DISCUSSION
Examples of bacterial pathogenicity factors being directly targeted to – and imported into – mitochondria have been described for a number of pathogens (Kozjak-Pavlovic et al. , 2008, Spier et al. , 2019) and numerous studies have explored the effects of mycobacterial infection on the function of mitochondria; however, there have been relatively few studies investigating the direct association of mycobacterial proteins with mitochondrial membranes at an intracellular level. Cpn60.2 is an example of a secreted Mtb protein (Joseph et al. , 2017) which was expressed intracellularly in RAW macrophages and was partially localized to mitochondrial membranes by both immuno-EM and immunofluorescence localization. However, there was no indication of the disruption of mitochondrial membranes that we observed at 48 hrs post-infection of Mtb H37Rv in A549 AECs (Fine-Coulson et al. , 2015) and mitochondrial morphology was not quantitatively assessed. Due to its membrane lytic activity, ESAT-6 was a likely candidate to cause disruption of the mitochondrial membranes directly; however, intracellular expression within A549 AECs in this study indicated a lack of colocalization with the mitochondria. Indeed, comparison of the degree of fragmentation induced by ESAT-6 versus a control GFP protein suggested that either high-level expression of GFP was sufficient to fragment mitochondria, or ESAT-6 induces increased tubulation, thus working in an opposite manner to GFP. These results support our conclusion that other secreted Mtb proteins may be directly responsible for the mitochondrial fragmentation phenotypes observed during Mtb pathogenesis, and not an activity of ESAT-6.
We initially became interested in PE17 following a report that PE17 localized to the mitochondria and decreased host cell secretion upon expression in epithelial cells (Stamm et al. , 2019). We found that PE17 did not completely co-localize with mitochondria when expressed in A549 cells but was found adjacent to fragmented mitochondria. Although PE17 did not induce exactly the same mitochondrial fragmentation pattern as was seen after M. tuberculosis infection of A549 cells, we did observe that PE17 expression resulted in an extensive decrease in mitochondrial mass (Figure 1), which corresponds to our previous finding (Fine-Coulsonet al. , 2015). This result indicates mitochondrial fragmentation can be dissociated from decreases in mitochondrial mass, suggesting that multiple bacterial factors are likely involved in the mitochondrial changes observed during M. tuberculosis infection. In 2019, Aguilar-Lopez et al. investigated two mycobacterial virulence factors which had been previously demonstrated to have N-terminal mitochondrial targeting sequences, LprG and PE_PGRS33 (Aguilar-Lopez et al. , 2019). Recombinant his-tagged proteins were added extracellularly to human monocyte derived macrophages (MDMs) and size, interconnectivity and elongation of mitochondria assessed. Surprisingly, these two proteins had opposite effects; in LprG stimulated cells there was a decrease in size, interconnectivity and elongation suggesting an increase in mitochondrial fission whereas the opposite was seen in PE_PGRS33 stimulated cells indicating increased fusion. Further evidence for a role of other members of the PE/PPE family in mitochondrial dynamics was reported in an elegant study by Cadieuxet al . A stably-transfected cell line expressing an inducible PE_PGRS33 was generated and the mitochondrial localisation confirmed by immunofluorescence microscopy (Cadieux et al. , 2011). They also detected a swelling of mitochondria in these cells compared to controls. Localisation of three additional PGRS-containing proteins, PE_PGRS1, PE_PGRS18 and PE_ PGRS24, was also determined and although none colocalized completely with mitochondria, both PE_PGRS18 and PE_PGRS24 localized in a compact structure close to the nucleus. Intriguingly, PE_PGRS24 cells also appeared to be closely adjacent to, possibly surrounding, spherical compact mitochondrial structures reminiscent of the pattern seen after PE17 expression. Thus, it is possible that PE17 and other mitochondrially targeted Mtb proteins, including other members of the PE/PPE family, can work cooperatively to alter the host mitochondria during M. tuberculosis infection and should be a focus of future studies.
An unexpected finding of this study was the extensive disruption in the mass and morphology of the Golgi stack and TGN (Figure 2) in response to the expression of PE17. Previous work had detected a secretion defect in PE17-expressing HeLa cells (Stamm et al. , 2019), and it is therefore possible that the structural changes identified in the present study contribute to the observed changes in host secretion during PE17 expression. Indeed, chemical disruption of the Golgi stack, commonly achieved through the fungal metabolite brefeldin A, decreases secretion (Misumi et al. , 1986, Oda et al. , 1987). Interestingly, the mass and morphology of the ER was unaffected by PE17 expression.
We also found a significant decrease in the mass of late endosomes and lysosomes – but not early endosomes or recycling endosomes – upon PE17 expression. In macrophages, M. tuberculosis is known to prevent fusion of the bacterium-containing compartment with late endosomes, increasing bacterial viability (Via et al. , 1997, Westman & Grinstein, 2020, Zulauf et al. , 2018). In A549 cells, it has been shown that the M. tuberculosis bacterium-containing compartment does acquire late endosomal markers, but avoids lysosomal fusion through some involvement of the autophagy pathway (Fine et al. , 2012). It is possible that the PE17-mediated decrease in the mass of the late endosomes and lysosomes is another mechanism by which M. tuberculosis avoids lysosomal degradation. Alternatively, as PE17 causes massive disruption of the TGN, which is a major source of input into late endosomes and lysosomes, a reduction of flux through this post-Golgi trafficking pathway could result in the observed loss in mass of endo-lysosomal organelles.
One of our most striking results is the association of PE17 with the surface of lipid droplets. Expression of PE17 in M. smegmatisincreased the bacterial viability during macrophage infection while increasing macrophage necrosis (Li et al. , 2019). WhileMycobacterium smegmatis is known to utilize lipids and fatty acids as a carbon source for growth, this bacterium lacks the molecular machinery to persist intracellularly within macrophages. It is possible that by providing PE17 to M. smegmatis , it allowed this bacterium to acquire enough host cell lipids to allow for some intracellular persistence in this study, unlike the wild-type parent. Therefore, it is tempting to speculate that PE17 is directly involved in either inducing the synthesis of, or ‘packaging’ of, host organellar membranes into lipid droplets to serve as a source of nutrients for the intracellularMycobacterium . Alternatively, perhaps PE17 coats host lipid droplets to prevent the host cell from consuming its own lipid droplets, thereby ensuring a pool of lipids to be acquired byMycobacterium .
Several infectious diseases caused by intracellular pathogens have been associated with effects on LD-organelle interfaces (Kory et al. , 2016), including the association of Mtb-containing phagosomes with host LDs within foamy macrophages (Peyron et al. , 2008). This process is mediated by Rab7 (Roque et al. , 2020) and is also characterized by LD redistribution and/or clustering upon infection. In this study, PE17 positive LDs appear to be largely localized perinuclearly and in close proximity to other LDs suggesting the potential of PE17 to cluster lipid droplets in vivo. Increasing lipid flux towards lipid droplet production without concomitant lipid replacement of affected membrane compartments could lead to the drastic PE17-dependent organellar phenotypes we have observed in this study. Alternatively, or in addition, the widespread effects seen during PE17 expression may result from direct interaction with a host protein that has a regulatory role in organelle homeostasis.
PE17 was detected in large spherical structures apparent in transmitted light images as well as surrounding BODIPY-positive structures in fluorescence images, a pattern reminiscent of perilipin A labelled LDs (Garcia et al. , 2003). Although there is no consensus regarding a specific protein targeting signal associated with proteins that bind lipid droplets from the cytosol (Class II proteins) (Ingelmo-Torreset al. , 2009, Nakamura & Fujimoto, 2003, Olzmann & Carvalho, 2019), truncation experiments indicate the central domain of perilipin, containing a combination of hydrophobic domains in conjunction with an acidic domain, is required for targeting and anchoring to LDs (Garciaet al. , 2003). In this report we show the unique C-terminal domain of PE17 is required for association with LDs in A549 cells. This domain does not contain any previously identified targeting motifs, nor is it predicted to be helical in nature, suggesting that PE17 may be binding to a LD associated protein rather than directly interacting with the LD membrane. A recent study used a proteomic approach to identify proteins associated with LDs after infection with live Mtb (Menonet al. , 2019). Unexpectedly, as vesicular transport from LDs does not occur, proteins associated with vesicular transport and lysosomal biogenesis were found to accumulate on LDs during Mtb infection. As we have demonstrated an effect of PE17 expression on multiple organelles, it is tempting to speculate that PE17 may interact with a protein or proteins associated not only with LDs, but with mitochondria, Golgi, and lysosomes as well. One candidate identified in the proteomic screen (Menon et al. , 2019) was Rab7, which increased its association with lipid droplets after Mtb infection and has been implicated in regulating LD-phagosome interactions (Roque et al. , 2020) as well as promoting lysosome-mitochondria contacts (Wong et al. , 2018) in addition to its more traditional role in Golgi-endosome trafficking (Guerra & Bucci, 2016). Another potential candidate would be Arf1. Arf1 is the substrate for the guanine exchange factor GBF1, both of which are perhaps best known for their role in recruiting COPI to coat vesicle for retrograde traffic from the Golgi stack to the ER (Kaczmarek et al. , 2017). At the LD surface, Arf1/COPI facilitates the transfer of proteins from the ER to LD, and regulates LD surface tension (Thiamet al. , 2013, Wilfling et al. , 2014). Arf1 is also associated with mitochondria as well as Golgi stack so offers a potential link between the organelles most affected by PE17 expression. PE17 may interact with and/or sequester Arf1/GBF1 at the LD surface, inhibiting its many cellular roles.
Although the accumulation and redistribution of LDs within the host cell is a hallmark of Mtb infection, the mechanisms involved in this process are still poorly understood and require further investigation. Our current findings suggest a potential role for PE17 in manipulation of host LDs, but further research is needed to determine how PE17 expression leads to changes in organelle mass and if these changes are mediated through either PE17:host protein or direct PE17:lipid interactions.