LC3-dependent EV loading and secretion (LDELS) promotes TFRC (transferrin receptor) secretion via extracellular vesicles

ABSTRACT

LC3-dependent EV loading and secretion (LDELS) is a secretory autophagy pathway in which the macroautophagy/autophagy machinery facilitates the packaging of cytosolic cargos, such as RNA-binding proteins, into extracellular vesicles (EVs) for secretion outside of the cell. Here, we identify TFRC (transferrin receptor), one of the first proteins found to be secreted via EVs, as a transmembrane cargo of the LDELS pathway. Similar to other LDELS targets, TFRC secretion via EVs genetically requires components of the MAP1LC3/LC3-conjugation machinery but is independent of other ATGs involved in classical autophagosome formation. Furthermore, the packaging and secretion of this transmembrane protein into EVs depends on multiple ESCRT pathway components and the small GTPase RAB27A. Based on these results, we propose that the LDELS pathway promotes TFRC incorporation into EVs and its secretion outside the cell.

Abbreviations: ATG: autophagy related; ESCRT: endosomal sorting complexes required for transport; EV: extracellular vesicle; EVP: extracellular vesicle and particle; ILV: intralumenal vesicle; LDELS: LC3-dependent EV loading and secretion; LIR: LC3-interacting region; MVE: multivesicular endosome; RBP: RNA-binding protein; TMT: tandem mass tag; TFRC: transferrin receptor.

Introduction

While ATG (autophagy related) proteins are essential for the lysosomal degradation of cytosolic material, they also play important roles in non-autophagic processes, including cellular secretion [1,2]. Genetic loss-of-function studies demonstrate that ATGs are required for the efficient secretion of inflammatory cytokines [3], release of bactericidal enzymes and tissue repair factors [4], exocytosis of autophagy cargo receptors during lysosome inhibition [5] and extracellular vesicle (EV) secretion [6,7]. Recently, we described a secretory autophagy pathway termed LC3-dependent EV loading and secretion (LDELS) that facilitates the packaging of cytosolic cargoes, most notably, RNA-binding proteins (RBPs) into EVs for secretion outside the cell [8]. This pathway utilizes a pool of LC3 localized at single membrane endosomes rather than double membrane autophagosomes. Hence, LDELS is akin to other autophagy-related pathways involving the conjugation of ATG8 to single membranes (CASM) [9], including LC3-associated endocytosis (LANDO) [10]. In addition, endosomal microautophagy (eMI) has been implicated in the turnover of cytosolic components and integral membrane proteins. In eMI, cytosolic proteins are engulfed via intralumenal budding mediated by the endosomal sorting complexes required for transport (ESCRT) machinery, thereby resulting in cargo incorporation into intralumenal vesicles (ILVs) [11–14].

Because LANDO and eMI both play important roles in the trafficking and clearance of membrane proteins, we hypothesized that the LDELS may similarly specify integral membrane proteins for EV loading and secretion. Here, we define a previously unrecognized role for the autophagy machinery in EV-mediated secretion of the integral membrane protein transferrin receptor (TFRC). Similar to LDELS secretion of RBPs, TFRC secretion requires the LC3-conjugation machinery but no other ATGs necessary for classical autophagosome formation. We also demonstrate that Atg8-family members directly bind to the cytosolic domain of TFRC, which may contribute to the packaging of this transmembrane protein into EVs. Finally, we delineate that TFRC secretion via EVs functionally requires ESCRT pathway components and the small GTPase RAB27A, both of which have been previously implicated in EV biogenesis and cargo loading [15–17]. Given the similarities between the mechanisms of TFRC secretion and EV-mediated release of RBPs, we propose that the LDELS pathway promotes the extracellular secretion of TFRC via EVs.

Results

ATG7 is genetically required for the secretion of transmembrane proteins in extracellular vesicle and particle (EVP) fractions

We recently described a secretory autophagy pathway termed LDELS that employs the LC3-conjugation machinery to load RBPs and small RNAs into EVs [8]. To gain a broader understanding of putative substrates and cargoes secreted by the LDELS pathway, we further scrutinized the tandem mass tag (TMT)-based quantitative proteomic data from this study. We focused on identifying membrane proteins enriched in the 100,000 g conditioned media fractions (100 K) from wild-type (WT) HEK293T cells versus those deficient for ATG7 or ATG12. Importantly, although these 100 K fractions are enriched for small EVs (exosomes), they contain a broader array of nanoparticles and proteins, collectively termed extracellular vesicles and particles (EVPs) [18]. From a total of 851 proteins enriched in EVPs from WT cells versus ATG7-deficient cells and ATG12-deficient cells, 30 proteins were classified as plasma membrane proteins by gene ontology analysis (GO) and also contained membrane-spanning regions (Figure 1(a), Figure S1, Table S1). Amongst these, the transmembrane proteins TFRC, IGF2R, and LAMP2, stood out on the basis of their high quantitative mass spectrometry intensity values, an indication of relative peptide abundance (Table S1). Interestingly, these proteins have all previously been detected in EVs [19–21]. In fact, TFRC was the first protein identified in EVs [22,23] and its secretion is proposed to be a crucial step in the maturation of reticulocytes into erythrocytes [24].

Figure 1.

ATG7 is genetically required for the secretion of transmembrane proteins via extracellular vesicles. (A) Volcano plot of the proteins identified in 100,000 g EVP-enriched fractions for WT and ATG7^−/− HEK293T cells quantified by TMT mass spectrometry in Leidal et al. 2020 [8]. TMT-labeled proteins are plotted according to their -log10 P values as determined by two-tailed t-test and log2 fold enrichment (WT/ATG7^−/−; n = 4). Gray dots: proteins not relatively enriched in EVPs from WT or ATG7^−/− cells identified with P > 0.05 and/or log2 fold change between −0.5 and 0.5. Green dots: proteins significantly enriched in EVPs from WT cells relative to ATG7^−/− cells. Red dots: proteins significantly enriched in EVPs from ATG7^−/− cells relative to WT cells. Yellow dots: membrane proteins identified by gene ontology (GO) analysis. (B) 100,000 g EVP fractions (100 K) from WT and ATG7^−/− cells were collected, normalized for protein concentration, and immunoblotted to detect the endogenous levels of the indicated proteins (n = 3). (C) Quantification of TFRC levels in EVPs from ATG7^−/− (cyan) cells relative to WT (gray) (mean ± s.e.m.; n = 3). Statistical significance calculated by one-way ANOVA coupled with Tukey’s post hoc test. (D) Schematic detailing the differential centrifugation protocol employed to isolate small EVP-enriched fractions from cell cultures. (E) Whole cell lysate (WCL) and fractionated conditioned media (CM) collected from serum-starved HEK293Ts. CM was subjected to serial differential ultracentrifugation to recover large EVs (10,000 g; 10 K) and small EVPs (100,000 g; 100 K). Equal volumes of fractionated CM were probed for the indicated proteins alongside WCL normalized to protein content of the 100 K fraction (n = 3). (F) Quantification of TFRC in the indicated fractions of CM from serum-starved cells relative to WCL (mean ± s.e.m.; n = 3). Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test. (G) Representative immunoblots of EVs from CM separated via linear sucrose density gradient ultracentrifugation, fractionated and immunoblotted to detect TFRC, LC3, and CD63 (n = 3). (H) Percent of total secreted TFRC, LC3, and CD63 detected in individual linear sucrose gradient fractions (mean ± s.e.m.; n = 3). (I) Representative immunoblots of EVs immunopurified from concentrated CM fractions using a normal mouse IgG isotype control or an antibody against TFRC (n = 3).

To corroborate our prior proteomic data, we first evaluated the secretion of TFRC, IGF2R, and LAMP2 in EVP fractions from WT and ATG7 deficient HEK293T cells via immunoblotting (100 K, Figure 1(b-c)). These analyses confirmed ATG7 to be necessary for the efficient release of TFRC, IGF2R, and LAMP2 in EVPs, with TFRC showing the most significant reduction in secretion upon loss of ATG7. Because cells secrete many different types of EVs and nanoparticles including large EVs bud from the plasma membrane (microvesicles, ectosomes) and small EVs that originate from the endosomal system (exosomes), we next performed serial differential centrifugation of conditioned media to determine whether TFRC was enriched in specific fractions (Figure 1(d)). Although 10,000 g (10 K) fractions corresponding to large EVs contained modest levels of TFRC, this membrane protein was highly enriched, along with LC3-II in small EVP populations isolated at 100 K (Figure 1(e-f)). Further purification of 100 K fractions via linear sucrose density gradient flotation demonstrated that TFRC co-fractionates with LC3 and the EV-associated tetraspanin CD63 at buoyant densities consistent with small EVs (Figure 1(g-h)). Finally, we employed antibodies targeting the extracellular domain of TFRC to specifically immuno-isolate EVs from the 100 K fraction of conditioned media. Immunoblotting of these immuno-isolated EVs revealed that LC3 specifically co-purified with TFRC, but not immunoglobulin controls (Figure 1(i)), suggesting that LC3 resides within TFRC⁺ EVs. Overall, these results demonstrate that TFRC is secreted extracellularly in association with small EVs via an ATG7-dependent mechanism.

TFRC secretion in EVs requires the LC3-conjugation pathway but is independent of other components involved in classical autophagosome formation

To determine whether autophagy pathway components are generally required for TFRC release in EVPs, we examined the secretion of this transmembrane protein in cells deficient for the early initiation factors ATG14 and RB1CC1/FIP200 as well as ATG7. To control for potential differences in EV production among these ATG deficient cells [8], we normalized the protein content of EV lysates. Importantly, analyses of 100 K EVP fractions revealed that, in contrast to ATG7, the initiation factors ATG14 and RB1CC1/FIP200 were completely dispensable for TFRC secretion in EVPs (Figure 2(a-b)). Additionally, EVP secretion of TFRC and LC3-II was reduced upon siRNA depletion of the LC3-conjugation components ATG3 and ATG5 (Figure 2(c-d)). These results support that TFRC secretion in EVPs, similar to the RNA-binding protein cargo of the LDELS pathway, requires LC3 processing and lipidation but is independent of other components necessary for classical degradative autophagy. Autophagic flux assays demonstrated that TFRC did not accumulate within cells as a result of impaired lysosomal acidification in nutrient replete conditions or following starvation-induced autophagy (Figure 2(e)). When taken together with our quantitative mass spectrometry data (Figure 1(a), Figure S1), TFRC secretion via EVPs requires LC3 processing and lipidation but is independent of other mediators of classical degradative autophagy. These findings are in line with our previous results observed for RBPs secreted via the LDELS pathway [8].

Figure 2.

TFRC (transferrin receptor) secretion in EVs requires the LC3-conjugation pathway but is independent of other components mediating classical autophagosome formation. (A) Whole cell lysate (WCL; top) and 100,000 g EV fractions (100 K; bottom) from the indicated cell types were collected, normalized for protein concentration and immunoblotted to detect the endogenous levels of the indicated proteins (n = 3). Individual lanes shown are from the same representative immunoblot. (B) Quantification of TFRC and LC3-II levels in EVs from the indicated ATG-deficient cell lines relative to WT (mean ± s.e.m.; n = 3). Statistical significance calculated by one-way ANOVA coupled with Tukey’s post hoc test. (C) WCL (top) and 100 K fractions (bottom) from equal numbers of wild-type HEK293T cells transfected with non-targeting (NT) control siRNA or siRNAs targeting ATG3 and ATG5 and immunoblotted for the indicated proteins (n = 3). (D) Quantification of TFRC and LC3-II levels in EV fractions from cells treated with siRNAs targeting the indicated proteins relative to cells treated with NT control siRNA (mean ± s.e.m.; n = 3). Statistical significance calculated by one-way ANOVA coupled with Tukey’s post hoc test. (E) HEK293Ts cultured in serum-starved or complete media and treated with DMSO or 20 nM bafilomycin A₁ (Baf A1) for 18 h were then lysed and immunoblotted for TFRC and the indicated proteins (n = 3). Quantifications of TFRC, SQSTM1/p62, and LC3-II relative protein levels (values normalized to GAPDH) in WCLs from cells cultured in the indicated conditions relative to complete, vehicle-treated media are shown. (F) WCL (top) and 100 K fractions (bottom) from serum-starved HEK293Ts treated with vehicle or 20 nM BafA1 for 16 h were collected and immunoblotted to detect the indicated proteins (n = 3). (G) Quantification of TFRC in EV fractions from BafA1-treated cells relative to vehicle control (mean ± s.e.m.; n = 3). Statistical significance calculated by an unpaired two-tailed t-test.

Remarkably, pharmacological inhibition of lysosomal acidification using bafilomycin A₁ (BafA1) enhanced TFRC and LC3 secretion in 100 K EVP fractions (Figure 2(f-g)). We recently described a pathway, called secretory autophagy during lysosome inhibition (SALI), which promotes the secretion of autophagy cargo receptors via extracellular nanoparticles in response to endolysosomal dysfunction [5]. However, the robust secretion of TFRC in cells lacking ATG14 and RB1CC1, both of which are functionally required for SALI, argues against a major role for this pathway in extracellular release of TFRC. Overall, based on these results, we propose that LDELS or an LDELS-like pathway promotes the extracellular secretion of TFRC via EVs.

Atg8-family proteins bind TFRC via a cytoplasmic domain LIR motif

In the LDELS pathway, LC3 directly binds to and mediates the packaging of specific RBP cargoes into EVs, such as scaffold-attachment factor B (SAFB), for secretion outside the cell [8]. To determine whether TFRC is packaged into EVs through similar mechanisms, we first tested for TFRC interaction with the various Atg8-family orthologs. Co-immunoprecipitation assays revealed that TFRC specifically bound to multiple Atg8 orthologs (Figure 3(a)), showing a preference for interaction with LC3A, LC3B, and LC3C. ATG8 proteins frequently bind their substrates via LC3-interacting region (LIR) motifs [25]. Because LC3 is localized to the interior of EVs [8], we postulated that the cytoplasmic tail of TFRC would be accessible for interaction with LC3. Bioinformatic analyses [26] predicted a LIR motif within the cytoplasmic tail of TFRC coinciding with its YTFR internalization motif, a region previously shown to directly interact with the ATG8 ortholog GABARAP [27] (Figure 3(b)). Therefore, we mutated this putative LIR within a soluble, recombinant protein comprising the cytoplasmic N-terminal domain of TFRC fused to luciferase. Simultaneous mutation of tyrosine-20 and phenylalanine-23 to alanine (Y20A, F23A; AA) within the TFRC cytoplasmic domain disrupted binding to both LC3B and GABARAP (Figure 3(c)), as well as the other Atg8-family members (Figure S2A). This suggests that LC3 at least partly interacts with TFRC through a LIR in the cytoplasmic domain of this transmembrane protein. To evaluate the impact of this LIR mutation on secretion, we generated cells that stably express full length TFRC (TFRC-HA; WT); mutants containing substitutions in critical LIR residues (TFRC^Y20A,F23A-HA; AA); or mutants deleted for all but four amino acids of the cytoplasmic domain (TFRC∆3-59-HA; ∆CD). Surprisingly, neither the LIR mutant nor the cytoplasmic domain truncation mutant significantly affected TFRC secretion (Figure 3(d-e)). This suggested that regions outside of the cytoplasmic domain of TFRC may also facilitate interactions with LC3, a notion further supported by the binding of the TFRC cytoplasmic domain truncation mutants to LC3B and GABARAP (Figure S2B). In support, utilizing antibodies specifically directed against the epitopes located in either the cytoplasmic or extracellular domains of TFRC, we determined that both domains of TFRC exhibited partial protease protection when EVs were treated with trypsin in the absence of detergent (Figure 3(f)). Importantly, these results are consistent with TRFC being incorporated as a transmembrane protein into EVs secreted via LDELS. At the same time, they broach that TFRC exists in both canonical and reversed topologies within the membranes of EVs. Indeed, recent studies revealed a number of transmembrane proteins exhibiting unconventional or reversed topologies in EV membranes [28]. Based on these results, we postulate that although the cytoplasmic LIR promotes TFRC interaction with LC3, TRFC existing in an unconventional or reversed topology may allow LC3 to interact with cryptic LIRs of the TFRC extracellular domain (Figure 3(b)). Alternatively, other features of the TFRC extracellular domain or interactions of TFRC with other proteins able to directly bind LC3 may facilitate its packaging into EVs and secretion outside the cell.

Figure 3.

Atg8-family proteins bind TFRC via a cytoplasmic domain LIR motif. (A) HEK293T cells co-transfected with HA-tagged TFRC and MYC-tagged LC3A, LC3B, LC3C, GABARAP (GR), GABARAPL1 (GRL1), GABARAPL2 (GRL2) were lysed, immunoprecipitated with anti-MYC antibody and immunoblotted with the indicated antibodies (n = 3). (B) Domain map and the putative LIRs of TFRC highlighted in yellow. (C) Cells co-transfected with luciferase-tagged cytoplasmic domain from WT TFRC(1–63) or mutant TFRC(1–63)^Y20AF23A (AA) along with MYC-tagged LC3B or MYC-tagged GABARAP as indicated. Cells were lysed, immunoprecipitated with anti-MYC antibody and immunoblotted with the indicated antibodies (n = 3). (D) WCL (top) and 100 K fractions (bottom) from cells stably expressing HA-tagged WT TFRC, mutant TFRC^Y20AF23A (AA), or truncated TFRC lacking the cytoplasmic domain (TFRC∆3-59; ∆CD) were immunoblotted for the indicated markers (n = 3). (E) Quantification of TFRC-Ha in EVs from equal numbers of HEK293T cells expressing HA-tagged WT TFRC, mutant TFRC^Y20AF23A (AA), or truncated TFRC lacking the cytoplasmic domain (TFRC∆3-59; ∆CD) (mean ± s.e.m.; n = 3). Statistical significance calculated by one-way ANOVA coupled with Tukey’s post hoc test. (F) Representative immunoblots of the indicated proteins from untreated EVs and EVs incubated with 100 µg ml⁻¹ trypsin and/or 1% Triton X-100 (TX-100) for 30 min at 4°C (n = 3).

ESCRT machinery and RAB27A promotes TFRC loading and secretion via EVs

EV biogenesis occurs within the endosomal system as endosomal limiting membranes undergo invagination to produce intralumenal vesicles (ILVs), thus forming multivesicular endosomes (MVEs) [29]. In our studies of RBP secretion via LDELS, the biogenesis of ILVs via the LDELS pathway functionally required SMPD3/neutral sphingomyelinase 2 (sphingomyelin phosphodiesterase 3), which converts sphingomyelin to ceramide to induce inward budding at the MVB [30]. Thus, we asked whether SMPD3 was similarly required for the secretion of TFRC in EVs. However, treatment with the SMPD3 catalytic inhibitor GW4869 did not significantly reduce TFRC secretion, arguing against a specific requirement for SMPD3 for EV secretion of this transmembrane protein (Figure 4(a-b)). We next assessed the role of ESCRT machinery in EV-mediated secretion of TFRC, employing short interfering RNAs (siRNAs) to deplete key ESCRT components including PDCD6IP/ALIX, HGS and TSG101 and then evaluating the impact on TFRC release. Genetic depletion of PDCD6IP and HGS impaired both TFRC and LC3-II secretion in EVs, whereas the loss of TSG101 had a negligible impact (Figure 4(c-d)). Taken together, these data suggest that TFRC incorporation into ILVs functionally requires an ESCRT-dependent intralumenal budding pathway. Upon incorporation of resident cargo within ILVs at the MVE, fusion of these MVEs with the plasma membrane is required for secretion outside of the cell [16]. RAB27A, a small GTPase of the RAB superfamily, has been implicated in the docking of MVBs at the plasma membrane to release EVs [17,31]. Accordingly, we examined the role of RAB27A in the release of EVs containing TFRC. Short hairpin RNAs (shRNAs) targeting RAB27A strongly suppressed both TFRC and LC3-II secretion (Figure 4(e-f)). Thus, RAB27A is functionally required for the secretion of TFRC and LC3-II. Altogether, these results suggest that, in contrast to the SMPD3-dependent release of RBPs, ESCRT- and RAB27A-dependent mechanisms functionally contribute to EV loading and secretion of TFRC via the LDELS pathway. Furthermore, by uncovering a genetic requirement for RAB27A in the EV secretion of both TFRC and LC3-II, our results corroborate our originally proposed model that EVs generated via the LDELS pathway originate from MVEs or late endosomes, rather than via direct outward budding from the plasma membrane [2,8].

Figure 4.

ESCRT machinery and RAB27A promotes TFRC loading and secretion via EVs. (A) Whole cells lysates (WCL; left) and EV lysates (100 K; right) from cells treated in the absence or presence of 5 µM GW4869 for 24 h and immunoblotted for the indicated proteins (n = 3). (B) Quantification of TFRC levels in EV fractions from GW4869-treated cells relative to vehicle control (mean ± s.e.m.; n = 3). Statistical significance calculated by an unpaired two-tailed t-test. (C) WCL (top) and 100 K fractions (bottom) from equal numbers of wild-type HEK293T cells transfected with non-targeting (NT) control siRNA or siRNAs targeting ATG7, PDCD6IP, HGS, and TSG101 and immunoblotted for the indicated proteins (n = 3). (D) Quantification of TFRC and LC3-II levels in EV fractions from cells treated with siRNAs targeting the indicated proteins relative to cells treated with NT control siRNA (mean ± s.e.m.; n = 3). Statistical significance calculated by one-way ANOVA coupled with Tukey’s post hoc test. (E) WCL (top) and 100 K fractions (bottom) from equal numbers of cells stably expressing non-targeting (NT) shRNA or shRNAs that deplete RAB27A immunoblotted for indicated proteins (n = 3). (F) Quantification of TFRC and LC3-II levels in EV fractions from RAB27A depleted cells (shRAB27A) relative to controls expressing NT shRNA. (mean ± s.e.m.; n = 3).

Discussion

Here, we demonstrate that the LC3-conjugation machinery functionally contributes to the EV-mediated secretion of TFRC. Interestingly, the term “exosome” was first coined in regard to EVs harboring TFRC released during reticulocyte maturation after being intralumenally bud within the multivesicular body [22,23]. Despite TFRC being the first exosome marker ever reported [32,33], until now, the precise mechanisms directing the secretion of this transmembrane protein in extracellular vesicles have remained elusive. Whereas previously described LDELS substrates, such as the RNA-binding proteins SAFB and HNRNPK, utilized an SMPD3-dependent mechanism for secretion, TFRC secretion requires specific ESCRT-associated components including PDCD6IP and HGS. Notably, our previous studies implicated the ESCRT component CHMP4B in the intralumenal budding of LC3 [8]. Our results here further support a role for the ESCRT machinery in the secretion of cargo via the LDELS pathway. Overall, these results broach that LDELS cargoes may employ differing and substrate-specific mechanisms of intralumenal budding. Importantly, we previously demonstrated that while LC3-II is present in EVs released from a broad spectrum of cell types, the repertoire of secreted RBP cargo varies from cell type to cell type [8]. Accordingly, an important follow-up question is delineating whether and how cell differentiation state as well as the response to specific stressors influence the secretion of TFRC, RBPs, and other LDELS cargo.

Our results here also expand the protein cargo secreted via LDELS beyond our original description of RBPs by demonstrating that this secretory autophagy pathway mediates the incorporation of transmembrane proteins into EVs released outside of the cell. An important unanswered question is delineating how the LDELS pathway modulates disease progression and physiological functions in vivo. In this regard, EVs containing TFRC have been previously implicated in reticulocyte development and therapeutic drug delivery [34–36]. Furthermore, treatment with excess iron increases the ubiquitination and subsequent trafficking of TFRC to the MVE [37]. Given the importance of MVEs in the secretion of cargo via EVs, one can speculate that exposure to excess iron may increases the secretion of TFRC via EVs. Further supporting this notion, a recent study demonstrated that oxidative stress induced in reticulocytes caused increased shedding of TFRC, which was proposed to mitigate oxidative stress via eliminating excess TFRC and ferric iron via EVs [34]. Similarly, cells that rely heavily on autophagy and EV secretion for cellular remodeling, such as reticulocytes [38], may utilize LDELS as a means to remove TFRC. Determining whether and how TFRC secretion via LDELS influences the shedding of TFRC within EVs during oxidative stress and reticulocyte development as well as more broadly regulates cellular iron uptake and homeostasis remains important topics for future study.

Materials and methods

Cell culture

HEK-293T (ATCC, CRL-3216) were cultured in Dulbecco’s Modified Eagle Medium (DMEM), high glucose, pyruvate (Gibco, 11,995,040) supplemented with 10% fetal bovine serum (FBS; Atlas Biologicals, F-0500-D), and 100 units/mL penicillin and 100 µL/mL streptomycin (Gibco, 14,140,163). This cell line was tested for Mycoplasma contamination (Sigma Aldrich, MP0035).

Unless otherwise indicated, CM and EVP preparations were collected following 24 h incubation in DMEM containing all supplements except FBS. For autophagy flux assays, cells were incubated with 20 nM bafilomycin A₁ (Sigma, B1793) as indicated for 16 h before CM collection and lysis. Treatment with 5 µM GW4869 (Cayman, 13,127) or vehicle (dimethylsulfoxide; Sigma Aldrich, 472,301) in serum-free DMEM for 24 h was used to inhibit SMPD3 activity.

Plasmid constructs

The following vectors are available or were obtained from Addgene: pcDNA3.2/DEST/hTfR-HA (69,610; Robin Shaw), pBABE-puro-GFP-LC3 (22,405; Jayanta Debnath). To generate pcDNA3-TFRC-HA, TFRC-HA was amplified with flanking primers (Fwd: atgcgaattcgccaccatgatggatcaagctagatcagcat; Rev: atgcgcggccgcttacgcgtaatctgggacgtcg) from pcDNA3.2/DEST/hTfR-HA and sub-cloned into pcDNA3 between the EcoRI and NotI sites. To generate pcDNA3-TFRC(∆3-59)-HA, TFRC-HA was amplified from pcDNA3.2/DEST/hTfR-HA with the same reverse primer as above but with a forward primer containing the first two amino-terminal codons of TFRC (Met-Met) in its overhang and sequence alignment beginning at Lys-60 (Fwd: atgcgaattcgccaccatgatgaaaaggtgtagtggaagtatct). This product was then sub-cloned into pcDNA3 between the EcoRI and NotI sites. To generate pcDNA3-TFRCY20AF23A-HA, site-directed mutagenesis of pcDNA3-TFRC-HA was performed via QuikChange PCR. First, overlapping primers carrying the desired Y20A mutation (Fwd: gaaccattgtcagctacccggttcag; Rev: ctgaaccgggtagctgacaatggttc) were used to amplify pcDNA3-TFRC-HA and templated plasmid was eliminated via DpnI digestion. Subsequently, individual clones were sequenced to verify mutagenesis of the desired site. This process was then repeated with the pcDNA3-TFRC^Y20A-HA construct and overlapping primers carrying F23A mutation (Fwd: gtcagctacccgggccagcctggctcggc; Rev: gccgagccaggctggcccgggtagctgac) in order to generate pcDNA3-TFRC^Y20AF23A-HA.

To generate pcDNA3-TFRC(1-63)-Luc, luciferase (Luc) was amplified with flanking primers (Fwd: atcggaattcatggaagacgccaaaaacataaagaaaggc; Rev: atcggcggccgcctactattacaatttggactttccgcccttcttgg) from pRetroX-Tight-Pur-Luc (Takara, 632,104) and sub-cloned into pcDNA3 between EcoRI and NotI sites to generate pcDNA3-Luc. Subsequently, TFRC(1–63) was amplified with flanking primers (Fwd: actgaagcttgccaccatgatggatcaagctagatca; Rev: actgggatccccactacacctttttggttttgtgacattg) from pcDNA3-TFRC-HA and sub-cloned into pcDNA3-Luc between the HindIII and BamHI sites. To generate pcDNA3-TFRC(1–63)^Y20A,F23A-Luc, the same process was repeated as above this time using pcDNA3-TFRC^Y20A-HA as template DNA for amplification with the same flanking primers, thus generating pcDNA3-TFRC(1–63)^Y20A-Luc. Subsequently, site-directed mutagenesis of pcDNA3-TFRC(1–63)^Y20A-Luc was performed via QuikChange PCR. Overlapping primers carrying the desired F23A mutation (Fwd: gtcagctacccgggccagcctggctcggc; Rev: gccgagccaggctggcccgggtagctgac) were used to amplify pcDNA3-TFRC(1–63)^Y20A-Luc and template plasmid was eliminated via DpnI digestion. Individual clones were sequenced to verify mutagenesis of the desired site.

To generate pBABE-puro-TFRC-HA, TFRC-HA was amplified with flanking primers (Fwd: actgaccggtgccaccatgatggatcaagctagatca; Rev: actggtcgacttacgcgtaatctgggacgtcg) from pcDNA3-TFRC-HA and sub-cloned into pBABE-puro-GFP-LC3 between AgeI and SalI, effectively replacing the GFP-LC3 open reading frame. To generate pBABE-puro-TFRC^Y20AF23A-HA, the above was repeated, this time using pcDNA3-TFRC^Y20AF23A-HA as template DNA for amplification. To generate pBABE-puro-TFRC∆3-59-HA, TFRC(∆3-59)-HA was amplified with flanking primers (Fwd: actgaccggtgccaccatgatgaaaaggtgtagtgga; Rev: actggtcgacttacgcgtaatctgggacgtcg) from pcDNA-TFRC∆3-59-HA and sub-cloned into pBABE-puro-GFP-LC3 as above. All constructs were verified by sequencing.

RNA interference

For transient siRNA-mediated knockdown, cells were transfected with siRNA using DharmaFECT no. 1 (Horizon Discovery, T-2001-03) according to the manufacturer’s instructions. ON-TARGETplus smart pools against ATG7 (10,533; Horizon Discovery, L-020112-00-0005), ATG3 (64,422; Horizon Discovery, L-015375-00-0005), ATG5 (9474; Horizon Discovery, L-004374-00-0005), PDCD6IP/ALIX (10,015; Horizon Discovery, L-004233-00-0005), TSG101 (7251; Horizon Discovery, L-003549-00-0005), HGS (9146; Horizon Discovery, L-016835-00-0005) and non-targeting siRNAs (Horizon Discovery, D-001810-10-20) were purchased from Dharmacon. To generate stable knockdowns, cells were transduced with pLKO.1 lentiviral vectors expressing shRNAs targeting RAB27A (Sigma Aldrich, TRCN0000005295 [shRab27A¹]; Sigma Aldrich, TRCN0000005296 [shRab27A²]), and non-targeting shRNA (Sigma Aldrich, SHC002).

Retroviral and lentiviral packaging, infection, and selection

Retroviral pBABE expression vectors were packaged and target cells were transduced according to established protocols. Briefly, Phoenix-AMPHO cells (gift from C. McCormick, Dalhousie University) were seeded and transfected with retroviral vectors using polyethylenimine. Virus-containing CM was collected 2 d after transfection and clarified using a 0.45-µM filter. Prior to infection, virus-containing medium was diluted 1:2 in DMEM growth medium and the mix was supplemented with polybrene to a final concentration of 8 µg ml⁻¹. Subsequently, the viral transduction mix (5 ml total volume/10 cm culture dish) was incubated with HEK293T cells for 24 h. Cells were selected 2 d post-transduction with 1 µg ml⁻¹ puromycin for a minimum of 2 d. To package lentivirus, HEK293T cells were seeded and co-transfected with the packaging vectors pRSV-Rev, VSV.G, and pMDLg, and individual pLKO.1 transfer vectors. Virus collection, infection, and puromycin selection of stable cell pools were carried out as above.

EV preparation and characterization

EVs were purified according to standard differential centrifugation protocols [39]. Briefly, cells seeded in 15-cm culture dishes at approximately 70% confluence were incubated with serum-free DMEM for 24 h. CM was collected and centrifuged serially at 200 g for 5 min to pellet cells, 2,000 g for 10 min to pellet cellular debris and apoptotic bodies, 10,000 g for 30 min to pellet large EVs, and 100,000 g in an ultracentrifuge for 3 h to pellet small EVPs. Crude EVP pellets were then gently triturated in PBS (ThermoFisher, 14,190,144), diluted further in PBS (12 ml), and ultracentrifuged for an additional 70 min at 100,000 g to generate EVP preparations for further analysis as described below. Importantly, for all comparisons of EVs or EVPs between experimental conditions, results from individual cohorts were normalized as indicated on the basis of total cell number or whole-cell lysate (WCL) protein concentration to ensure that quantification was not confounded by seeding differences.

Sucrose density gradient separation was utilized to generate highly purified EV preparations and to analyze the co-fractionation of TFRC, LC3-II, and EV marker proteins on linear sucrose gradients. Briefly, the 100,000 g EV pellets generated via differential centrifugation as described above were thoroughly resuspended in 100 µl 10% sucrose (Sigma Aldrich, S0389) solution and gently layered onto a continuous 5–60% sucrose gradient formed on a gradient station (BioComp Instruments) and then ultracentrifuged at 210,000 g for 18 h. Subsequently, 1 ml fractions from the gradient were top unloaded, weighed and diluted in 10 ml of PBS. The diluted fractions were spun at 100,000 g for 70 min and the pellets were resuspended in RIPA lysis buffer for analysis by immunoblotting.

For protease protection assays, equal amounts of EVs were resuspended in PBS or PBS containing 1% Triton X-100 (EMD-Millipore, M1122980101) in the absence or presence of 100 µg ml⁻¹ trypsin (Sigma Aldrich, T1426-100 MG) for 30 min at 4°C. Subsequently, the reactions were stopped by the addition of 2X protein sample buffer and the lysates were subjected to immunoblotting.

Antibodies

Rabbit anti-IGF2R/M6PR (Abcam, ab124767; 1:5,000), mouse anti-LAMP2 (Santa Cruz Biotechnology, sc-18,822; 1:200), mouse anti-TFRC/CD71 (Santa Cruz Biotechnology, sc-32,272; 1:200), rabbit anti-LC3 (Cell Signaling Technology, 3868; 1:1,000), rabbit anti-CD63 (Abcam, ab134045; 1:1,000), mouse anti-TSG101 (BD Transduction Laboratories, 612,696; 1:500), rabbit anti-TFRC/CD71 (Cell Signaling Technology, 13,113; 1:1,000), mouse anti-GAPDH (Millipore, MAB374; 1:5,000), rabbit anti-SQSTM1/p62 (Cell Signaling Technology, 39,749; 1:1,000), rabbit anti-HA (Cell Signaling Technology, 3724; 1:5,000), mouse anti-MYC (Millipore, M5546; 1:5,000), goat anti-firefly luciferase (Abcam, ab181640; 1:1,000), mouse anti-TFRC/CD71 (Santa Cruz Biotechnology, sc-65,882; 1:200), rabbit anti-ATG7 (Cell Signaling Technology, 8558; 1:1,000), mouse anti-PDCD6IP/Alix (Cell Signaling Technology, 2171; 1:1,000), rabbit anti-HGS (Abcam, ab155539; 1:1,000), rabbit anti-RAB27A (Cell Signaling Technology, 69,295; 1:1,000), peroxidase-AffiniPure donkey anti-rabbit IgG (H + L) (Jackson, 711–035-152; 1:5,000), peroxidase-AffiniPure donkey anti-mouse IgG (H + L) (Jackson, 715–035-150; 1:5,000), peroxidase-AffiniPure donkey anti-goat IgG (H + L) (Jackson, 705–035-147; 1:5,000).

Immunoblotting

To generate WCL and EVP lysates, cells and EVP preparations were lysed in RIPA buffer (Sigma Aldrich, R0278; 50 mM Tris, pH 8.0, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS) plus protease inhibitor cocktail (Sigma Aldrich, P8340) and PhosSTOP (Roche, 4,906,845,001). Lysates were cleared by centrifugation, quantified by BCA assay (Thermo Fisher, 23,225), mixed with sample buffer, resolved by SDS–polyacrylamide gel electrophoresis (PAGE) and transferred to polyvinylidene fluoride membrane (Bio-Rad, 1,620,177). Membranes were blocked for 1 h in 5% milk in PBS with 0.1% Tween 20 (Sigma Aldrich, P1379), incubated in primary antibody overnight at 4°C, washed, incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (1:5,000; Jackson, 715–035-150 and 711–035-152), washed and visualized via enhanced chemiluminescence (Millipore, WBLUF0100). Immunoblots were quantified by densitometry using Fiji and Image Lab.

Immunoprecipitation

For immunoprecipitation, cells transiently transfected with MYC-tagged LC3-family members and TFRC-HA, TFRC(1-63)-Luc, or mutants thereof were lysed 24 h post-transfection in NP40 buffer (25 mM Tris, pH 8.0, 150 mM NaCl, 1% NP40 (IGEPAL; Sigma Aldrich, I8896), 5% glycerol, 2 mM EDTA, 2 mM EGTA, 10 mM β-glycerophosphate, 10 mM NaF) plus protease inhibitor cocktail and PhosSTOP. Immune complexes were then captured by incubation with anti-MYC magnetic beads (Thermo Scientific, 88,842) for 2 h at 4°C and then washed five times with NP40 buffer plus inhibitors, eluted with sample buffer and analyzed by immunoblotting.

Immuno-isolation of EVs

The following antibody-conjugated magnetic beads were employed for immuno-isolation of EVs: rabbit IgG isotype control magnetic bead conjugate (Cell Signaling Technology, 8726S) mouse anti-CD63 (Abcam, ab8219), mouse anti-TFRC (Abcam, ab218326; ProSci, 33–825; NSJ, V3481), and mouse IgG2b kappa isotype control (eBioscience, 14–4732-82) and rabbit anti-TFRC magnetic bead conjugate (LSBio, LS-C171694). Briefly, 10 μg of magnetic beads conjugated with antibody against TFRC or normal mouse IgG (isotype control) was mixed with EVP fractions purified from approximately 4.6 × 10⁸ cells by differential centrifugation, resuspended in 100 μL of PBS (containing 0.1% BSA), split equally between 2 to 4 microcentrifuge tubes (25 μL each), with each tube resuspended up to 100 μL PBS (containing 0.1% BSA). Each sample was incubated with antibody-bound beads for 2 h at 4°C with end-over-end rotation to allow capture of EVs. Next, bound EVs were captured using the DynaMag-2 magnetic stand (Invitrogen, 12321D), washed 3x with ice-cold PBS (containing 0.1% BSA) plus one final wash with PBS, and then eluted with 40 µL of RIPA lysis buffer and sample buffer. Samples were then resolved via SDS–PAGE and immunoblotted for TFRC, LC3, and EV marker proteins.

Statistics and reproducibility

Statistical analyses were performed using Prism GraphPad 8 software. Groups were compared using unpaired Student’s t-test where indicated or one-way ANOVA followed by Tukey’s post-hoc test for multiple comparisons. The sample size was chosen on the basis of the size of the effect and variance for the different experimental approaches. P values of less than 0.05 were considered to be significant.

Funding Statement

This work was supported by the Canadian Institutes of Health Research; National Cancer Institute [CA213775, CA126792]; National Institute on Aging [AG057462]; Samuel Waxman Cancer Research Foundation; and Mark Foundation for Cancer Research.

Disclosure statement

JD is a member of the Scientific Advisory Board of Vescor Therapeutics, LLC.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2022.2140557.