2 h in the presence or absence of EIPA. Representative flow cytometry histograms show phospho-S6 levels in CD4+ T cells. All panels are from the same experiment. The mid-blue-dashed line indicates all AA in the absence of EIPA. d Mean±1 SEM of the fold increase in mTORC1 activation in CD4+ T cells at 14 h relative to the 0 AA control in the absence of inhibitors (n = 8 and 7 independent experiments for 20 AA and LQRS, respectively, in the absence of inhibitors; n = 4 for 20 AA and n = 3 independent experiments for LQRS and 0 AA in the presence of inhibitors). **P < 0.01 by Student's 1-sample 2-sided t-test. Source data are provided as Source Data file.

Methods
Animals. Wild-type mice were bred in house and were on a mixed 129S6/SvEv X C57BL/6 genetic background. One exception was recipient mice in in vivo experiments that were on a CD45.1 C57BL/6 background (JAX). OTII TCR Tg mice (JAX) and Rasgrp1 mutant mice (JAX) were on a C57BL/6 genetic background. Mice ranged in age from 6 weeks to 3 months. Mice of both sexes were used in experiments. All experiments performed with mice were in compliance with University of Michigan guidelines and were approved by the University Committee on the Use and Care of Animals.
Confocal microscopy. Murine splenic pan T cells or CD4+ T cells (isolated by column depletion) were stimulated with CD3 and CD28 mAb for 12 h as above before incubation in the presence or absence of BSA-Alexa 488 (0.4 mg/ml) or DQ Red BSA (50 μg/ml) at 37°C or 4°C in the presence or absence of EIPA (50 μM) for a further 4 or 8 h. Cells were harvested, washed, resuspended in PBS and sedimented for 1 h on ice at 1 g onto coverslips previously coated with 0.1% poly-L-lysine (Sigma). Cells were fixed in situ by addition of an equal volume of 4% paraformaldehyde and incubation for 30 min at room temperature. Coverslips were then washed and cells were stained with rat anti-mouse CD4 (R&D Systems, clone GK1.5), CD8α (R&D Systems, clone 53-6.7), LAMP-1 (eBioscience, clone 1D4B, cat. no. 14-1071-82, dilution 1:100), or LAMP-2 (Invitrogen, clone M3/84, cat. no. MA5-17861, dilution 1:100) mAb overnight at 4°C. The following day, coverslips were washed, blocked with 5% donkey serum for 1 h, incubated with Alexa 488-or Alexa 594-labeled donkey anti-rat secondary antibody (Thermo Fisher, cat. nos. A-21208 and A-21209, dilution 1:200), and stained with 10 μg/ml Hoechst 33258 (Thermo Fisher). For imaging of polymerized filamentous actin, cells were permeabilized with Triton X-100 and stained with Actistain-488 fluorescent phalloidin (0.1 μM; Cytoskeleton, Inc.) prior to the antibody staining steps. Coverslips were mounted in ProLong Gold anti-fade (Thermo Fisher) and images were captured on a Leica upright SP5 confocal microscope (Leica). Data showing the percentage of T cells with indicated numbers of macropinosomes are based upon counts of >150 cells.
Scanning EM. Murine BM-derived macrophages were prepared as described 31 . After 6 d culture, macrophages were stimulated for 15 min with CSF1. Murine CD4+ T cells were stimulated for 16 h with CD3/28 mAb and prepared as indicated in confocal microscopy.
Macrophages and T cells were processed for scanning EM as described 32 except that the fixed samples were dehydrated in an acetone series followed by hexamethyldisalizane. Air-dried samples were gold coated and imaged on an Amray 1900 field emission scanning electron microscope.
Flow cytometric analysis of mTORC1 and NFκB activation. Murine splenocytes were stimulated as above for the indicated times in the presence or absence of inhibitors at concentrations indicated above. Cells were harvested, fixed in 4% paraformaldehyde for 20 min at room temperature, washed with PBS supplemented with 5% FCS, and permeabilized by drop-wise addition of ice-cold, 90% methanol with gentle vortexing. Cells were washed and stained with APC-Cy7-CD4, APC-CD8α and PE-Cy7-phospho-S6 mAb (Cell Signaling Technology, clone D57.2.2E, cat. no. 34411, dilution 1:50) or anti-phospho-NFκB p65 (pS356) (Cell Signaling Technology, clone 93H1, cat. no. 3033, dilution 1:100) followed by Alexa488-labeled donkey anti-rabbit secondary antibody (Jackson Immunoresearch, cat. no. 711-545-152, dilution 1:100) and analyzed by flow cytometry. The gating strategy is illustrated in Supplementary Fig. 8. In each experiment, the effect of inhibitors upon mTORC1 activation between 12 and 20 h or NFκB activation between 12 and 14 h was calculated as a percentage of mTORC1 or NFκB activation observed in the absence of inhibitor as follows: [(MFI phospho-S6 or NFκB of CD3/28 stimulated T cells in presence of inhibitor−MFI phospho-S6 or NFκB of unstimulated T cells in absence of inhibitor)/(MFI phospho-S6 or NFκB of CD3/ 28 stimulated T cells in absence of inhibitor−MFI phospho-S6 or NFκB of unstimulated T cells in absence of inhibitor)] × 100.
For experiments that determined if AA were sufficient to sustain mTORC1 activation, splenocytes were stimulated for 12 h with CD3/28 mAb in RPMI 1640 with 10% FCS as above. Cells were then washed extensively in PBS and recultured in wells of 96 well U-bottomed plates at 1 × 10 6 cells per well in RPMI 1640 with 10% FCS, or with modified RPMI 1640 without AA (USBiological) to which different combinations of AA (Thermo Fisher or Sigma) had been added at the concentrations normally found in RPMI 1640. EIPA and J/B were added to some wells at the same concentrations indicated above. After 2 h, cells were harvested and analyzed for phospho-S6 by flow cytometry.
Statistical analysis. P values were calculated using Student's 1-sample or 2-sample 2-sided t-tests as appropriate for normally distributed data.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The flow cytometry and imaging data supporting these findings are available from the corresponding author upon request. The source data underlying Fig. 1c, h, k, n, 3d, f, h, 4a, 5e, and Supplementary Figs. 1e, g, 2a-c, e-g, 3b, e, h, 4a-d, f, 5b-e, 6b, d, f, and 7b is provided as a Source Data File.