Glutamine deprivation alters the origin and function of cancer cell exosomes

Abstract Exosomes are secreted extracellular vesicles carrying diverse molecular cargos, which can modulate recipient cell behaviour. They are thought to derive from intraluminal vesicles formed in late endosomal multivesicular bodies (MVBs). An alternate exosome formation mechanism, which is conserved from fly to human, is described here, with exosomes carrying unique cargos, including the GTPase Rab11, generated in Rab11‐positive recycling endosomal MVBs. Release of Rab11‐positive exosomes from cancer cells is increased relative to late endosomal exosomes by reducing growth regulatory Akt/mechanistic Target of Rapamycin Complex 1 (mTORC1) signalling or depleting the key metabolic substrate glutamine, which diverts membrane flux through recycling endosomes. Vesicles produced under these conditions promote tumour cell proliferation and turnover and modulate blood vessel networks in xenograft mouse models in vivo. Their growth‐promoting activity, which is also observed in vitro, is Rab11a‐dependent, involves ERK‐MAPK‐signalling and is inhibited by antibodies against amphiregulin, an EGFR ligand concentrated on these vesicles. Therefore, glutamine depletion or mTORC1 inhibition stimulates release from Rab11a compartments of exosomes with pro‐tumorigenic functions, which we propose promote stress‐induced tumour adaptation.


Expanded View Figures
▸ Figure EV1. Exosomes are made inside Rab11 compartments in Drosophila secondary cells.

A
Bar chart showing the number of ILVs within specific size ranges in the non-acidic compartments of SCs expressing CD63-GFP. Each dot represents number of vesicles within each size range in a single compartment. The size of all ILVs was measured from 3D-SIM images of at least three compartments for three SCs per gland from three different animals (producing at least 27 dots per size range). B Bar chart shows average number of large (diameter greater than two micrometres), non-acidic dense-core granule compartments (identified using DIC microscopy) in SCs from w 1118 , YFP-Rab11 gene trap, SC>Btl-GFP and SC>CD63-GFP male flies. Note that the number of these compartments is increased by CD63-GFP expression, as previously reported (Redhai et al, 2016), but not by the other transgenes. Counts are for three SCs per gland from 10 different animals. C Bar chart shows average number of large (diameter greater than two micrometres), LysoTracker Red ® -positive, acidic compartments (LELs) in different genetic backgrounds. The number of acidic compartments is slightly increased by the YFP-Rab11 gene trap and SC > Btl-GFP. Counts are for three SCs per gland from 10 different animals. D Bar chart shows average size of the largest acidic LEL compartment in these genetic backgrounds. This is slightly reduced by the YFP-Rab11 gene trap and increased by CD63-GFP. Sizes were measured for three SCs per gland from 10 different animals. E Transmission electron micrograph of a w 1118 non-transgenic fly SC in the AG epithelium with boxes showing location of enlarged images, marked by coloured dots.
In the magnified green dot image, most large compartments lacking dense-core granules are equivalent to the acidic compartments seen in live fluorescence imaging (Corrigan et al, 2014). Arrows mark exosome-sized (30-150 nm diameter) vesicles. Arrowheads mark complex membranous structures, characteristic of lysosomal compartments. In the magnified red and yellow dot images, large non-acidic, dense-core granule compartments have previously been reported to be Rab11-positive (Redhai et al, 2016). Arrows mark ILVs. Arrowhead in each image marks dense-core granule, which is surrounded by filamentous material (Corrigan et al, 2014). F-K Wide-field fluorescence images of living fly secondary cells (SCs) and paired accessory glands (AGs; K), and confocal images of fixed AG lumens (G-I). Focal planes indicated in Fig 1A schematics. Acidic compartments are marked by the vital dye LysoTracker ® Red (magenta); GFP-and YFP-tagged constructs are shown in yellow.
(F) Basal view of an SC expressing a YFP-Rab11 gene trap, with cell outline approximated by dashed white circle. Rab11 compartment highlighted by box contains three large ILVs (arrowhead) with Rab11 marking their outlines. (G) Transverse view of an SC from a fly in which CD63-GFP is expressed specifically in SCs under GAL4/UAS control. CD63-GFP is present on both the limiting and ILV membranes of acidic and non-acidic compartments, in addition to the apical plasma membrane (arrows in merge). Image is also shown schematically. CD63-GFP puncta (arrows) are present in the AG lumen of these flies. (H) Transverse view of an SC expressing a YFP-Rab11 gene trap, which marks non-acidic compartments, and also the cytosol, but is not trafficked to the apical plasma membrane (arrows in merge). Image is also shown schematically. Absence of YFP-Rab11 throughout the apical plasma membrane is shown in the Z-stack in Movie EV2. YFP-Rab11 puncta (arrows) are present at low levels in the AG lumen. (I) Basal view of an SC from a fly expressing SC-specific YFP-Rab11 (yellow) under GAL4/UAS control, with cell outline approximated by dashed white circle. Boxed non-acidic compartment is magnified in Zoom; arrows highlight YFP-Rab11-positive ILVs (Merge) and puncta present at low levels (AG lumen). (J) Basal view of an SC from a fly expressing the YFP-Rab11 gene trap (yellow) and SC-specific CD63-mCherry (magenta) under GAL4/UAS control, with cell outline approximated by dashed white circle. Large non-acidic compartments with CD63-mCherry at their limiting membrane appear to exclude YFP-Rab11 from their surface (boxes 1 and 2). In the compartment lumen, co-localisation of the markers is observed (white arrowheads; box 1), but also non-overlapping fluorescence (coloured arrowheads; box 2), consistent with ILV heterogeneity in a single compartment. (K) AGs from a fly expressing nuclear GFP under the control of a well-characterised btl-GAL4 enhancer trap (Hayashi et al, 2002). Boxed region of AG epithelium containing SCs is enlarged in Zoom, with SC outlines approximated by dashed white circles. GFP localisation in binucleate SCs suggests that btl is normally expressed in SCs.

The EMBO Journal
A Bar chart showing relative levels of phosphorylation of mTORC1 downstream readouts, S6 and 4E-BP1 (measured as a ratio of p-S65-4E-BP1 to pan 4E-BP1 or of the c-phosphorylated form of 4E-BP1 relative to total 4E-BP1), following growth of cells for 24 h in glutamine-depleted (2.00 mM) versus glutamine-replete (0.15 mM) conditions. Data are from triplicate experiments, analysed as in Fig 4A. B Nanosight Tracking Analysis of EV size and number for diluted samples (normalised to cell lysate protein levels) produced from ultracentrifugation (UC) of medium from cells cultured in glutamine-replete and glutamine-depleted conditions for 24 h, as in Fig 4B. C Electron micrographs (EMs) of EV samples from glutamine-replete and glutamine-depleted cells as analysed in (B). Arrows indicate representative EVs with the characteristic cup-shaped morphology previously reported. Scale bars represent 500 nm (left panels) and 200 nm (right panels). D Comparative Western analysis of putative exosome and non-exosome markers in HCT116 cell lysate (cells) and EV preparation produced by UC from cells grown in glutamine-replete conditions (EVs). Lanes are loaded with equal amounts of protein. Note that unlike other "classical" exosome markers, Rab11a and Cav-1 are present, but not enriched in EVs, while Golgi (TGN46), ER (calnexin) and early endosome (EEA1) markers are absent. E Bar charts show changes in levels of putative exosome proteins in EVs produced by UC from cells grown in glutamine-depleted versus glutamine-replete conditions and analysed as in Fig 4B. Protein quantities in each sample were normalised to EV particle number, CD81, Syn-1 and Tsg101 in the four graphs (n = 4, except for particle number, where n = 6). All show an increase in Rab11a and Cav-1, and a decrease in CD63, while small or no change is seen for other markers. F Western blot analysis of EV preparations isolated from HCT116 cells cultured in glutamine-replete and glutamine-depleted conditions for 24 h using size-exclusion chromatography (SEC; fractions two to four). EV loading was normalised to protein levels in cell lysates. Bar chart shows relative levels of putative exosome markers normalised to cell lysate protein mass (n = 3). (F 0 ) Nanosight Tracking Analysis of EV size and number for samples produced as in (F).
Data information: Bar charts derived from at least three independent experiments and analysed by the Kruskal-Wallis test; *P < 0.05; n.s. = not significant. Bars and error bars denote mean AE SD. Significantly decreased levels are in blue and increased levels are in red.  Figure EV4. Changes in EV proteins in cell lysates and EVs following reduction in PI3K/Akt/mTORC1 signalling in HCT116 cells.

Panels show Western blot analyses of cell and EV proteins, as well as Nanosight Tracking Analysis (NTA) of EV size and number.
A Western blot analysis of lysates from HCT116 cells cultured in the presence or absence of 100 nM Torin 1 for 24 h. Total protein levels were reduced by approximately 20% following drug treatment. Bar chart shows the abundance of putative exosome proteins relative to tubulin in these lysates and relative levels of phosphorylation of mTORC1 downstream readouts, S6 and 4E-BP1 (measured as a ratio of p-S65-4E-BP1 to pan 4E-BP1). Significantly decreased levels are in blue and increased levels are in red. (A 0 ) Western blot analysis of EV preparations isolated by ultracentrifugation from HCT116 cells cultured in the presence or absence of 100 nM Torin 1 for 24 h. EV loading was normalised to protein level in cell lysates. (A″) NTA for EV samples produced as in (A 0 ). B Western blot analysis of lysates from HCT116 cells cultured in the presence or absence of 10 nM rapamycin for 24 h. Total protein levels were reduced by 7 AE 2% following drug treatment. Bar chart shows the abundance of putative exosome proteins relative to tubulin in these lysates and relative levels of phosphorylation of mTORC1 downstream readouts, S6 and 4E-BP1 (measured as a ratio of p-S65-4E-BP1 to pan 4E-BP1). Significantly decreased levels are in blue and increased levels are in red. (B 0 ) Western blot analysis of EV preparations from HCT116 cells isolated by ultracentrifugation from HCT116 cells cultured in the presence or absence of 10 nM rapamycin for 24 h. EV loading was normalised to protein level in cell lysates. See Appendix Fig S7E for analysis of SEC-isolated EVs under these conditions. (B″) NTA for EV samples produced as in (B 0 ). C With relevance to EVs shown in Fig 5D, Western blot analysis of lysates from HCT116 cells subjected to 4 days of raptor or non-targeting (NT) shRNA knockdown.
Total protein levels were not significantly altered by knockdown versus control. Bar chart shows the abundance of putative exosome proteins relative to tubulin in these lysates and relative levels of phosphorylation of mTORC1 downstream readouts, S6 and 4E-BP1 (measured as a ratio of p-S65-4E-BP1 to pan 4E-BP1).
Significantly decreased levels are in blue and increased levels are in red. (C 0 ) NTA for EV samples produced as in Fig 5D.  Figure EV5. Analysis of EV function and proteins secreted by control and knockdown HCT116 colorectal cancer cells.
A Number of DAPI-stained cells after 120-h culture in serum-depleted conditions (1% serum), following pre-incubation either with EVs from HCT116 cells cultured in glutamine-replete or glutamine-depleted conditions for 24 h, or with vehicle (PBS). Bar charts derived from three independent experiments. B Overall levels of apoptosis, measured in arbitrary units (AU), of fluorescence, and levels of apoptosis per cell (right-hand graph) detected in HCT116 cells by analysis of caspase-3 and caspase-7 activities, following treatments with EVs isolated as in (A) or with vehicle (PBS). C Growth curves for HCT116 recipient cells in 1% serum conditions following 30-min pre-incubation with 0.5 × 10 3 , 1.0 × 10 3 and 2.0 × 10 3 EVs per cell, isolated by SEC from glutamine-replete and glutamine-depleted HCT116 cells or PBS. D-F Western blot analysis of cell lysate proteins, and Nanosight Tracking Analysis (NTA) of EV size and number in D 0 and E 0 . Bar charts indicate putative exosome protein levels normalised to Tubulin on Western blots. (D) With relevance to EVs shown in Fig 6C, Western blot analysis of lysates from EV-secreting cells with and without knockdown of Rab11a in glutamine-replete and glutamine-depleted conditions. Total protein levels were increased by 1 AE 3% following knockdown. Bar chart shows the abundance of putative exosome proteins relative to tubulin in these lysates and relative levels of phosphorylation of mTORC1 downstream readouts, S6 and 4E-BP1 (measured as a ratio of p-S65-4E-BP1 to pan 4E-BP1). Significantly decreased levels are in blue and increased levels are in red.(D 0 ) NTA of EVs isolated from HCT116 cells with and without knockdown of Rab11a. (E) With relevance to EVs shown in Fig 6F, Western blot analysis of proteins isolated from EV-secreting HCT116 cells with and without knockdown of Rab7. Bar chart shows the abundance of putative exosome proteins relative to tubulin in these lysates and relative levels of phosphorylation of mTORC1 downstream readouts, S6 and 4E-BP1 (measured as a ratio of p-S65-4E-BP1 to pan 4E-BP1). Significantly decreased levels are in blue and increased levels are in red. (E 0 ) NTA of EVs isolated from HCT116 cells with and without knockdown of Rab7. (F) With relevance to EVs shown in Fig 7C, Western blot analysis of cell lysates from HCT116 cells cultured in glutamine-replete (2.00 mM) and glutamine-depleted (0.15 mM) medium for 24 h. Gel was loaded with equal protein amounts. The activity of mTORC1 was assessed via phosphorylation of S6 and 4E-BP1. Bar chart shows the abundance of selected exosome/EV proteins relative to tubulin in these lysates. Significantly decreased levels are in blue and increased levels are in red.
Data information: Growth curves were reproduced in three independent experiments and analysed by two-way ANOVA. Bar charts derived from three independent experiments and analysed by the Kruskal-Wallis test: **P < 0.01, *P < 0.05 Bars and error bars denote mean AE SD.