These authors contributed equally to this work.
Cargo-Dependent Degradation of ESCRT-I as a Feedback Mechanism to Modulate Endosomal Sorting
Version of Record online: 13 JUN 2011
© 2011 John Wiley & Sons A/S
Volume 12, Issue 9, pages 1211–1226, September 2011
How to Cite
Malerød, L., Pedersen, N. M., Sem Wegner, C. E., Lobert, V. H., Leithe, E., Brech, A., Rivedal, E., Liestøl, K. and Stenmark, H. (2011), Cargo-Dependent Degradation of ESCRT-I as a Feedback Mechanism to Modulate Endosomal Sorting. Traffic, 12: 1211–1226. doi: 10.1111/j.1600-0854.2011.01220.x
- Issue online: 11 AUG 2011
- Version of Record online: 13 JUN 2011
- Accepted manuscript online: 12 MAY 2011 11:15AM EST
- Received 23 September 2010, revised and accepted for publication 11 May 2011, uncorrected manuscript published online 12 May 2011, published online 13 June 2011
- endosomal sorting;
- Top of page
- Materials and Methods
- Supporting Information
Ligand-mediated lysosomal degradation of growth factor receptors, mediated by the endosomal sorting complex required for transport (ESCRT) machinery, is a mechanism that attenuates the cellular response to growth factors. In this article, we present a novel regulatory mechanism that involves ligand-mediated degradation of a key component of the sorting machinery itself. We have investigated the endosomal localization of subunits of the four ESCRTs—Hrs (ESCRT-0), Tsg101 (ESCRT-I), EAP30/Vps22 (ESCRT-II) and charged multivesicular body protein 3/Vps24 (ESCRT-III). All the components were detected on the limiting membrane of multivesicular endosomes (MVEs). Surprisingly, however, Tsg101 and other ESCRT-I subunits were also detected within intraluminal vesicles (ILVs) of MVEs. Tsg101 was sequestered along with cargo during endosomal sorting into ILVs and further degraded in lysosomes. Importantly, ESCRT-mediated downregulation of two distinct cargoes, epidermal growth factor receptor (EGFR) and connexin43, mutually made cells refractory to degradation of the other cargo. Our observations indicate that the degradation of a key ESCRT component along with cargo represents a novel feedback control of endosomal sorting by preventing collateral degradation of cell surface receptors following stimulation of one specific pathway.
Growth factor receptors are essential for proliferation, survival, differentiation and migration of cells. Correct feedback response of growth factor receptor activation at the plasma membrane is essential for proper cell signaling. Several lines of evidence suggest that the failure of such regulation might contribute to oncogenic transformation (1). Upon activation of growth factor receptors at the plasma membrane, the receptors are taken into the cell via endocytosis and further delivered to early endosomes, multivesicular endosomes (MVEs) and lysosomes. For each of these processes, a broad protein network has been described.
Upon entering early endosomes, endocytosed membrane receptors are either recycled back to the plasma membrane, as for the constitutively endocytosed transferrin receptor, or directed for degradation in lysosomes as is the case with epidermal growth factor receptor (EGFR) upon EGF binding. Endosomal sorting of ubiquitinated membrane proteins such as EGFR is mediated by the conserved endosomal sorting complex required for transport (ESCRT) machinery, which was originally discovered in yeast and consists of four different complexes (0, I, II and III). This sorting process and its corresponding transport machinery have been described in detail (2,3).
A functional ESCRT machinery is essential for normal cell signaling by regulating activated growth factor receptors at the plasma membrane. Consistent with this, several studies have shown that knockdown of ESCRT-0 or -I causes prolonged EGF signaling (4,5), whereas depletion of any ESCRT subunit impairs receptor degradation (4–11). In the fruit fly Drosophila melanogaster, the ESCRTs are responsible for proper Notch receptor signaling and degradation (12–14). Alternative functions of the ESCRT machinery have also been discovered, such as cell abscission during cytokinesis and cell cycle control (15–19), virus budding (20–25) and transcriptional regulation (26–28). The widespread roles of the ESCRT machinery in cellular functions make it relevant to ask how the different ESCRTs are controlled in time and space to exert their diverse functions.
For the ESCRT-I complex, several alternative Vps37 and Mvb12 subunits have been identified, suggesting that ESCRT-I can be tuned to perform distinct functions. The ESCRT-I subunit, Tsg101, has been reported to have an ‘autoregulatory’ mechanism that controls its expression level (29), and several distinct ubiquitin ligases have been reported to be involved in this regulation. The RING E3 ubiquitin ligase, Tsg101-associated ligase (Tal) ubiquitinates the C-terminal part of Tsg101, thereby directing it for proteasomal degradation (30,31). Whether Tsg101 is polyubiquitinated or not by Tal is dependent on its oligomeric state. In complex with Vps37 and Vps28, Tsg101 might be protected against polyubiquitination and hence degradation, allowing Tsg101 to sort cargo. Mahogunin, another ubiquitin ligase found to interact with Tsg101, was reported to multi-monoubiquitinate Tsg101 and regulate its function in endosome-to-lysosome trafficking (32,33). These results imply that the regulation of the sorting machinery is more complex than first anticipated.
To date, the regulation and turnover of the ESCRTs in response to cell signaling have not been clarified, and in this article we have investigated the fate of the ESCRT subunits both upon serum starvation and upon activation-dependent internalization of two distinct plasma membrane proteins, EGFR and connexin43. We show that the ESCRT-I subunit Tsg101, in contrast to subunits of the other ESCRT complexes [Hrs, EAP30 and charged multivesicular body protein 3 (CHMP3)], localizes both to the endosomal membrane and to intraluminal vesicles (ILVs) of MVEs. Surprisingly, we found that Tsg101 was further degraded in lysosomes together with its cargo upon ligand stimulation. This suggests that there exists a cargo-dependent regulation, not only of degradation of the growth factor receptors and their ligands, but also of the ESCRT machinery itself. We propose that this regulation could play a key role in normal responses to extracellular signals.
- Top of page
- Materials and Methods
- Supporting Information
ESCRT-I, in contrast to other ESCRTs, is localized in the lumen of endosomes
All ESCRTs are recruited from the cytosol to endosomes during endosomal sorting of cargo destined for lysosomal degradation. However, with exception of ESCRT-0 (34), the detailed endosomal localization of most ESCRTs has so far not been determined. To determine this, we have used HEp2 cells stably expressing Rab5(Q79L) which upon induction with CdCl2 form enlarged endosomes (typically 2–10 µm in diameter) enabling us to discriminate between the limiting membrane and interior of the endosomes by confocal microscopy. As Figure 1 shows, Hrs (Figure 1A), EAP30 (Figure 1C) and CHMP3 (Figure 1D) were found on the limiting membrane of these enlarged endosomes visualized by staining against the endosomal antigen, early embryonic antigen 1 (EEA1). Tsg101 (Figure 1B) on the other hand was detected both on the limiting membrane and in the endosomes, indicating that this protein is sorted into ILVs.
To confirm the confocal microscopy results biochemically in cells not expressing Rab5(Q79L), cytosol and membrane fractions were isolated from untransfected HEp2 cells, and the purity of the fractions was confirmed by the presence of tubulin and Lamp1, respectively. Treatment of membrane fractions with increasing concentrations of proteinase K leads to degradation of proteins not enclosed and hence protected by membranes. As shown in Figure 2A, proteinase K degraded membrane-associated Hrs, EAP30 and CHMP3, indicating that these subunits of ESCRT-0, -II and -III are all located on the limiting membrane of endosomes as observed by confocal microscopy (Figure 1A,C,D). In contrast, a major fraction of Tsg101 as well as Lamp1 (35) was resistant to proteinase K, confirming that Tsg101 is protected within the endosomal lumen, in ILVs. Additionally, the localization of Tsg101 within the lumen of MVE was verified both by immunofluorescence and biochemically using an alternative antibody recognizing Tsg101 (data not shown). The two different Tsg101 antibodies colocalize and were both found inside enlarged endosome. When comparing HEp2-Rab5(Q79L) and untransfected HEp2 cells, we detected an increased membrane-associated and ILV-localized fraction of Tsg101 (Figure S1A). Furthermore, other components of the ESCRT-I complex (Vps28 and Vps37B) were also shown to be proteinase K resistant (Figure 2B). These results imply that the whole ESCRT-I complex is able to enter into ILVs. The different ESCRT-I subunits were sensitive to proteinase K when the detergent Triton X-100 was added to dissolve the endosome membranes (Figure 2B, lane 4). To further confirm the localization of Tsg101 within ILVs, we also used an alternative protease, trypsin, to measure Tsg101 protease protection, and obtained similar results as with proteinase K (Figure S1B). Moreover, Tsg101 was also found to be protease protected in additional cell lines such as PC-3 and U2OS cells (Figure S1C).
To confirm the results from immunofluorescence and biochemical experiments, pre-embedding immunoelectron microscopy was performed to show the localization of Tsg101 at the ultrastructural level. As seen in Figure 2C,D, HeLa cells transiently transfected with green fluorescent protein (GFP)-Tsg101 showed labeling of anti-GFP inside both early MVEs (Figure 2C) and lysosomes (Figure 2D) confirming the immunofluorescence and biochemical data.
The current understanding of MVE biogenesis implies that Hrs (36,37), Tsg101 (38) and SNX3 (39) are important for sequestering ILVs in MVE biogenesis. We therefore assessed Tsg101 localization in cells where ILV formation was prohibited by depletion of SNX3 or Hrs in HEp2-Rab5(Q79L) cells (Figure 3A). Compared with control transfected cells, endosomes in SNX3 or Hrs-depleted cells exhibited significantly less endosomal Tsg101 (Figure 3B,C). To confirm these data biochemically, a proteinase K assay was performed in Hrs and SNX3 knockdown cells (Figure S2) and we found that less Tsg101 was protected in ILVs in cells depleted of Hrs and SNX3. This indicates that Tsg101 localization to ILVs is dependent on proper formation of internal vesicles of MVEs. Together, these data confirm that Tsg101 is indeed able to enter into ILVs of control cells and could suggest that this occurs via the same route as cargo sorting.
Endosomal localization of Tsg101 is serum regulated
In many experimental protocols, cells are typically serum starved prior to stimulation with growth factors such as EGF in order to obtain clearer changes upon incubation with ligands. Interestingly, we observed that enlarged endosomes were almost devoid of intraluminal Tsg101 after 24 h of serum starvation (Figure 4A,B), in agreement with previous observations (7). Likewise, HEp2 cells exhibited significantly less membrane-associated Tsg101 upon serum depletion compared with cells cultured in 10% fetal calf serum (FCS) (Figure 4C), thus confirming our observations made by confocal microscopy in HEp2-Rab5(Q79L) cells. Interestingly, the ratio of membrane-associated Tsg101 versus proteinase K-protected Tsg101 was not significantly changed upon serum starvation, suggesting that ILV sequestration of Tsg101 is serum independent. In line with this notion, we observed by electron microscopy that MVEs contain similar numbers of ILVs irrespective of whether the cells were cultured with or without serum (Figure S3A). Furthermore, the endosomal marker CD63 was detected inside ILVs both when cells were cultured with and without serum (Figure S3B). Together, these observations indicated that endosomes were not depleted of Tsg101 due to prohibited ILV formation during serum depletion. On the other hand, serum contains several growth factors and other molecules that can activate cell surface receptors prior to sorting by the ESCRT machinery. In the absence of serum, these cargoes are not internalized and thus Tsg101 is not recruited to endosomes.
Tsg101 follows cargo into ILVs
Connexin43, a transmembrane protein forming channels called gap junctions that connect neighboring cells (40), represents an interesting candidate for serum-regulated cargo, as internalization of connexin43 gap junctions is impaired during serum starvation (41). Furthermore, Tsg101 and Hrs have been shown to mediate lysosomal sorting of connexin43, suggesting an ESCRT-dependent regulation of connexin43 (42,43). Accordingly, we detected connexin43 inside Rab5(Q79L)-enlarged endosomes (Figure 5A), although, curiously, connexin43 localized closer to the limiting membrane than Tsg101 (see Discussion). Furthermore, in cells depleted of connexin43 by RNA interference (Figure 5B), significantly less Tsg101 was localized to the enlarged endosomes (Figure 5C,D), indicating that Tsg101 is indeed recruited into ILVs by its cargo. The endosomes were not completely devoid of Tsg101 in connexin43-depleted cells, which is consistent with the fact that ESCRTs sort a variety of cell surface proteins. Retransfecting connexin43-depleted cells with GFP-connexin43 significantly restored the endosomal localization of Tsg101, which confirmed that connexin43 directly recruits Tsg101 into endosomes. In addition, the total cellular level of Tsg101 was unchanged upon depletion of connexin43 (data not shown). Next, we investigated whether serum starvation regulates the endosomal localization of connexin43 as previously observed for Tsg101 (Figure 4A,B). As shown in Figure 5E,F, significantly less connexin43 was detected in endosomes upon serum starvation compared with cells cultured in 10% FCS. These results indicate that connexin43 could be one of the serum-regulated cargoes of Tsg101.
To further explore the cargo-dependent recruitment of Tsg101 into ILVs of MVEs, we used the most studied cargo of the mammalian ESCRT machinery, namely EGFR. We stimulated cells with Alexa647-EGF, which activates EGFR and facilitates its internalization and endosomal sorting for lysosomal degradation. HEp2-Rab5(Q79L) cells were serum starved for 24 h and stimulated with serum-containing medium with or without Alexa647-EGF for increasing time at 37°C (Figure 6A,B). By quantifying the intra-endosomal Tsg101 intensities of confocal microscopy images, we observed a significant EGF-facilitated recruitment of Tsg101 into endosomes compared with control cells stimulated only with serum-containing medium.
To investigate the importance of cargo for ILV localization of Tsg101, ubiquitination and thus endosomal sorting of EGFR were prevented by depleting HEp2-Rab5(Q79L) cells of the E3 ubiquitin ligases Cbl-b and c-Cbl, known to mediate ubiquitination and degradation of EGFR (44,45). To verify the functional effect of the Cbl knockdown, ubiquitination (Figure S4A) and degradation of EGFR (Figure S4B) were examined, as these are the processes that are established to be affected by depletion of Cbl. The endosomal localization of Tsg101 was investigated by confocal microscopy in control cells compared to Cbl-b and c-Cbl small interfering RNA (siRNA) transfected cells and showed that significantly less Tsg101 was recruited into endosomes in Cbl-depleted cells (Figure 7B,C). Moreover, EGF was not able to facilitate recruitment of Tsg101 to endosomes in Cbl-depleted cells as observed in control cells where EGFR is properly ubiquitinated (Figure 7B,C). These observations further support that Tsg101 is sequestered along with cargo into ILVs during endosomal sorting.
Ubiquitination of cargo is important for sorting into ILVs of MVE, and ubiquitination of Tsg101 could also be involved in the translocation of Tsg101 into ILVs. To explore the involvement of ubiquitin in Tsg101 localization, we investigated the role of Mahogunin, an ubiquitin ligase reported to ubiquitinate Tsg101 (46). HEp2-Rab5(Q79L) cells were depleted for Mahogunin and Tsg101 localization was compared in control and Mahogunin-depleted cells (Figure S5). We found that less Tsg101 localized to ILVs in Mahogunin-depleted cells, which could imply that Tsg101 is ubiquitinated. However, as we were unable to detect direct ubiquitination of Tsg101 under our assay conditions, the observed effect of Mahogunin depletion could alternatively be due to ubiquitination of other proteins involved in this process.
Tsg101 sorted into ILVs is degraded in lysosomes
The localization of Tsg101 in ILVs raised the question whether this component of the sorting machinery might be transported along with cargo to the lysosomes. To investigate whether MVEs appeared devoid of Tsg101 upon serum starvation due to lysosomal degradation of the already ILV-localized Tsg101, HEp2-Rab5(Q79L) cells were serum starved in the presence or absence of inhibitors of lysosomal degradation (leupeptin + NH4Cl). As illustrated in Figure 8A, endosomes exhibited significantly higher levels of Tsg101 upon inhibition of lysosomal degradation, almost reaching the levels of endosomes of cells cultured in 10% FCS. Thus, it seems that Tsg101 follows serum-regulated cargo to lysosomes for simultaneous degradation. This is not replaced by cytosolic Tsg101 as endocytosis of the cargo is strongly impaired in the absence of serum. Interestingly, degradation of connexin43 during serum starvation has also been reported to be counteracted by lysosomal inhibitors (41). In addition to lysosomal degradation, proteins may also be degraded by proteasomes and autophagy. However, inhibiting proteasomal degradation by lactacystin (Figure 8A) or autophagy by 3-methyladenine (data not shown) did not affect the serum-regulated localization of Tsg101 in endosomes. Thus, the endosomal Tsg101 seems to be controlled by lysosomal degradation rather than proteasomal or autophagic activity. By blocking lysosomal degradation using leupeptin and NH4Cl in normal HEp2 cells, we also detected increased membrane-associated and proteinase K-resistant Tsg101 (Figure 8B), indicating that a significant amount of Tsg101 is sorted into ILVs during endosomal sorting of serum-regulated cargo. Likewise, HEp2 cells serum starved for 24 h expressed less cellular Tsg101 compared with control cells (grown with 10% FCS), and the effect of serum starvation was significantly counteracted by blocking lysosomal degradation (Figure 8C). The expression of cellular Tsg101 shown in Figure 8C was not changed at the translational level as we did not detect any changes in mRNA expression in 10% FCS-treated cells compared to serum-starved cells (data not shown). In summary, these experiments indicate that Tsg101 is sorted for lysosomal degradation and that this occurs to some extent even in the absence of serum.
To measure the kinetics of Tsg101 degradation in more detail, HeLa (Figure 9A,B) and HEp2 (data not shown) cells were treated with the protein synthesis inhibitor cycloheximide for different time points and then analyzed for Tsg101 content. In contrast to the short-lived protein Aurora B (47) and the long-lived protein EGFR (48), the half-life of Tsg101 was found to be approximately 12 h. As expected from Figure 8, blocking lysosomal degradation by leupeptin and NH4Cl prolonged the half-life of Tsg101 (Figure 9C).
We examined lysosomal degradation of Tsg101 in cells actively sorting EGFR or connexin43. To this end, cells were treated with cycloheximide in order to prevent synthesis of new proteins, before stimulated with EGF (Figure 10A,B) or the protein kinase C activator 12-O-tetradecanoylphorbol 13-acetate (TPA), which is known to cause internalization and ESCRT-dependent degradation of connexin43 (42,43,49). The cells were incubated for 4 h with either agonist in the absence or presence of the lysosomal inhibitors leupeptin and NH4Cl. As expected, EGFR and connexin43 were efficiently degraded in agonist-stimulated cells (Figure 10). Interestingly, the same was observed for Tsg101. The proteins were degraded in lysosomes as the degradation was significantly impaired in the presence of leupeptin and NH4Cl. Because of the short half-life of connexin43 (50–52), increased expression was observed after preventing lysosomal degradation for 4 h (Figure 10C,D). In summary, it seems that during ESCRT-mediated sorting of cargoes such as EGFR or connexin43, ESCRT-I is sequestered along with cargo into ILVs and transported to lysosomes for degradation. This could represent a novel mechanism for regulating endosomal sorting of different cargoes destined for lysosomal degradation.
Lysosomal degradation of Tsg101 represents a novel feedback regulatory mechanism in endosomal sorting
The cargo-dependent degradation of Tsg101 raised the question whether different cargoes might affect each other's degradation through downregulation of Tsg101. The cell's capacity to degrade two cargoes (EGFR and connexin43) sequentially was therefore investigated. All the experiments were performed in the presence of cycloheximide thereby omitting replacement of any degraded Tsg101. HeLa cells were stimulated with either EGF or TPA for 2 h in 10% FCS-containing medium and then further treated with 10% FCS with and without TPA or EGF for another 2 h. As shown in Figure 11A,B, EGFR and connexin43 were significantly degraded after 4-h treatment. However, if the cells were first stimulated with EGF for 2 h and then with TPA for another 2 h, significantly less connexin43 was degraded compared to cells stimulated with TPA alone (compare lanes 3 and 4). Vice versa, EGFR was less efficiently degraded in cells pretreated with TPA (lane 2 versus lane 5). In cells simultaneously treated with both ligands, connexin43 was efficiently degraded, whereas EGFR was not (lane 6). In light of our previous observations that Tsg101 and cargo were simultaneously degraded in lysosomes, the second cargo is most probably inefficiently sorted for lysosomal degradation due to lower Tsg101 levels. To test this further, we investigated degradation of EGFR and connexin43 in PAE cells stably expressing EGFR or not. When the cells were simultaneously stimulated with EGF and TPA, less connexin43 was degraded in EGFR-expressing PAE cells compared with control PAE cells not expressing EGFR (Figure 11C). We therefore propose that the purpose of degrading an ESCRT component together with cargo could represent a novel mechanism to regulate the endosomal sorting process mediated by the ESCRTs.
- Top of page
- Materials and Methods
- Supporting Information
Ligand-mediated lysosomal downregulation of cell surface receptors is a well-described mechanism of negative feedback regulation of cell signaling. In this work, we have uncovered a novel mechanism of cargo-mediated feedback regulation, namely the lysosomal degradation of a key component of the endosomal sorting machinery itself.
Proteins comprising the ESCRT machinery exert additional cellular functions besides endosomal sorting, such as regulating transcription, virus budding and cytokinesis (recently reviewed in 3). The many roles of ESCRTs might explain the different distribution of the proteins between cytosol and membranes observed in this and previous studies (36,52–54,55). By investigating the endosomal localization of ESCRT subunits biochemically and by confocal microscopy, we found that Hrs, as already described (34), Tsg101, EAP30 and CHMP3 were detected at the limiting membrane of the early endosome. This localization is in accordance with the current understanding of ESCRT-mediated endosomal sorting, where the ESCRTs are sequentially recruited from the cytosol to endosomes by associating with endosomal lipids and ubiquitinated cargo.
Tsg101 and two other ESCRT-I proteins (Vps28 and Vps37B) were also found within the lumen of endosomes, suggesting that the whole ESCRT-I complex is sorted into ILVs of MVEs. The proteinase K assay indicated that Tsg101 was located both at the limiting membrane and within ILVs (i.e the proteinase K-resistant fraction), whereas the confocal images showed Tsg101 staining predominantly in ILVs. As the confocal images were acquired at settings below pixel saturation, the very intense staining of luminal Tsg101 therefore limited detection of the relatively fainter Tsg101 staining at the limiting membrane. Furthermore, the apparently higher endosomal localization of Tsg101 detected by immunofluorescence compared with biochemical assays could also be explained by the fact that we detected increased membrane-associated and ILV-localized Tsg101 in HEp2-Rab5(Q79L) cells compared with normal HEp2 cells (Figure S1). Rab5(Q79L) is predicted to increase the synthesis of PtdIns3P on endosomes (53), which facilitates recruitment of PtdIns3P-binding proteins such as Hrs, and thus indirectly Tsg101. It is also likely that preparation of membrane fractions using homogenization and ultracentrifugation partly destroys the endosomes, which then become leaky and causes underestimation of the amount of membrane-protected proteins. Accordingly, we observed a modest degradation of Tsg101 and Lamp1 by the highest concentration of proteinase K used (Figure 2A). Investigating endosomal localization of Tsg101 using confocal microscopy did not discriminate between Tsg101 in ILVs completely abscissed and only invaginated, and thus the method may overestimate the true amount of Tsg101 in ILVs.
Previous proteomic analyses of exosomes identified Tsg101 (54) and thus the specific intra-endosomal localization of ESCRT-I components was not completely unexpected. But how is ESCRT-I sequestered into ILVs and what is the purpose of this localization? In line with previously published data (7), we found that cargoes, such as EGFR and connexin43, recruited Tsg101 to the limiting membrane of endosomes. Consequently, inhibiting sequestration of EGFR by depleting Cbl-b and c-Cbl (44,45) simultaneously decreased the EGF-facilitated recruitment of Tsg101 to ILVs. Furthermore, less endosomal Tsg101 was also observed in Cbl-depleted cells without EGF stimulation suggesting that Cbl also ubiquitinates other, yet unidentified, cargoes of Tsg101 than EGFR, probably explained by the broad target specificity of Cbl (reviewed in 55). On the other hand, Cbl was recently reported to control endosome fusion, which could also explain why endosomes exhibit less Tsg101 in Cbl-depleted cells (56). It is also interesting to note that the lack of a fourth ESCRT-I subunit, Mvb12, triggers sorting of the Tsg101 homolog Vps23 into ILVs in yeast (57), and we cannot exclude the possibility that mammalian cells may contain alternative ESCRT-I complexes with differential abilities to enter into ILVs.
The limited colocalization between Tsg101 and cargoes such as connexin43 and EGFR within the lumen of Rab5(Q79L) endosomes was somewhat surprising, but control experiments with antibody switching indicated that it was not an artifact of the fluorescence microscopy (data not shown). We assume that transmembrane ILV cargoes (EGFRs and connexin43) could behave differently than soluble cargoes (Tsg101) because only the latter will escape out of the ILVs once they are permeabilized by lipases in the MVE lumen.
Cargo recognized and sorted by the ESCRTs is sequestered into ILVs whose content is degraded upon fusion with lysosomes. Likewise, we observed simultaneous lysosomal degradation of Tsg101 and EGFR or connexin43 in cells stimulated with EGF or TPA, respectively. This combinatorial degradation of cargo and ESCRTs could represent a novel way to regulate endosomal sorting. To test this idea further, we sequentially stimulated cells with EGF and TPA, or vice versa. In contrast to previous reports (58,59), we did not observe EGF-stimulated degradation of connexin43, most probably because of cell-specific differences (60). Interestingly, the sequential treatment strongly impaired degradation of the second activated cargo, indicating that there exists a ‘lag phase’ after endosomal sorting because of the low Tsg101 expression which efficiently suppresses the sorting process. The reduced degradation of the second cargo could, in theory, be because of the saturation of the endocytic pathway by the first cargo–ligand complex, which thus would prevent ligand-induced internalization of the second cargo. However, we observed the same degree of internalization of EGFR or connexin43 in cells pretreated with TPA or EGF as in untreated cells (data not shown).
During massive stimulation of the cell by growth factors, it could be beneficial to downregulate one signaling pathway at a time. Correspondingly, it seems that the cell executes only one Tsg101-dependent process at a time as we have observed that mitotic cells very poorly endocytose and traffic EGFR. Moreover, prolonged or massive stimulation with growth factors could deplete the pool of cell surface receptors, implying that the cell is non-responsive to the specific stimulus until the receptors have been replaced by transcription and translation. Simultaneous treatment with EGF and TPA caused degradation of connexin43 but not EGFR, in agreement with previous data showing that TPA strongly reduced EGF binding to cell surface receptors (61,62). The inhibitory effect of TPA on EGF receptor association seems to be transient as sequential treatment of TPA and then EGF promoted EGFR degradation.
It is striking that only ESCRT-I, and not other ESCRTs, is degraded in response to receptor activation. Efficient regulation of the endosomal sorting process is probably obtained by targeting an early step, such as ESCRT-0 or -I. In contrast to Tsg101, we did not detect degradation of Hrs during EGF stimulation of cells (data not shown), which could perhaps be explained in terms of evolution, because ESCRT-0 is less evolutionarily conserved compared with the other ESCRT complexes (63). Alternatively, ESCRT-0 and -II are recruited to endosomes by binding phosphoinositides (64,65) and their endosomal association might be indirectly controlled by the turnover of these lipids.
Cargo-mediated degradation of endosomal Tsg101 could represent an alternative way to modulate cellular Tsg101 expression. Work from several laboratories has shown that the expression of Tsg101 is strictly regulated, exemplified by the observation that cells stably expressing mCherry-Tsg101 scarcely express endogenous Tsg101 (17). The levels of endosome-associated Tsg101 seem to be regulated by several pathways. Whereas Tal-mediated polyubiquitination of Tsg101 facilitates dissociation from endosomes into cytosol and subsequent proteasomal degradation (30,31), this study suggests that endosomal Tsg101 is also regulated by lysosomal degradation upon ligand stimulation. The tight control of Tsg101 expression could be associated with the many cellular roles of Tsg101 in endosomal sorting, transcriptional regulation and cytokinesis.
In conclusion, we have identified an unexpected downregulation of the ESCRT-I complex during endosomal sorting of cargo destined for lysosomal degradation such as EGFR or connexin43. The fast turnover of Tsg101 during endosomal sorting could serve to strictly control the expression of Tsg101 in response to external stimuli. The functional consequence of such regulation would be that ligand-mediated degradation of one cell surface protein makes the cell transiently refractory to degradation of other membrane proteins. This could function as a safeguard that restricts degradation to those membrane proteins that are activated by external stimuli.
Materials and Methods
- Top of page
- Materials and Methods
- Supporting Information
Reagents and antibodies
Alexa647-EGF was obtained from Invitrogen. Leupeptin was delivered by Peptide Institute Inc. Human EGF, cycloheximide, lactacystin and protease inhibitor cocktail (#P8340) were from Sigma Aldrich, whereas NH4Cl was from MERCK. FuGENE6 was from Roche Diagnostics GmbH. All additional chemicals were of analytical grade. RNAiMax used for RNA interference (RNAi) was purchased by Invitrogen. Mouse anti-Tsg101 (ab83) and mouse anti-Aurora B were obtained from Abcam. Mouse anti-α-tubulin was from Sigma Aldrich. Human anti-EEA1 was a gift from Ban-Hock Toh (Monash University). Mouse anti-Lamp1 was obtained from Developmental Studies Hybridoma Bank (University of Iowa). Rabbit antibodies raised against Hrs, EAP30, CHMP3 and Vps28 have previously been described (4,5,66,67). Guinea pig anti-ESCRT-I was prepared by Eurogentec. Rabbit anti-connexin43 has previously been described (60). Sheep anti-EGFR was obtained from Fitzgerald, mouse anti-transferrin receptor was from Zymed Laboratories, goat anti-SNX3, rabbit anti-c-Cbl and mouse anti-Cbl-b were from Santa Cruz. Rabbit anti-Hrs has been previously described (66). Rabbit anti-GFP was a gift from Terje Johansen, Tromsø University, Norway. All secondary antibodies used for confocal microscopy and horseradish peroxidase (HRP) labeled secondary antibodies used for western blotting were obtained from Jacksons ImmunoResearch Laboratories. Precise Protein Gels and SuperSignal West Dura Extended Chemiluminscent substrate were from Pierce. Secondary antibodies for immunoelectron microscopy, Nano-Gold anti-rabbit and Nano-Gold anti-mouse were from Molecular Probes. Fluorescently labeled secondary antibodies used for western blotting detection using the Odyssey developer were purchased from LI-COR® Biosciences GmbH.
Cell cultures, transfections and siRNA oligonucleotides
HEp2 cells and HeLa cells were grown in DMEM containing 10% FCS, 2 mm glutamine, 100 U/mL penicillin and 100µg/mL streptomycin in a 5% CO2 incubator at 37°C according to ATCC. HEp2 cells stably expressing an inducible Rab5(Q79L) plasmid was previously described (68). Expression of Rab5(Q79L) giving rise to enlarged endosomes (2–10 µm) was induced by incubating with 15 µm CdCl2 overnight. HeLa cells stably expressing Rab5(Q79L) were also made similarly to HEp2-Rab5(Q79L) (see 68 for details) and the expression was induced by 5 µm CdCl2 (overnight). PAE cells were grown in Ham's F-12 (Cambrex Bio Science) supplemented with 10% FCS and 0.5× penicillin–streptomycin mixture. The PAE (porcine aortic endothelial) cell line expressing EGFR was grown under G418 selection (400 µg/mL). For RNA interference experiments, cells were transfected with 50 nm connexin43 Stealth Select siRNA (#HSS104123) or 75 nm Hrs siRNA (CGACAAGAACCCACACGUC) from Invitrogen or 50 nm Smart pool siRNAs against c-Cbl (#sc-29242) and 50 nm Smart pool Cbl-b (#sc-29950) obtained from Santa Cruz, 15 nm ON-TARGET Plus SNX3 siRNA (CUGAAAGAAGAAUCUAUGUUAUU) and ON-TARGET plus siCONTROL (#D-001810-01) (used as a negative control) were obtained from Dharmacon. To obtain efficient protein depletions of connexin43, Cbl-b and c-Cbl, cells were transfected, split after 3 days and the experiments were performed on the fifth day. Knockdown experiments for Hrs and SNX3 siRNA cells were split the day after transfection and experiments were performed 3 days after transfection. In rescue experiments, control RNA- and connexin43 siRNA-transfected cells were transiently transfected with 0.5 µg GFP-connexin43 using FuGENE6 during the last 24 h of RNAi treatment. HeLa cells were transiently transfected with pEGFP-Tsg101 using FuGENE6 reagents according to the manufacturer.
Confocal fluorescence microscopy
Cells grown on coverslips were rinsed twice with PBS and permeabilized [using PEM buffer (80 mm PIPES, 5 mm EGTA and 1 mm MgCl2× 6H2O, pH 6.8) containing 0.05% saponin] for 5 min on ice and fixed in 3% formaldehyde for 15 min on ice. Coverslips were incubated with primary antibodies diluted in PBS–0.05% saponin for 1 h at room temperature or overnight at 4°C. After washing using PBS–0.05% saponin (3 × 5 min), the coverslips were incubated with secondary antibodies for 1 h at room temperature. The coverslips were washed twice in PBS–0.05% saponin and once in PBS before mounting in Mowiol (Merck, Calbiochem, Darmstadt, Germany). The specimens were examined using a Zeiss LSM 510 META confocal microscope. The Zeiss lsm 510 software (version 3.2) was used to quantify the intensities of the images, all scanned using the same pinhole, offset gain and amplifier values below saturation. Endosomes with a size larger than 7 µm were not included in the quantifications. The imaris software (version 6.1.5, Bitplane Scientific software) was used to make three-dimensional reconstructions of Z-stacks illustrating the localization of ESCRT proteins within EEA1-positive endosomes.
Preparation of membrane and cytosolic fractions and proteinase K assay
HEp2 cells grown in 10 cm dishes were rinsed with PBS, scraped and pelleted at 900 ×g for 5 min (4°C). The cell pellet was resuspended in 100 µL homogenization buffer (0.25 m sucrose in 1 mm imidazole containing protease inhibitor cocktail) and passed through a 23G syringe for homogenization. The postnuclear supernatant obtained after spinning at 3500 ×g for 5 min (4°C) was transferred to an ultracentrifuge tube. The membranes were pelleted at 150000 ×g for 15 min (4°C) and dissolved in homogenization buffer (without protease inhibitor cocktail). The subsequent supernatant represents the cytosolic fraction. The protein concentration of membrane and cytosolic fractions was measured using the Bio-Rad Protein Assay according to the manufacturer's instructions (Bio-Rad Laboratories). Equal amounts of membrane fraction (35 µg) were treated with 0, 2 or 4 ng proteinase K/µg membrane protein for 10 min at 20°C. The samples were boiled in 2× loading buffer and subjected to western blotting. Tubulin and Lamp1 were used as markers of cytosolic and membrane fractions, respectively.
Cells were rinsed twice with PBS prior to lysis [in 25 mm HEPES (pH 7.2), 125 mm potassium acetate, 2.5 mm magnesium acetate, 5 mm EGTA, 0.5% NP40 and 1 mm DDT] containing protease inhibitor cocktail. Equal amounts of protein were separated by SDS–PAGE on Precise Protein Gels from Pierce and blotted onto PVDF membranes from Millipore Corporation. Immunodetection was performed either using HRP-conjugated or fluorescent-labeled secondary antibodies. Western blots were quantified using software of the Syngene or Odyssey developer.
Pre-embedding immunolabeling and monolayer epoxy embedding
Cells on coverslips were fixed in 4% formaldehyde and 0.1% glutaraldehyde in 0.1 m PHEM buffer at room temperature for 1 h and washed. Cells were permeabilized with 0.01% saponin in PHEM buffer for 15 min on ice. The coverslips were further incubated with primary antibody diluted in PHEM, 0.01% saponin, 0.5% BSA, washed five times and then incubated with secondary antibody (1 h each, room temperature). Coverslips were then silver enhanced for 7–8 min 69 gold toned, washed and postfixed for 1 h with 2% OsO4 and 1.5% K4[Fe(CN)6] in H2O. After staining with 4% uranyl acetate in 50% ethanol (30 min), the coverslips were dehydrated in graded series of ethanol and embedded in Epon. Ultrathin sections were cut (Leica EM FCS ultramicrotome) and collected on carbon grids and contrasted with 1% Pb citrate. Sections were observed at 60–80 kV in a JEOL JEM-1230 electron microscope and micrographs were recorded with a digital camera (Morada) using iTEM (SIS) software and image processing was done with Adobe Photoshop CS2.
Comparisons of differences in measured intensity levels of Tsg101 or connexin43 were in most cases based on ordinary one-sample t-tests. In the case of EGF-facilitated recruitment of Tsg101 into endosomes (Figure 4), we used analysis of variance (mixed factor models) to take into account the experimental design, with treatments handled as fixed factors and experiments as random factors. In this case, the measured intensities also caused markedly skewed distributions for the observations, and a logarithmic transform was used to satisfy the requirement of approximately normal observations. All p values are two sided.
- Top of page
- Materials and Methods
- Supporting Information
We thank Ida Gedde and Marianne Smestad for technical assistance, Terje Johansen for GFP antibody and Inger Helene Madshus for the kind gift of untransfected and EGFR-transfected PAE cells. This work was financially supported by the European Research Council, the Research Council of Norway, the Norwegian Cancer Society and the Hartmann family foundation.
- Top of page
- Materials and Methods
- Supporting Information
- Top of page
- Materials and Methods
- Supporting Information
Additional Supporting Information may be found in the online version of this article:
Appendix S1: Materials and methods.
Figure S1: Increased amounts of Tsg101 are membrane associated and sorted into ILVs upon expression of Rab5(Q79L), and the localization of Tsg101 in the MVE lumen is detected by trypsin protection in various cell lines. A) Cytosol and membrane fractions were isolated from normal HEp2 or HEp2-Rab5(Q79L) cells. Equal amounts of membrane fractions (35 µg) were treated without (lanes 2 and 5) or with (lanes 3 and 6) 4 ng proteinase K/µg membrane fraction to determine whether the relative amount of Tsg101 sorted into ILVs is affected by the expression of Rab5(Q79L) giving rise to the enlarged endosomes. The experiment was repeated twice and a representative blot is shown. B) Cytosol and membrane fractions were isolated from HEp2 cells and equal amounts of protein (35 µg) were subjected to SDS–PAGE. The membrane fraction was digested (or not) with increasing concentrations of an alternative protease, trypsin, for 15 min at 37°C. The samples were separated by SDS–PAGE and immunodetectable levels of Tsg101, Hrs, Lamp1 and tubulin were assessed. C) Equal amounts of cytosol and membrane proteins (35 µg), the latter treated (or not) with 4 ng proteinase K/µg membrane protein, were isolated from PC-3 or U2OS cells. Immunoreactive Tsg101 was detected by western blotting analysis. The purity of the cytosol and membrane fractions was verified with antibodies against the specific markers, tubulin and Lamp1. The experiments were performed at least twice, and representative results are shown.
Figure S2: Decreased amounts of Tsg101 are protease protected in cells depleted of Hrs and SNX3. Hep2 cells were transfected with siRNA against Hrs and SNX3, and split the next day. Three days after transfection, cytosol and membrane fractions were isolated and equal amounts of membrane fractions (35 µg) were treated without (−) or with (+) 4 ng proteinase K/µg membrane fraction. This was done to determine the relative amount of Tsg101 protected in ILVs in cells depleted of Hrs or SNX3. The experiment was repeated at least three times and a representative blot is shown.
Figure S3: MVE biogenesis does not require serum stimulation of cells. A) HEp2-Rab5(Q79L) cells cultured in the presence and absence of serum were examined by electron microscopy. The amounts of ILVs were approximately equal in cells grown in 10% FCS and serum-starved cells (zoomed-in areas are shown in the right column). B) To determine the distribution of the MVE marker CD63 we labeled cells with a mouse anti-CD63 antibody, followed by a bridging secondary mouse anti-rabbit IgG and 10-nm protein A gold. We detected intraluminal CD63 labeling (arrowheads) in both control and serum-starved cells. Scale bars as indicated.
Figure S4: Knockdown of c-Cbl and Cbl-b inhibits ubiquitination and degradation of EGFR upon incubation with EGF. HEp2-Rab5(Q79L) cells transfected with siRNA against c-Cbl and Cbl-b, the day after transfection cells were spilt and experiment were performed on the fourth day. Induction of endosomes was induced by adding CdCl2 the day prior to the experiment. A) Cells were incubated with or without EGF (50 ng/mL) for 10 min before hot lysis and immunoprecipitation of EGFR under denaturating conditions. The immunoprecipiatated proteins were subjected for western blotting using antibodies against ubiquitin and EGFR. The experiment was done at least three times and a representative blot is shown. B) To measure the degradation of EGF, cells were incubated with 125I-EGF (50 ng/mL) at 37°C for 15 min, before cells were washed in PBS to remove excess of 125I-EGF and further chased at 37°C for the time indicated. The ratio of degraded, recycled and internalized 125I-EGF was calculated as described in Appendix S1 and plotted against time. The data represent the average of the three independent experiments (±SE).
Figure S5: Depletion of the ubiquitin ligase Mahogunin decreases the level of Tsg101 in ILVs of MVEs. HEp2-Rab5(Q79L) cells were transfected with control or Mahogunin siRNA for 5 days. A) Cells were lysed and subjected to western blotting with antibodies against Mahogunin and actin. One representative blot is shown. B) The day after the transfection, cells were split onto coverslips and enlarged endosomes were induced on the fourth day upon transfection by adding CdCl2. To determine the distribution of Tsg101 in enlarged endosomes, cells were labeled with Tsg101 and EEA1 antibodies. The graph represents quantifications of Tsg101 localization to enlarged endosomes in control versus Mahogunin siRNA cells. Images were acquired using the same setting by the Zeiss LSM 510 META confocal microscope and the graph represents the average of the three independent experiments (±SE).
|TRA_1220_sm_s1.doc||31K||Supporting info item|
|TRA_1220_sm_f1.jpg||451K||Supporting info item|
|TRA_1220_sm_f2.jpg||500K||Supporting info item|
|TRA_1220_sm_f3.jpg||632K||Supporting info item|
|TRA_1220_sm_f4.jpg||237K||Supporting info item|
|TRA_1220_sm_f5.jpg||230K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.