Inhibition of the SEC61 translocon by mycolactone induces a protective autophagic response controlled by EIF2S1-dependent translation that does not require ULK1 activity

Mycolactone induces an increase in autophagic flux

We have previously shown an increase in LC3B lipidation in HeLa cells in response to mycolactone [Citation16]. To determine whether this response also occurs in disease-relevant primary cells from skin, we examined LC3B puncta formation in human dermal microvascular endothelial cells (HDMEC), which are highly sensitive to the effects of mycolactone and may play a role in disease pathology [Citation6], and in primary human fetal foreskin fibroblasts (HFFF) by immunofluorescence microscopy. In both cell types, exposure to mycolactone for 8 h induced a significant increase in cytosolic LC3B puncta compared to the DMSO vehicle (Figure 1A and 1B). Similar results were obtained with both murine embryonic fibroblasts (MEF) and HeLa cell lines, although the kinetics were slower in the latter, requiring 16 h to reach a significant increase (Fig. S1A and S1B). As expected, LC3B puncta were increased by incubation with saturating levels of either bafilomycin A₁ (or chloroquine; Fig. S1B) and co-incubation with mycolactone led to a further increase(Figure 1A,1B and S1B). A time-dependent increase in lipidated LC3B-II was detected in HFFF cells exposed to mycolactone and this was further increased in the presence of bafilomycin A₁ (Figure 1C, quantified in Fig. S1C).

Figure 1. Mycolactone induces an increase in autophagic flux in primary dermal microvascular endothelial cells and fibroblasts. Cells were exposed to mycolactone (Myco) at the minimal inhibitory concentration alongside a solvent control (0.05% DMSO, equivalent dilution to the mycolactone) for the longest timepoint tested. where included, bafilomycin A₁ (BAF) was used for 4 h at cell-specific saturating concentrations, and at the longest timepoint tested when in combination with mycolactone. PP242 was used at 1 μM for 1 h. All immunofluorescence images were obtained using a Nikon A1 confocal laser scanning microscope. For all panels, unless otherwise indicated, statistical comparisons are to the DMSO-treated cells of that particular cell type; ns = not significant; * = p< 0.05; ** = p< 0.01; *** = p < 0.001; **** = p < 0.0001. (A) HDMEC exposed to 10 ng/ml mycolactone with and without 100 nM BAF for 8 h were fixed with methanol for LC3B staining (scale bar: 50 µm). values represent the number of puncta detected per cell for at least 36–49 cells. Data are representative of 2 independent experiments (scale bar: 50 µm). (B) HFFF exposed to 31.25 ng/ml mycolactone with and without 200 nM BAF for 8 h were fixed with methanol for LC3B staining (scale bar: 50 µm). Values represent the number of puncta detected per cell for at least 63–90 cells. Data are representative of 2 independent experiments. (scale bar: 50 µm) experiments. (C) HFFF cells exposed to 31.25 ng/ml mycolactone and/or 200 nM BAF were lysed directly in gel sample buffer prior to immunoblotting. The migration relative to known molecular mass markers is shown. Data are representative of 2 independent experiments. (D) Expression of tandem-tagged LC3B in HeLa cells. Cells transiently transfected with pDEST-mCherry-eGFP-LC3B were exposed to mycolactone for 4 h before the addition of 100 nM (BAF) and incubation for a further 4 h before fixation. Values represent the number of puncta detected per cell for at least 8–10 transfected cells. Data are representative of 4 independent experiments (scale bar: 10 μm). (E) HDMEC and HFFF were exposed to mycolactone for various times or PP242, then fixed for WIPI2 staining (scale bar: 50 µm). Values represent the puncta detected in each cell for 25–50 cells. Data are representative of 2 independent experiments. (F) HDMEC were exposed to 10 ng/ml mycolactone for 8 h, or PP242 then fixed for RB1CC1 or ATG16L1 staining (scale bar: 50 µm). Values represent the puncta detected in each cell for 25–50 cells. Data are representative of 2 independent experiments. (G) HDMEC were exposed to 10 ng/ml mycolactone for 8 h then fixed for co-staining with WIPI (green) and RB1CC1 or ATG16L1 (red) (scale bar: 10 μm).

Display full size

To determine whether this observation was due to a bona fide activation of autophagy or a decrease in the turnover of autophagosomes owing to inhibition of the terminal degradative step in the pathway, we utilized a previously described [Citation31] mCherry-GFP-LC3B tandem tag to follow the fusion of autophagosomes with the lysosome. HeLa cells transfected with the tandem tag construct showed a basal level of autophagy, with detection of both red and yellow puncta in cells incubated with DMSO (Figure 1D and S1D). As expected, no puncta formation was detected in cells expressing a control mCherry-GFP (Fig. S1E). Exposure of cells transfected with a plasmid encoding the LC3B-fusion protein to mycolactone had little effect on the number of autophagosomes (yellow puncta), but significantly increased the number of LC3B-positive autolysosomes (red puncta), suggesting that mycolactone does not block fusion of autophagosomes with lysosomes or their acidification (Figure 1D). Addition of a saturating concentration of bafilomycin A₁ (Fig. S1F) reduced the number of acidified red structures while also causing an accumulation of neutral, yellow puncta, as expected. Co-incubation with mycolactone and bafilomycin A₁ slightly increased levels of acidified vesicles while significantly increasing the number of neutral, yellow autophagic vesicles compared to bafilomycin A₁ alone (Figure 1D). Taken together this data suggests that mycolactone does indeed induce an increase in autophagic flux.

As an alternative approach to assessing changes in autophagic flux we also assessed mycolactone’s ability to promote phagophore formation by following appearance of WIPI2 puncta by immunofluorescence staining [Citation32]. In both HDMEC and HFFF, WIPI2 puncta increased with time of exposure to mycolactone (Figure 1E). The response is slower than that induced by the MTOR inhibitor PP242 [Citation33], but a significant increase in puncta could be detected at 4 h in HDMEC and 8 h in HFFF cells. Similar changes were observed in MEF and HeLa cell lines (Fig. S1G and S1H). The production of early autophagosomes was further assessed by staining HDMEC cells for RB1CC1 and ATG16L1. Both markers showed an increase in puncta comparable to the positive control PP242 8 h after mycolactone addition (Figure 1F). Co-staining revealed close association of both RB1CC1 and ATG16L1 with WIPI2 in HDMEC and MEFs, suggesting they were present on the same structures (Figure 1G and S1H). RB1CC1 and ATG16L1 puncta were also detected in HFFF cells but the effect was less marked (Fig. S1J). Since all three markers are only transiently associated with autophagosomes and (unlike LC3B) are not destroyed by fusion with the lysosome, the most likely explanation for the increase in puncta detection is an increase in autophagic flux. However, a “dual effect” of an increase in autophagic flux together with inhibition of autophagosome maturation cannot be completely ruled out at this stage.

Mycolactone-induced autophagy involves the selective autophagic receptor protein, SQSTM1 and polyubiquitination

Mycolactone significantly increased the appearance of puncta staining positively for SQSTM1 in both HDMEC and HFFF cells within 8 h (Figure 2A and 2B). This was further increased in the presence of bafilomycin A₁, indicating that this observation is unlikely to be due to inhibition of lysosomal clearance. In HeLa cells SQSTM1-positive puncta accumulated over time but more slowly and did not reach statistical significance until 16 h after mycolactone addition (Fig. S2A).

Figure 2. Accumulation of SQSTM1 and ubiquitinated proteins in mycolactone-exposed cells. Cells were exposed to mycolactone (Myco) at the minimal inhibitory concentration, alongside a solvent control (0.05% DMSO, equivalent dilution to the mycolactone) for the longest timepoint tested. Where included, bafilomycin A₁ (BAF) was used at cell-specific saturating concentrations. All immunofluorescence images were obtained using a Nikon A1 confocal laser scanning microscope. For all immunoblots, the migration relative to known molecular mass markers is shown. For all panels, unless otherwise indicated, statistical comparisons are to the DMSO-treated cells of that particular cell type, ns = not significant; * = p< 0.05; ** = p< 0.01; *** = p < 0.001; **** = p < 0.0001. (A) HDMEC exposed to 10 ng/ml mycolactone with and without 100 nM BAF for 8 h were fixed with PFA for SQSTM1 staining with a mouse anti-SQSTM1 antibody (scale bar: 50 µm). Values represent the number of puncta detected per cell for 45–100 cells. Data are representative of 2 independent experiments. (B) HFFF exposed to 31.25 ng/ml mycolactone with and without 200 nM BAF for 8 h were fixed with PFA for SQSTM1 staining with a mouse anti-SQSTM1 antibody (scale bar: 50 µm). Values represent the number of puncta detected per cell for 45–100 cells. Data is representative of 2 independent experiments. (C) HeLa cells were exposed to 31.25 ng/ml mycolactone for 24 h, in the presence or absence of 100 nM BAF for the final 12 h, then fixed with methanol for SQSTM1 with a mouse anti-SQSTM1 antibody (green) and LC3B (red) staining. Nuclei were counterstained with DAPI (blue) (scale bar: 10 µm). Values are means of 3 independent experiments ± SEM. (E) HeLa cells were left untreated, exposed to mycolactone for different times (as indicated) or 50 µM chloroquine (CQ) for 12 h. The 48-h timepoint was performed without and with 50 µM CQ (for the final 12 h). Cells were harvested in RIPA lysis buffer, then soluble and insoluble fractions were prepared as described in Methods prior to immunoblotting. Data are representative of 3 independent experiments. (F) HeLa cells exposed to 31.25 ng/ml mycolactone for various times or 10 µM MG132 for 8 h were lysed directly in gel sample buffer prior to immunoblotting. Data are representative of 2 independent experiments. (G) HeLa cells were exposed to 31.25 ng/ml mycolactone for the times indicated, or 10 µM MG132 for 8 h then fixed with PFA for total ubiquitin staining with FK2 (scale bar: 10 µm). Data are representative of 2 independent experiments. (H) HeLa cells were exposed to 31.25 ng/ml mycolactone for 16 h then fixed with PFA for total ubiquitin with FK2 (green) and SQSTM1 with a rabbit anti-SQSTM1 antibody (red) staining (scale bar: 10 µm).

Display full size

We next assessed the extent of colocalization between SQSTM1 and LC3B in response to mycolactone. As expected, high levels of colocalization were observed in the presence of bafilomycin A₁, an effect also elicited by mycolactone alone (Figure 2C and S2B). Exposure to both mycolactone and bafilomycin A₁ gave rise to a further, significant increase in double-positive puncta numbers compared to either compound alone, providing additional evidence for a genuine increase in autophagic flux.

As SQSTM1 aggregates are frequently insoluble [Citation34], we analyzed both detergent-soluble and -resistant fractions of HeLa cells exposed to mycolactone by immunoblotting. Levels of soluble SQSTM1 decreased with longer exposure to mycolactone, while processing of LC3B to LC3B-II increased over time (Figure 2D). By contrast, SQSTM1 levels in detergent-resistant fractions showed a marked increase after 24-h mycolactone exposure (approximately 48 h before these cells die) [Citation16]. Addition of chloroquine to mycolactone-exposed cells prior to lysis caused a further increase of LC3B-II. Similar results were obtained in macrophage-like RAW264.7 and MEF cell lines (Fig. S2C and S2D).

To investigate the role played by polyubiquitin in mycolactone-induced selective autophagy we carried out immunoblot analysis of HeLa cells using a well-characterized pan-ubiquitin antibody, FK2. Total levels of polyubiquitin-conjugated proteins were strongly increased, as expected [Citation19], in whole cell lysates following exposure to mycolactone to levels comparable to the positive control, MG132 (Figure 2E). By immunofluorescence, ubiquitin was predominantly nuclear in control cells but shifted progressively to a more cytoplasmic location over time with mycolactone, with large numbers of cytoplasmic puncta detectable after 16–24 h (Figure 2F). While the proteasome inhibitor MG132 caused an increase in overall cytoplasmic staining, this was diffuse, and cells contained relatively few discrete puncta. Co-staining with anti-SQSTM1 antibodies revealed a high degree of overlap with cytoplasmic ubiquitin puncta after 24 h incubation with mycolactone (Figure 2G). A close association between SQSTM1 and ubiquitin was also observed in HFFF (Fig. S2E).

These experiments suggest that SQSTM1 accumulation in mycolactone-exposed cells may be a response to the increase in cytosolic ubiquitinated proteins. The enhanced levels of SQSTM1 puncta in the presence of bafilomycin A₁ suggest that this is dependent on an increased autophagic flux, but protein aggregation and changes in SQSTM1 expression may also contribute to these changes.

Mycolactone-induced autophagy is dependent on SEC61 inhibition

To confirm that the increase in autophagy seen after mycolactone exposure was a consequence of SEC61 inhibition, we used stably transfected HeLa cells harboring a missense mutant of SEC61A1 (SEC61A1^D60G). This is one of a series of mutations in SEC61A1 found to provide resistance to mycolactone in reverse genetic screening due to reduced mycolactone binding [Citation16,Citation35,Citation36]. The expression of the Flag-tagged mutant protein was confirmed by immunoblotting and immunofluorescence (Fig. S3A and S3B) and the HeLa cells were resistant to mycolactone-induced cytotoxicity (Fig. S3C). Mycolactone induced formation of both WIPI2 and LC3B puncta in HeLa cells as before; however, this did not occur in cells overexpressing the mutant SEC61A1 (Figure 3A and 3). Conversely, both wild-type and mutant cells exhibited increased WIPI2 and LC3B puncta formation in response to PP242. For SQSTM1, we saw a more muted induction in wild type HeLa than in primary cells, as before (Figure 3C). Here, co-incubation with mycolactone and bafilomycin A₁ seemed to result in very strong staining for puncta that could potentially be aggregates due to the dual pressure of the different inhibitors. The potential that this may have resulted in undercounting might explain why the absolute number of puncta is lower versus bafilomycin A₁ alone. However, cells expressing SEC61A1^D60G no longer showed an increase in SQSTM1 puncta after exposure to mycolactone (Figure 3C).

Figure 3. Mycolactone-induced changes in autophagy are dependent on SEC61 inhibition.

Parental (WT) and HeLa cells stably overexpressing mycolactone-resistant SEC61A1^D60G were exposed to 0.05% DMSO, 31.25 ng/ml mycolactone (Myco) and/or 1 μM PP242 for 1 h or 100 nM BAF for 4 h. All immunofluorescence images were obtained using a Nikon A1 confocal laser scanning microscope (scale bar: 10 μm). For all panels, unless otherwise indicated, statistical comparisons are to the DMSO-treated cells of that particular cell type; ns = not significant; * = p< 0.05; ** = p < 0.01; *** = p < 0.001; **** = p < 0.0001. (A) Cells were exposed to mycolactone for 8 h or 1 μm PP242 for 1 h and fixed and stained for WIPI2. Values represent the number of puncta detected per cell for at least 140 cells. Data are representative of 3 independent experiments. (B) Cells were exposed to mycolactone for 16 h or PP242 then fixed with methanol for LC3B staining. Values represent the number of puncta detected per cell for at least 128 cells. Data is representative of 2 independent experiments. (C) Cells were exposed to mycolactone for 16 h with or without BAF, then fixed with PFA for SQSTM1 staining with a mouse anti-SQSTM1 antibody. Values represent the puncta detected per cell for 30–100 cells per treatment. Data are representative of 3 independent experiments.

Display full size

Of the ATG proteins, only one has transmembrane domains, ATG9A, making it a potential target of SEC61 inhibition. However, ATG9A is a multipass membrane protein and these are generally insensitive to the inhibitory effects of mycolactone [Citation17]. Instead, mycolactone primarily targets type I and II single-pass membrane and secreted proteins [Citation9,Citation10]. Hence, these effects of mycolactone cannot be explained by changes in ATG9A levels as mycolactone appears to have little effect on ATG9A expression in HFFF cells (Fig. S3D). Furthermore, mycolactone induces a dispersal of ATG9A from its perinuclear location, in a manner similar to that reported for mammalian cells subjected to starvation or treated with rapamycin [Citation37] (Fig. S3E).

Taken together the data suggest the responses we see are driven by blocking of translocation of newly synthesized proteins into the ER which leads to a subsequent buildup of mislocalized, ubiquitinated proteins in the cytoplasm that in turn associate with SQSTM1 and LC3B. The secondary nature of the response to mycolactone may explain the slower appearance of early autophagy markers such as WIPI2 when compared to direct inhibitors of autophagy regulating enzymes such as PP242, or acute nutrient starvation.

Expression of SQSTM1 protects against mycolactone cytotoxicity but is not required for increased autophagy initiation

Next, we sought to characterize the role played by SQSTM1 in mycolactone-induced cytotoxicity by comparing the response between Sqstm1 WT and sqstm1^−/- MEFs [Citation34,Citation38]. Immunoblotting confirmed SQSTM1 deletion in knockout cells, while mycolactone induced a shift of SQSTM1 into detergent-resistant fractions in matched WT cells (Figure 4A). Expression of SQSTM1 was not required for LC3B processing because similar levels of LC3B-II could be seen in both Sqstm1 WT and sqstm1^−/- cells after mycolactone exposure (Figure 4A). Similarly, both Sqstm1 WT and sqstm1^−/- MEFs exhibited an increase in WIPI2 puncta, showing that SQSTM1 itself is not required for initiation of autophagy triggered by SEC61 inhibition or MTOR inactivation by PP242 (Figure 4B). This is unlikely to be due to other ubiquitin-binding receptor proteins taking the place of SQSTM1 in the knockout MEF cells since CALCOCO2, NBR1, TAX1BP1 and OPTN all showed a limited response to mycolactone compared to SQSTM1 in both Sqstm1 WT and sqstm1^−/- cells (Fig. S4A).

Figure 4. Increased expression of SQSTM1 enhances cell survival in the presence of mycolactone but is not required for initiation of autophagy. Cells were exposed to 31.25 ng/ml mycolactone (Myco), alongside a solvent control (0.05% DMSO, equivalent dilution to the mycolactone) for the longest timepoint tested. All immunofluorescence images were obtained using a Nikon A1 confocal laser scanning microscope (scale bar: 10 µm). For all immunoblots, the migration relative to known molecular mass markers is shown. For all panels, unless otherwise indicated, statistical comparisons are to the DMSO-treated cells of that particular cell type, ns = not significant; * = p< 0.05; ** = p< 0.01; *** = p < 0.001; **** = p < 0.0001.(A-C) Matched Sqstm1 WT and sqstm1^−/- MEFs were exposed to mycolactone for the indicated times. As controls, either 100 nM bafilomycin A₁ (BAF) for 12 h or 1 µM PP242 for 1 h were used, as indicated. (A) Cells were harvested in RIPA lysis buffer, then soluble and insoluble fractions were prepared as described in Methods prior to immunoblotting. Migration relative to known molecular mass markers is shown. Twelve-hour mycolactone exposure. Data are representative of 3 independent experiments. (B) Cells were fixed and stained for WIPI2 after 8 h mycolactone exposure. Values represent the means of 3 independent experiments ± SEM. (C) Cell viability in Sqstm1 WT and sqstm1^−/- MEF as measured by CellEvent assay after 24 h mycolactone exposure. Values represent the mean of 3 independent experiments ± SEM. (D) WT HeLa cells exposed to mycolactone for 16 h before fixing with PFA for immunofluorescent staining with mouse anti-SQSTM1, anti-TAX1BP1, anti-CALCOCO2, anti-NBR1 or anti-OPTN antibodies. Data are representative of duplicate experiments. (E) Cell viability for parental HeLa (WT), PentaKO cells (5KO) and PentaKO cells with reconstituted SQSTM1 expression (5KO/SQSTM1) as measured by CellEvent assay after 48 h mycolactone exposure. Values represent the mean of 3 independent experiments ± SEM. (F) Skin sections from samples from healthy donor (left hand panel) and Buruli ulcer patient lesions (right hand panels), stained with anti-SQSTM1 antibody or isotype-matched control. Boxes show locations of magnified SQSTM1-stained sections. Comparable results were found for all 8 patient biopsies analyzed regardless of whether the lesion was an ulcer or a plaque.

Display full size

Both pro-survival and pro-apoptotic roles have been demonstrated for SQSTM1; however, given its involvement in eliminating toxic protein aggregates and restoring cellular homeostasis, we hypothesized that elevated SQSTM1 could be an adaptive response to the increased cytosolic (polyubiquitinated) protein burden, rather than a contributing factor to mycolactone’s cytotoxicity. To address this question, we assessed the viability of Sqstm1 WT and SQSTM1-deficient MEFs after mycolactone exposure. Here, sqstm1^−/- MEFs showed a significantly accelerated loss of viability compared to their Sqstm1 WT counterparts (Figure 4C and S4B). From these data, we conclude that SQSTM1 is protective against, rather than a driver of, mycolactone’s cytotoxicity.

To investigate the relative role of SQSTM1 in the cellular response to mycolactone in comparison with other mammalian autophagy receptors, we used HeLa cells in which expression of the five human receptors (SQSTM1, TAX1BP1, NBR1, CALCOCO2 and OPTN) had been ablated by gene editing [Citation39], known as PentaKO (5KO) cells. In WT HeLa cells, all receptors except OPTN showed some increase in puncta in the presence of mycolactone (Figure 4D). As previously reported [Citation16], WT HeLa cells exposed to mycolactone exhibit a slower rate of death than MEFs. However, in keeping with the viability assays on sqstm1^−/- MEFs, the death rate induced by mycolactone was accelerated in PentaKO cells (Fig. S4C). Immunofluorescence analysis revealed a general increase in cytosolic ubiquitin in PentaKO cells, but in contrast to WT HeLa cells, this was diffuse and did not localize to discrete puncta (Fig. S4D), confirming that receptor proteins are needed for this step. Interestingly, reconstitution of SQSTM1 expression (Fig. S4E) failed to restore viability (Figure 4E). Thus, although SQSTM1 plays an important role in cell survival, baseline levels of expression are insufficient to protect the cells from the buildup of mislocalized proteins and upregulation of expression may also be a requirement.

Since mycolactone is the pathogenic mediator of BU, we used immunohistochemistry to determine whether SQSTM1 expression changes in BU lesions (Table S1). In normal human skin, very little SQSTM1 staining was detected, except for a relatively weak signal seen close to sebaceous glands (Figure 4F, left-hand panel). In biopsies taken from the infected skin of BU patients, their lesions contained necrotic regions (N), which display features of coagulative necrosis and non-necrotic regions which include infiltrated regions containing inflammatory cells. Focusing on the non-necrotic regions, we observed SQSTM1 staining in all eight lesions examined, particularly in areas displaying heavy infiltration of inflammatory cells (Figure 4F). The levels varied between region and cell type; in some areas, SQSTM1 expression was quite widespread while, in others, strong staining was observed in individual cells (Figure 4F, middle panel). In two patients, SQSTM1 was detected close to blood vessels in sub-endothelial regions, associated with long, spindle-shaped cells, which might be smooth muscle cells (Figure 4F right-hand panel). This suggests that the upregulation in SQSTM1 protein expression in response to mycolactone also occurs in vivo.

Mycolactone induction of autophagy requires RB1CC1 but is ULK-independent

Key proteins in the formation of the phagophore are the members of the ULK-complex; RB1CC1 and ULK1/2. In order to dissect the mechanism of autophagy induced by mycolactone, we examined the role of these proteins using knockout MEFs. As observed in HDMEC and HFFF, exposure of Rbcc1 WT MEFs to mycolactone for 8 h led to an increase in RB1CC1-positive puncta (Figure 5A and S5A). In cells lacking RB1CC1, not only were there no RB1CC1 puncta detected (Fig. S5A), but WIPI2 puncta were also drastically reduced in both mycolactone-exposed and PP242-exposed cells; this was also true for LC3B puncta (Figure 5B and 5E). Although RB1CC1 deficiency caused a reduction in SQSTM1-labeled bodies, these were not significantly reduced in mycolactone exposed cells. The area of the SQSTM1 puncta was, however, generally larger than that seen in control cells (Figure 5B and S5B). Hence, the early stages of autophagosome formation following mycolactone exposure are dependent on RB1CC1, while condensation of SQSTM1 aggregates is not. In line with this, survival of RB1CC1-deficient cells was significantly reduced compared to the parental wild-type MEF cells (Fig. S5C) confirming an important protective role for autophagy in cells exposed to mycolactone.

Figure 5. Mycolactone stimulates a non-canonical pathway of autophagosome biogenesis.

(A-E) Matched parental Rb1cc1 WT and rb1cc1^−/- MEFs, or matched parental Ulk1-Ulk2 WT and ulk1^−/-ulk2^−/- MEFs (as indicated) were exposed to 31.25 ng/ml mycolactone (Myco) for 8 h alongside a solvent control (0.05% DMSO, equivalent dilution to the mycolactone) for the longest timepoint tested. When needed 1 µM PP242 for 1 h was used as a control. All immunofluorescence images were obtained using a Nikon A1 confocal laser scanning microscope (scale bar: 10 µm). For all panels, unless otherwise indicated, statistical comparisons are to the DMSO-treated cells of that particular cell type, ns = not significant; * = p< 0.05; ** = p < 0.01; *** = p < 0.001; **** = p < 0.0001. Results are representative of 3 independent experiments (A-D) or the mean ± SEM of 3 independent experiments (E).(A) Cells were fixed and stained for RB1CC1. (B and D) Cells were fixed with PFA for WIPI2 and SQSTM1 staining or fixed with methanol for LC3B staining. (C) Cells were lysed directly in gel sample buffer prior to immunoblotting. Migration relative to known molecular mass markers is shown. (E) Quantification of WIPI2, SQSTM1 and LC3B puncta for cells shown in B and D. Values represent the means of 3 independent experiments ± SEM. (F and G) HFFF cells were double transfected with non-targeting (ctrl), RB1CC1 or ULK1 SMARTpool siRNA oligonucleotides for a total of 60 h. (F) Cells were lysed directly in gel sample buffer prior to immunoblotting. Migration relative to known molecular mass markers is shown. (G) Cells were exposed to mycolactone for 6.5 h then fixed for WIPI2 staining. Values represent the mean puncta per cell for 13 fields containing at least 60 cells.

Display full size

Concerning the role of ULK1 and ULK2, we first examined the phosphorylation of ULK1 in Ulk1-Ulk2 WT MEF. As expected, the MTOR inhibitor PP242 caused rapid dephosphorylation of ULK1 at Ser727 within 1 h. In contrast, an 8 h exposure to mycolactone reduced ULK1 phosphorylation to a much smaller extent, which did not reach statistical significance (Figure 5C and S5D). As expected [Citation40], no ULK1 protein was detected in the double knockout (Figure 5C). However, despite the slightly reduced levels of ULK1 phosphorylation in mycolactone-exposed cells, the absence of the ULK kinases had no impact on the increase in autophagy initiation by mycolactone, as assessed by the appearance of WIPI2, SQSTM1 and LC3B puncta (Figure 5D and 5E). By contrast, the response to PP242 was reduced by approximately 50% for each of these proteins. This suggests that ULK1 and/or ULK2 are not involved in the early stages of the mycolactone-dependent increase in autophagy initiation; accordingly, cells lacking both ULK kinases died at the same rate as their parental wild-type cells (Fig. S5E).

These findings were replicated in primary HFFF cells using an siRNA knockdown approach. Knockdown of RB1CC1 (Figure 5F) caused a significant reduction in WIPI2 puncta in mycolactone-exposed cells (Figure 5G). By contrast, loss of ULK1 had no effect (Figure 5F and 5G). This supports the conclusion that the response to mycolactone is dependent on RB1CC1 but not ULK kinases. This surprising finding led us to explore potential mechanisms that might regulate this unusual autophagic response.

SQSTM1-dependent selective autophagy induced by mycolactone is controlled at the level of translation

Since baseline SQSTM1 expression was not sufficient to protect PentaKO cells from mycolactone exposure (Figure 4E), we considered whether regulation of its expression was important. In our previously published data set for translational profiling of mycolactone-treated cells, the Sqstm1 mRNA showed a 1.8-fold increase in translational efficiency after mycolactone exposure compared to control cells (p = 0.004) [Citation16]. Hence, it seemed possible that the mycolactone-induced ISR was involved in the regulation of SQSTM1 expression. In MEFs lacking the EIF2S1 kinase EIF2AK3 [Citation16], resting levels of SQSTM1 were equivalent to WT cells but levels of SQSTM1 protein were no longer increased after exposure to mycolactone (Figure 6A). However, the lack of EIF2AK3 had less impact on mycolactone-dependent LC3B processing and the accumulation of SQSTM1 in response to chloroquine was not affected. Similar results were obtained when using bafilomycin A₁ as a control (Fig. S6A). A small increase in the steady state level of Sqstm1 mRNA in mycolactone exposed cells was not affected by the absence of EIF2AK3 (Figure 6B), hence the increased expression of SQSTM1 appears to be regulated post-transcriptionally. To rule out transcriptional regulation of SQSTM1 by NFE2L2/Nrf2 (nuclear factor, erythroid 2 like 2) following mycolactone exposure, we quantitated expression of the NFE2L2 target Osgin1 (oxidative stress induced growth inhibitor 1) [Citation41,Citation42]. While the NFE2L2 activator monomethyl fumarate efficiently induced an increase in steady state Osgin1 expression in both Eif2ak3 WT and eif2ak3^−/- MEFs, mycolactone had no effect at either 6 or 24 h (Fig. S6B).

Figure 6. Mycolactone-induced autophagy markers are dependent on the integrated stress response.

Cells were exposed to 31.25 ng/ml mycolactone (Myco), alongside a solvent control (0.05% DMSO, equivalent dilution to the mycolactone) for the longest timepoint tested. For all immunoblots, the migration relative to known molecular mass markers is shown. All immunofluorescence images were obtained using a Nikon A1 confocal laser scanning microscope (scale bar: 10 µm). For all panels, unless otherwise indicated, statistical comparisons are to the DMSO-treated cells of that particular cell type, ns = not significant; * = p < 0.05; ** = p < 0.01; *** = p < 0.001; **** = p < 0.0001. (A-C) Matched Eif2ak3 WT or eif2ak3^−/- MEFs were left exposed to mycolactone for 24 h (or the times indicated). Either 50 µM chloroquine (CQ) 12 h or 10 μM MG132 for 16 h were used as controls. Where co-incubated, CQ was included for the final 12 h. Values represent the mean of 3 independent experiments ± SEM. (A) Cells were lysed directly in gel sample buffer prior to immunoblotting. Quantification of SQSTM1 abundance represents pixel intensity relative to GAPDH controls. (B) Total RNA was extracted from cells and subjected to qRT-PCR; data were normalized to Gapdh mRNA. (C) Cells were exposed to mycolactone for the times indicated, with or without 100 nM BAF for 4 h then fixed with PFA for SQSTM1 staining. Values represent the number of puncta in each cell for 60–100 cells ± SD. Data are representative of 2 independent experiments. (D) HeLa cells were exposed mycolactone for 24 h, or 5 µg/ml tunicamycin (Tuni) for 6 h, in the presence or absence of 100 nM ISRIB then lysed directly in gel sample buffer. Data are representative of 3 independent experiments. (E and F) WT MEF were exposed to mycolactone for 8 h or 1 µM PP242 for 1 h in the presence or absence of 200 nM ISRIB. Cells were fixed with PFA for WIPI2 and RB1CC1 staining, or with methanol for LC3B staining. (F) Quantification of puncta/cell for WIPI2, RB1CC1 and LC3B. Values represent the mean of triplicate experiments ± SEM.

Display full size

Immunofluorescence staining for SQSTM1 in WT MEFs showed a time-dependent increase of SQSTM1 puncta in response to mycolactone similar to that seen in HeLa cells (Figure 6C); however, these changes were markedly impaired in EIF2AK3-deficient cells. Moreover, there was no difference between these cell lines in number of SQSTM1 puncta observed in response to bafilomycin A₁ (Figure 6C).

In the absence of EIF2AK3, other EIF2S1 kinases are still able to contribute to the mycolactone-induced ISR [Citation16]. Hence, we also asked if the small molecule ISR inhibitor (ISRIB) could prevent the upregulation of SQSTM1. ISRIB reverses the effects of EIF2S1 phosphorylation by enhancing the guanine nucleotide exchange factor activity of the EIF2B complex [Citation43], enabling the rapid formation of the 43S ternary complex required for efficient translation initiation even in the presence of p-EIF2S1. In HeLa cells, both mycolactone and the ISR inducer, tunicamycin, efficiently promoted ATF4 translation as previously reported [Citation16,Citation44], and this was completely reversed by ISRIB (Figure 6D). ISRIB also completely reversed the upregulation of SQSTM1 elicited by mycolactone, providing additional evidence that EIF2S1 phosphorylation is required for the translational control of SQSTM1.

To investigate the role of the ISR in the development of the early autophagic response, WT MEF were incubated with mycolactone in the presence or absence of ISRIB at a concentration that blocked ATF4 induction in these cells (Fig. S6C). While the effect was not as marked as that seen in RB1CC1-deficient cells, numbers of RB1CC1, WIPI2 and LC3B puncta were all reduced in the presence of ISRIB (Figure 6E and 6F). Interestingly, the response to PP242 was also reduced, particularly with respect to the increase in LC3B-positive puncta. This last effect was reproduced in MEFs expressing a non-phosphorylatable mutant of EIF2S1, EIF2S1^S51A (Fig. S6D and S6E). These cells were also highly sensitive to mycolactone, and approximately 70% were dead after 24 h exposure (Fig. S6F). Taken together, the results show that the ISR is critical to the RB1CC1-dependent, ULK-independent pathway of protective autophagy that is by induced by mycolactone.

Reagents and cell culture

For all experiments, we used synthetic mycolactone A/B [Citation76], which was generously donated by Prof. Yoshito Kishi (Harvard University). HeLa cells were obtained from ATCC (CCL-2). PentaKO cells were a gift from Prof. Michael Lazarou (Monash University, Australia) [Citation39]. Human Caucasian fetal foreskin fibroblast cells (HFFF2) were purchased from Merck (86031405). HDMEC were from Promocell (C-12,210). eif2ak3^−/-, Eif2s1^S51A, and sqstm1^−/- and respective WT MEFs have been described previously [Citation34,Citation38,Citation77,Citation78] and were kindly donated by Prof. Ronald Wek (Indiana University), Dr Mark Caldwell (University of Southampton) and Prof. Terje Johansen (Arctic University of Norway), respectively. Rb1cc1 WT and rb1cc1^−/- MEFs [Citation79] were a gift from Prof Jun Lin Guan (University of Cincinnati) while Ulk1-Ulk2 WT and ulk1^−/-ulk2^−/- double-knockout MEFs [Citation40] were from Sharon Tooze (The Francis Crick Institute, London). All cells except HDMEC were grown at 37°C, 5% CO₂ in high-glucose DMEM (Merck, D6429) supplemented with 10% FBS (Gibco, 10500064). In the case of WT and mutant MEF cell lines and PentaKO HeLa cells, cells were additionally supplemented with 1 mM non-essential amino acids (Gibco, 11140035) 50 µM β-mercaptoethanol (Gibco, 31350010) and 100 µg/ml penicillin/streptomycin (Gibco, 5140122). HDMEC were cultured at 37°C, 5% CO₂ in Endothelial Cell Growth Medium MV (Promocell, C22120).

Mycolactone was diluted from a 500 µg/ml stock in DMSO (Merck, D2560) and was used at the minimal inhibitory concentration in all experiments. This was usually 31.3 ng/ml (~42 nM) in all cells except for HDMEC, which are more susceptible to mycolactone and therefore require only 10 ng/ml (~13 nM) mycolactone for maximal effect [Citation6]. Chloroquine (Merck, C6628) was diluted to 50 µM from a stock in sterile ultrapure water and bafilomycin A₁ (Invivogen, tlrl-baf1) was used at 100–200 nM, diluted from a DMSO stock. PP242 (Bio-Techne Ltd, 4257) was used at 1 µM. MG132 (Merck, M8699) was used at a final concentration of 10 µM and monomethyl fumarate (Merck, 651419) at 100 µM. ISRIB, diluted from a DMSO stock (Merck, SML0843), was used at 100 nM for HeLa cells and HDMEC and 200 nM for MEF and HFFF. Tunicamycin (Merck, T7765) was used at 5 µg/ml.

Transfection

A synthetic mutant SEC61A1^D60G gene (Thermo Fisher Scientific) was generated and cloned into the pNLF1-C vector (Promega, N1361), replacing the nanoluciferase insert. The plasmid was digested with MluI (New England Biolabs, R0198S) and 2 µg transfected into HeLa cells with Fugene 6 (Promega, E2691) according to the manufacturer’s instructions. Cells were selected for one week in the presence of 10 ng/ml mycolactone epimers [Citation80,Citation81] (kind gift from Dr Nicolas Blanchard, CRNS France) and then for 1 week with 600 µg/ml hygromycin (Gibco, 10687010). The tandem tag vector pDEST-mCherry-eGFP-LC3B and control pDEST-mCherry-eGFP-C1 were a gift from Terje Johansen [Citation31]. The plasmid named “HA-p62”, encoding hemagglutinin-tagged SQSTM1 was from Addgene (280277), deposited by Qing Zhong [Citation82]. Plasmids (2 µg) were transfected into WT HeLa or PentaKO cells with Fugene 6. For tandem tag assays, WT cells were incubated at 37°C and 5% CO₂ for 24 h, then trypsinized and plated onto sterile coverslips and incubated for a further 24 h. After treatment, cells were fixed with 4% paraformaldehyde (PFA; Merck, P6148) and mounted on Aqua-Polymount (Tebu-Bio Ltd, 18606–20) and imaged on a Nikon A1 confocal laser scanning microscope. Images were analyzed using ImageJ software (NIH). For the selection of stably transfected PentaKO cells expressing SQSTM1, resistant colonies that emerged after 2 weeks treatment with 400 µg/ml zeocin (Gibco, R25005) were picked, expanded and verified by western blotting.

Small interfering RNA transfection

HFFF cells plated into 6-well plates were transfected with Dharmacon OnTarget Plus SMARTpool siRNAs (L-021117-00-0005 for RB1CC1 and L-005049-00-0005 for ULK1) or OnTarget Plus Non-Targeting pool (D-001810-10-20) (see Table S2 for details) according to manufacturer’s instructions using 10 µl siRNA and 4.5 µl DharmaFECT 1 Transfection Reagent (Horizon Discovery, T-2001) per well. Cells were incubated overnight then the medium was replaced. After a further 24 h the transfection was repeated. After overnight incubation, cells were trypsinized, plated onto coverslips and incubated overnight again before assaying.

Immunofluorescence microscopy

Cells were grown on pre-sterilized glass coverslips. For LC3B visualization, cells were fixed and permeabilized in methanol (Merck, 34885-M) and blocked in 2% FBS, 1% goat serum (Merck, NS02L) in phosphate-buffered saline (PBS; Merck, 524650) for at least an hour. For all other markers, cells were fixed with 3.75% PFA in 200 mM HEPES (Merck, H0887), pH 7.5, permeabilized with 0.25% Nonidet P-40 Alternative in Netgel (150 mM NaCl [Thermo Fisher Scientific, S/3161/53], 5 mM ethylenediaminetetraacetic acid [Merck, E5134], 50 mM Tris-Cl [Thermo Fisher Scientific, BP152-1], pH 7.4, 0.05% Nonidet P-40 Alternative [Merck, 492016], 0.25% gelatin [Merck, G7401] and 0.02% sodium azide [Merck, S2002]). Antibodies used in this study are as follows: mouse anti-SQSTM1 (Abcam, 56416), rabbit anti-SQSTM1 (Enzo Biosciences, BML-PW9860), anti-LC3B (MBL International, PM036), anti-WIPI2 (Bio-Rad, MCA5780GA), anti-RB1CC1 (Bio-techne, NB100-77279), anti-ATG9A (Merck, ZRB1648), anti-ATG16L1 (Cell Signaling Technology, 8089), anti-NBR1 (Cell Signaling Technology, 9891), anti-CALCOCO2 (GeneTex, GTX115378), anti-TAX1BP1 (Proteintech 14,424-1-AP), anti-OPTN (Cayman Chemical, 100000), anti-rabbit Alexa Fluor 488 (Invitrogen, A11034; 1:500), anti-rabbit Alexa Fluor 594 (Invitrogen, R37117; 1:500), anti-mouse Alexa Fluor 488 (Invitrogen, A21202; 1:600). Nuclei were stained using a 0.1 µg/ml DAPI (Invitrogen, 62248) in PBS before mounting onto glass slides with VectaShield (Vector Laboratories, H-1000) or Aqua-Poly/mount. Slides were imaged on a Zeiss Axiovert S100 TV inverted fluorescence microscope or Nikon A1 confocal laser scanning microscope and NIS Elements software. Images were analyzed using ImageJ software. For quantification, puncta in each cell were counted using the ImageJ “Analyze Particles” function. This function measures number, area and intensity of fluorescent signals within a selected space. Where numbers of puncta are presented as averages in bar charts the number of puncta per field was divided by the number of nuclei. Where puncta per cell is represented by individual points, each point represents the number of puncta in an individual cell. For each experimental treatment puncta in each cell were counted for each experimental treatment in at least 3 fields.

Cell lysis and immunoblotting

For analysis of proteins within detergent-soluble and -resistant fractions, cells were first lysed in RIPA buffer (Merck, 20–168) supplemented protease inhibitor cocktail (Merck, P9599), 1 mM sodium pyrophosphate (Merck, 71501), 5 mM sodium fluoride (Merck, S7920) and 1.75 mM β-glycerophosphate (Merck, G5422) for 5 min before centrifugation at 20,000 x g at 4°C. Supernatants (soluble) were removed and protein content was quantified by Pierce BCA assay (Thermo Fisher Scientific, 23227) before normalization to the lowest concentration. Pellets (insoluble) were dissolved in “gel sample buffer” (50 mM Tris, pH 6.8, 10% glycerol [Thermo Fisher Scientific, G/0650/17], 2.5% β-mercaptoethanol, 2% sodium dodecyl sulfate [Thermo Fisher Scientific, 1281–1680], 0.01% bromophenol blue [BDH Chemicals, 44305]) and sonicated for 30 s. Alternatively, whole cell lysates were prepared directly in gel sample buffer and sonicated for 30 s.

Lysates were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis alongside a broad range polypeptide marker (Spectra BR; Thermo Fisher Scientific, 26623) and transferred to PVDF membranes (Thermo Fisher Scientific, 11556345). Antibodies used in this study are as follows: anti-SQSTM1 (Abcam, Ab56416), anti-TUBB/β-tubulin (Bio-Techne Ltd, NB600-936SS), anti LC3B (Cell Signaling Technology, 4108), anti-EIF2AK3 (Cell Signaling Technology, 9606), anti-ubiquitin (FK2, Calbiochem, ST1200), anti-GAPDH (Ambion, AM4300), anti-rabbit-HRP (GE Healthcare, NA934V), anti-mouse-HRP (GE Healthcare, NA931V). Blots were developed using enhanced chemiluminescence with Immobilon™ western chemiluminescence HRP substrate (Thermo Fisher Scientific, 11556345) and imaged on a Fusion FX Imager (Vilber-Lourmat), which provides a warning if the areas of the image are saturated. Quantification was performed via ImageJ analysis of pixel density from non-saturated images and data are presented as normalized values to GAPDH controls.

qRT-PCR

Total RNA was extracted using a QIAGEN RNeasy kit (74004) and real-time quantification of mRNA in triplicate samples was performed by TaqMan PCR on a QuantStudio7Flex real-time PCR system (Applied Biosystems) using the following program: 50°C for 2 min and DNA polymerase activation at 95°C for 10 min, followed by 40 cycles of at 95°C for 15 s and 60°C for 1 min. The following Applied Biosystems TaqMan probes were used in this study: SQSTM1 (Mm_00448091_m1), OSGIN1 (Mm_00660947_m1) and GAPDH (Mm_99999915_g1). Threshold cycle values were normalized to GAPDH internal controls and fold-change calculated from ΔΔCt values.

Cell viability assays

Cell death induced by mycolactone was assessed using CellEvent™ Green Detection Reagent (Thermo Fisher Scientific, C10423) as previously described [Citation16]. Briefly, cells were seeded in 96-well plates and treated as indicated, after which propidium iodide (Thermo Fisher Scientific, 11519206) and CellEvent were added at a final concentration of 0.3 μg/mL and 1% v:v, respectively. After a 30-min incubation at 37°C, 5% CO₂, plates were imaged on a Zeiss Axiovert S100 TV inverted fluorescence microscope. For at least two technical replicates per condition, a total of 3–4 bright-field images were taken, as well as corresponding images under red (PI) and green (CE) fluorescence channels. The proportion of cells in late apoptosis (with PI⁺/CE⁺ nuclei) were expressed as a percentage of total cells enumerated from bright-field images (mean cell survival = $(1-\frac{CE{\hspace{0.17em}}_{\hspace{0.17em}}^{+}PI{\hspace{0.17em}}_{\hspace{0.17em}}^{+}nuclei}{totalcells})\mathrm{X}100$.

For viability assays (which measure both growth and death simultaneously), cells were seeded into 96-well plates at a density of 1000/well and treated with various concentrations of mycolactone for 96 h, then resazurin (Sigma Aldrich, R7017) added to a concentration of 0.01 mg/ml. After 4 h further incubation, emission at 620 nm at excitation 580 nm was measured using a BMG Fluostar Optima Microplate Reader. Data are expressed as a viability index, representing the relative signal compared to the negative control (DMSO-treated cells).

Immunohistochemistry of human skin samples

Fixed and embedded 4 mm punch biopsies from Buruli ulcer patients were described previously [Citation6]. In ulcerated BU lesions (6 patients) punch biopsies were taken 1 cm inside the outer margin of the induration surrounding the ulcer. For BU plaque lesions (2 patients) punch biopsies were collected from the non-ulcerated center of the lesion. Healthy control tissue was obtained at The Whitely Clinic or purchased from AMS Biotechnology (Europe) Ltd (500041028). After removal, all tissues were fixed in neutral formalin, embedded into paraffin (pfm Medical, 9000R2010) and sectioned (5 μm). Sections were deparaffinized, endogenous peroxidase quenched, epitopes unmasked (pH 6 citrate buffer; Sigma, C999) and blocked with horse serum (Vector Laboratories, S-2000-20). The tissue sections were then incubated with either anti-SQSTM1 antibody (Abcam, ab56416) or IgG2a isotype control (Santa Cruz Biotechnology, sc-3878) overnight at 4°C and biotin-conjugated secondary antibody (Vector Laboratories, BA2000). Staining was performed using VECTASTAIN Elite ABC kit (Vector Laboratories, PK-6100) and Vector NovaRED peroxidase substrate (Vector Laboratories, SK-4800). Counterstaining was performed with Shandon Harris Hematoxylin (Thermo Fisher Scientific, 6765001).

Statistical analysis

All analyses were performed in GraphPad Prism v7: experiments assessing one independent variable across more than two groups were analyzed using a one-way analysis of variance (ANOVA), while those concerning two independent variables were analyzed by two-way ANOVA. Tukey’s post-hoc test was used to identify significant differences between groups and correct for multiple comparisons. Unless indicated with a line on the graph, statistical comparisons are to the DMSO-treated cells of that particular cell type. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, ns = not significant.

Ethics statement

Ethical approval for analyzing patient specimens was obtained from the Ethikkommission beider Basel, Basel, Switzerland and the provisional national ethical review board of the Ministry of Health Benin (N° IRB00006860) as well as from the Cameroon National Ethics Committee and the Ethics Committee of the Heidelberg University Hospital, Germany (ISRCTN72102977). Written informed consent from the patients or their guardians was obtained before specimens were collected for reconfirmation of BU as well as for detailed histopathological analysis and all patient data have been anonymized. Favorable ethical opinion for analyzing normal human skin was given by Faculty of Health and Medical Science Ethics Committee of the University of Surrey (1174-FHMS-16). The normal human skin samples were collected at The Whiteley Clinic or purchased from AMS Biotechnology (Europe) Ltd. Written informed consent was obtained from all donors. The research related to human tissues complies with the ethical processes of University of Surrey.