Hansenula polymorpha cells lacking the ER-localized peroxins Pex23 or Pex29 show defects in mitochondrial function and morphology

Peroxisomes are organelles that contain a proteinaceous matrix and are enclosed by a single membrane. They perform several metabolic functions, depending on the organism, tissue and developmental stage. Common functions include fatty acid β-oxidation and H₂O₂-based respiration (Smith and Aitchison, 2013). Proteins involved in peroxisome biogenesis are called peroxins and are encoded by PEX genes. So far, 37 PEX genes have been identified (Jansen et al., 2021).

Most peroxins localize to peroxisomes. However, members of the Pex23 protein family localize to the endoplasmic reticulum (ER). Proteins of this family are characterized by a transmembrane domain and a C-terminal Dysferlin (DysF) domain (Jansen et al., 2021). The function of the DysF domain is still unknown (Bulankina and Thoms, 2020).

Pex23 family members only occur in yeast and filamentous fungi. The number of Pex23 family members varies. Saccharomyces cerevisiae has five (called Pex28, Pex29, Pex30, Pex31 and Pex32), while Hansenula polymorpha contains four Pex23 family members (Pex23, Pex24, Pex29 and Pex32) (Jansen et al., 2021).

S. cerevisiae Pex23 family proteins have been extensively studied. These proteins play a role in de novo peroxisome biogenesis and in the formation of peroxisome-ER membrane contact sites (David et al., 2013; Mast et al., 2016). Proteomics studies showed that S. cerevisiae Pex29, Pex30 and Pex31 are components of a larger complex together with the ER reticulon-like proteins Rtn1, Rtn2 and Yop1 at ER-peroxisome contact sites. This protein complex defines a specialized domain of the ER where pre-peroxisomal vesicles (PPVs) bud off during de novo peroxisome formation (David et al., 2013; Farré et al., 2019). Deletion of S. cerevisiae PEX30 or PEX31 changes the kinetics of PPVs formation, indicating that these proteins regulate PPV formation (David et al., 2013; Joshi et al., 2016; Mast et al., 2016). ScPex30 also colocalizes with certain lipid droplet (LD) biogenesis factors. Hence, the same specialized ER domain plays a role in the biogenesis of PPVs and LDs (Choudhary et al., 2020; Joshi et al., 2018; Wang et al., 2018).

ScPex30 also functions at nuclear–vacuolar junctions (NVJs). To fulfil its different functions ScPex30 forms complexes with other members of the ScPex23 protein family. For the formation of peroxisome-ER contact sites, ScPex30 associates with ScPex28 and ScPex32, while it associates with ScPex29 in NVJs (Ferreira and Carvalho, 2021).

We previously studied the four members of the Pex23 family of the yeast H. polymorpha. We showed that HpPex24 and HpPex32 play key roles in peroxisome biogenesis and are required for the formation of ER-peroxisome contacts (Wu et al., 2020). The latter is underscored by the observation that the peroxisomal defects of pex24 and pex32 mutants could be suppressed by the introduction of an artificial ER-peroxisome tethering protein. Interestingly, cells lacking HpPex29 showed no peroxisome phenotype, whereas the absence of HpPex23 has only a minor effect on peroxisomes (Wu et al., 2020).

While HpPex24 and HpPex32 accumulate at peroxisome-ER contact sites, HpPex23 and HpPex29 localize to multiple regions of the ER. Moreover, like ScPex30, HpPex23 also accumulates at NVJs (Wu et al., 2020).

Our observation that cells lacking HpPex29 do not show any peroxisomal phenotype suggests that Pex23 family proteins may have additional functions. In the current study we found that the absence of HpPex23 or HpPex29, but not of HpPex23 or HpPex32, results in reduced numbers of LDs and mitochondrial defects. pex23 and pex29 cells show retarded growth on media containing glucose as sole carbon source and reduced mitochondrial activities. Microscopy analyses revealed that in pex23 and pex29 cells mitochondria are fragmented and clustered. In these cells the levels of Fzo1 were decreased, suggesting that the phenotype is related to reduced mitochondrial fusion. This explanation is underscored by the partial repression of the morphological and functional phenotypes by deletion of DNM1. Introduction of an artificial ER-mitochondrion tether protein partially suppressed the growth phenotype, suggesting a role of HpPex23 and HpPex29 in ER-mitochondrion membrane contacts.

To study whether the absence of H. polymorpha Pex23 family proteins affects other organelles in addition to peroxisomes, we analysed the morphology of vacuoles, LDs and mitochondria in glucose-grown pex23, pex24, pex29 and pex32 cells. Fluorescence microscopy (FM) revealed that in cells of all deletion strains vacuole morphology (marked with FM4-64) was similar as in WT control cells (Fig. 1A). In contrast, LDs (stained with BODIPY 493/503) were easily detected in WT, pex24 and pex29 cells, whereas in pex23 and pex29 cells only a few faint spots were detected per cell. The latter may reflect decreased LD numbers or changes in their lipid composition, which could influence BODIPY 493/505 staining. We therefore also used Erg6-GFP as a protein marker for LDs. In pex23 and pex29 cells producing endogenous Erg6-GFP the number of fluorescent spots was less relative to pex24, pex32 and WT controls (Fig. 1A,B). Western blotting showed that Erg6-GFP protein levels were similar in all five strains tested (Fig. 1C), indicating that the lower number of LDs detected in pex23 and pex29 cells is not due to reduced levels of the marker Erg6-GFP. We therefore conclude that the reduced number of BODIPY 493/505 or Erg6-GFP marked spots is indeed reflecting a lower LD abundance in pex23 and pex29 cells.

Interestingly, pex23 and pex29 cells, but not pex24 or pex32 cells, also displayed aberrant mitochondrial profiles, as evident from confocal laser scanning microscopy (CLSM) analysis of Mitotracker-stained cells (Fig. 1D, upper panel). Inspection of 3D models, constructed from these CLSM images, revealed that in WT, pex24, and pex32 cells mitochondria exhibited a tubular network. However, in pex23 and pex29 cells mitochondria appeared more clustered (Fig. 1D, lower panel). Quantification of the mitochondrial branch length revealed that mitochondria were shorter in pex23 and pex29 cells compared to WT cells (Fig. 1E).

Summarizing, our FM data show that the absence of Pex23 and Pex29, but not of Pex24 or Pex29, affects mitochondria and LDs. Aberrant LD formation was reported previously for S. cerevisiae cells lacking Pex30, an S. cerevisiae member of the Pex23 family (Choudhary et al., 2020; Joshi et al., 2018; Wang et al., 2018). We now for the first time observed alterations in mitochondrial morphology in cells lacking specific Pex23 family members.

To better understand the alterations in mitochondrial morphology, electron microscopy (EM) analysis was performed on KMnO₄ fixed/Epon embedded cells. Although this method can introduce changes in organellar shape, it is the method of choice to quantify the number of mitochondrial profiles or distances between organellar membranes (Wu et al., 2020). As shown in Fig. 2A, mitochondrial profiles were present throughout the cells in WT, pex24 and pex32 cells, in line with our FM observations. Quantification of the number of mitochondrial profiles per cell section revealed that thin sections of WT cells contained up to 4 mitochondrial profiles, with most sections containing 1 to 3 mitochondrial profiles. A very similar distribution was observed in sections of pex24 cells. However, in sections of pex23 and pex29 cells, and to a lesser extend of pex32 cells, a higher number of mitochondrial profiles was observed (Fig. 2A).

Subsequently, we quantified the distance between mitochondrial profiles, which revealed that mitochondria are more clustered in pex23 and pex29 mutants relative to pex24, pex32 and WT cells (Fig. 2B).

To study mitochondrial morphology in more detail, serial sections were analysed of cryo-fixed and freeze substituted cells. This method better preserves the native shape of cell organelles. As shown in Fig. 2C mitochondrial fragmentation and clustering was similar as observed in the chemically fixed cells (Fig. 2A). Analysis of the serial sections and 3D rendering confirmed that pex23 and pex29 cells contain multiple, highly clustered mitochondria (Fig. 2C; Fig. S1).

Yeast Dnm1 and Fzo1 are crucial proteins in mitochondrial fission and fusion, respectively (Hermann et al., 1998; Sesaki and Jensen, 1999). As expected, the number of mitochondrial profiles in thin sections of H. polymorpha dnm1 cells is strongly reduced, whereas an increase is observed in fzo1 cell sections (Fig. 2A; corresponding FM images are shown in Fig. S2). In dnm1 cells, the distance between mitochondrial profiles is similar as in the WT control. However, in fzo1 cells mitochondria are much more clustered, similar as observed in pex23 and pex29 cells (Fig. 2B).

In conclusion, the mitochondrial phenotype of pex23 and pex29 cells resembles that of fzo1 cells and is characterized by the presence of enhanced numbers of highly clustered mitochondria.

To determine whether the altered mitochondrial morphology resulted in changes in mitochondrial activity, we stained glucose-grown cells with Rhodamine123 (Rh123). Rh123 is a cell-permeable cationic green, fluorescent dye that is taken up by active mitochondria and commonly used for the detection of the mitochondrial membrane potential in yeast (Drakulic et al., 2005; Ludovico et al., 2001). As shown in Fig. 3A, Rh123 fluorescence intensities were reduced in pex23 and pex29 cells compared to pex24, pex32 and wild-type (WT) control cells (Fig. 3A). Moreover, the absence of Pex23 and Pex29, but not of Pex24 and Pex32, resulted in slightly retarded growth on media containing glucose as sole carbon source (Fig. 3B).

The presence of Pex23 family members at peroxisome-ER membrane contact sites and at NVJs suggests that these proteins could be common ER contact site proteins. To analyse whether HpPex23 occurs at ER-mitochondria contact sites, we performed correlative light and electron microscopy (CLEM). Pex23-GFP was slightly overproduced to obtain sufficient fluorescence signal for CLEM analysis. CLEM analysis revealed that Pex23-GFP localizes to the peripheral ER, including at mitochondrion-ER contact sites (Fig. 4A). In line with our earlier observation, Pex23-GFP accumulates at the NVJs (Fig. 4A), but we did not observe enhanced levels of Pex23-GFP at mitochondrion-ER contact sites.

EM analysis of thin sections revealed that the percentage of mitochondria that are closely associated with the ER is similar in all four mutant strains and the WT control (Fig. 4B). This indicates that the absence of Pex23 family proteins does not result in a reduction of the physical contacts between the ER and mitochondria.

Previous studies in S. cerevisiae revealed that an artificial ER-mitochondrion tether can suppress the phenotype of mutants lacking a component of the membrane contact site ERMES (ER-mitochondrion encounter structure) (Kornmann et al., 2009). Similarly, we constructed an artificial tether, called ERMIT, consisting of full-length Tom70 at the N-terminus followed by GFP and finally the C-terminal domain of the ER tail anchored protein Ubc6 (Fig. 4C). EM analysis revealed extensive regions were mitochondria were in close contact with the ER and nuclear envelope in WT cells producing ERMIT (Fig. 4D). FM analysis confirmed the localization of the GFP containing ERMIT protein at mitochondria in all three strains (Fig. 4E; Fig. S3). Quantitative analysis showed that the introduction of ERMIT resulted in a larger percentage of mitochondria that were closely (<10 nm distance) associated with the ER, as expected (Fig. 4H). Interestingly, the introduction of ERMIT partially restored the growth defect of pex23 and pex29 cells (Fig. 4I). However, the morphological phenotypes of the mutants (enhanced numbers of mitochondrial profiles, Fig. 4F, and more clustering, Fig. 4G) were not changed by ERMIT. A slight increase in distance between mitochondria was observed for the WT control (Fig. 4G). Possibly, the artificial ER-mitochondrial contacts keep the mitochondria a bit more apart from each other.

Summarizing our data show that introduction of ERMIT results in closer physical contacts between mitochondria and the ER in pex23 and pex29 cells accompanied by partial suppression of the growth phenotype. This suggests that the absence of Pex23 and Pex29 influences the function of mitochondrion-ER contact sites.

The increased number of mitochondrial profiles in pex23 and pex29 cells may be due to changes in mitochondrial fusion and/or fission. Western blot analysis of cells producing Fzo1-GFP or Dnm1-GFP under control of their endogenous promoters, revealed that the levels of Fzo1-GFP were reduced in pex23 and pex29, while Dnm1-GFP levels were the same as in the WT control (Fig. 5A,B). FM analysis showed that the localisation of Fzo1-GFP and Dnm1-GFP was similar to what observed in the WT control in pex23 and pex29 cells (Fig. 5C). Reduced fusion, accompanied by normal fission, may explain the enhanced numbers of mitochondrial profiles in pex23 and pex29. Interestingly in pex32 cells both Fzo1-GFP and Dnm1-GFP levels were reduced (Fig. 5A). This may explain why the number of mitochondrial profiles is much less increased in pex32 cells (Fig. 2A).

Deletion of DNM1 in S. cerevisiae fzo1 cells largely restored the fragmented mitochondrial phenotype (Sesaki and Jensen, 1999). We hypothesized that in H. polymorpha pex23 and pex29 cells, which have reduced Fzo1 levels, the mitochondrial phenotype is suppressed by DNM1 deletion too. FM analysis of pex23 and pex29 cells in which DNM1 was deleted showed that mitochondrial fragmentation was partially restored, as expected (Fig. 6A). This was also evident from quantitative analysis of the length of mitochondrial branches (Fig. 6B), measurements of the number of mitochondrial profiles (Fig. 6C) and distances between mitochondria (Fig. 6D). Also, mitochondrial activities increased (Fig. 6E) and cell growth on glucose was partially restored in the double deletion strains (Fig. 6F,G).

Taken together, these results show that deletion of DNM1 in pex23 and pex29 partially restored the mitochondrial defects, similar to what was observed for S. cerevisiae fzo1. As a consequence, these observations point to a direct or indirect role of Pex23 and Pex29 in mitochondrial fusion.

In this study, we show for the first time that the absence of members of the Pex23 protein family can lead to alterations in mitochondrial morphology and activity. Our data indicate that deletion of PEX23 or PEX29 in the yeast H. polymorpha leads to aberrant mitochondrial morphology. In the same two mutants the number of LDs is also reduced. Similarly, the absence of the Pex23 family protein Pex30 in S. cerevisiae resulted in less and smaller LDs (Joshi et al., 2018; Wang et al., 2018).

Different from what was reported for S. cerevisiae, peroxisome biogenesis and LD formation appeared not to be linked in H. polymorpha. In S. cerevisiae the same, specialized subdomain of the ER, which contains Pex30 and proteins involved in LD formation, is involved in peroxisome and LD formation (Choudhary et al., 2020; Joshi et al., 2018; Wang et al., 2018). We here show that in H. polymorpha Pex24 and Pex32 have crucial roles in peroxisome biology (Wu et al., 2020), while Pex23 and Pex29 are important for LD formation (this study). Moreover, Pex23 family proteins involved in peroxisome biology and LD formation localize to different regions of the ER, namely at peroxisome-ER contacts (HpPex24 and HpPex32) or other ER regions, including NVJs (HpPex23, HpPex29) (Wu et al., 2020).

The altered mitochondrial morphology in H. polymorpha pex23 and pex29 cells is not indirectly due to peroxisome biogenesis defects, because pex24 cells, which show very severe peroxisome biogenesis defects, did not exhibit any mitochondrial abnormalities. Conversely, in pex29 cells, which show mitochondrial defects, peroxisome biogenesis is normal (Wu et al., 2020).

EM analysis of pex23 and pex29 cells revealed an increased number of smaller mitochondria that are highly clustered, suggesting reduced mitochondrial fusion. In S. cerevisiae a block in mitochondrial fusion, caused by the absence of Fzo1, also results in fragmented mitochondria that cluster in one area of the cell (Hermann et al., 1998). We observed a similar phenotype for H. polymorpha fzo1 (Fig. 2A). Indeed, the mitochondrial morphology of H. polymorpha fzo1 cells highly resembles that of H. polymorpha pex23 and pex29 cells. A defect in mitochondrial fusion is underscored by the observation that deletion of DNM1 suppressed the morphological mitochondrial phenotype of pex23 and pex29 cells (i.e. less mitochondrial profiles and less clustering). Moreover, DNM1 deletion also restored the reduced mitochondrial membrane potential and resulted in suppression of the growth phenotype. Together these observations point to a reduction of mitochondrial fusion. Essentially similar observations were made when DNM1 was deleted in the mitochondrial fusion mutant S. cerevisiae fzo1 (Bleazard et al., 1999; Sesaki and Jensen, 1999). Interestingly, the protein levels of Fzo1 were reduced in H. polymorpha pex23 and pex29 cells (Fig. 5A), while Dnm1 levels were unchanged. Why Fzo1 levels are reduced in the absence of Pex23 or Pex29 remains unknown. In S. cerevisiae Fzo1 levels are highly regulated, including degradation of the protein by the ubiquitin proteasome system (Cavellini et al., 2017). However, so far, the regulation of Fzo1 degradation has not been studied in H. polymorpha.

Cells of the PEX32 deletion strain also showed increased mitochondrial numbers, however, these organelles were not clustered. In these cells both Fzo1 and Dnm1 levels were reduced, most likely resulting in less fission and fusion of the organelles. Why the levels of these proteins are reduced in this mutant is still unknown.

HpPex24 and HpPex32 are important for the formation of ER-peroxisomal contact sites (Wu et al., 2020). We therefore asked whether the effects of the absence of HpPex23 and HpPex29 on mitochondria may be related to a role in ER-mitochondrion contact sites. Although the absence of these proteins did not result in reduced ER-mitochondrion contacts (Fig. 4B), enhancing the contacts by the introduction of an artificial ER-mitochondrion tether protein partially repressed the retarded growth of pex23 and pex29 cells (Fig. 4I). While mitochondrial function was partially restored by the tether, no significant effects could be detected in mitochondrial fragmentation and clustering (Fig. 4F,G). Still our observations suggest that the absence of Pex24 or Pex29 may influence (the function of) ER-mitochondrial membrane contact sites.

In conclusion, we here show that proteins of the Pex23 family proteins are not only involved in peroxisome and LD formation, but also play a role in the biology of mitochondria. This highlights the importance and functional diversity of this protein family.

H. polymorpha cells were grown in batch cultures at 37°C on mineral medium (Van Dijken et al., 1976) supplemented with 0.5% glucose or 0.25% glycerol as carbon source. When required, leucine was added to a final concentration of 60 μg/ml. For growth on plates, Yeast extract–Peptone–Dextrose (YPD) media (1% yeast extract, 1% peptone and 1% glucose) were supplemented with 2% agar. Transformants were selected using 100 μg/ml zeocin (Invitrogen), or 100 μg/ml nourseothricin (WERNER BioAgents) or 300 μg/ml hygromycin (Invitrogen).

The Escherichia coli strain DH5α used for cloning. E. coli cells were grown at 37°C in Luria broth (LB) media (1% Bactotryptone, 0.5% yeast extract and 0.5% NaCl) supplemented with 100 μg/ml ampicillin or 50 µg/ml kanamycin. For plates 2% agar was added.

The strains, plasmids and primers used in this study are listed in Table S1, S2 and S3, respectively. Plasmids integration was performed as described previously (Faber et al., 1994). All integrations were confirmed by PCR (Thermo Fisher Scientific). Gene deletions were confirmed by PCR and Southern blotting. All strains and plasmids are available on request.

To construct an fzo1 strain, a PCR fragment containing the FZO1 deletion cassette was amplified with primers dfzo1 fw and dfzo1 rev using plasmid pHIPN4 as a template. The PCR product was then transformed into yku80 cells to obtain the fzo1 mutant.

To construct the dnm1 pex23 and dnm1 pex29 double deletion strains, a PCR fragment containing the 3 kb DNM1 deletion cassette was obtained by PCR using primers Dnm1 5F and Dnm1 3R and pDEST DNM1-LEU as template. The PCR product was then transformed into H. polymorpha pex23 or pex29 cells to obtain the dnm1 pex23 and dnm1 pex29 double deletion strains. Correct integrations were confirmed by PCR and Southern blot analysis.

A PCR fragment containing PEX23-GFP was amplified with primers Pex23-F and Pex23-R using strain Pex23-GFP as a template. The obtained PCR product was digested with HindIII and SalI and inserted between the HindIII and SalI sites of plasmid pHIPH4, resulting in pHIPH4-PEX23-GFP. PflmI-linearized pHIPH4-PEX23-GFP was transformed into yku80. NsiI-linearized pHIPN4-DsRed-SKL was integrated into P_AOXPex23-GFP.

To introduce an artificial ER-mitochondrion tether in the WT, pex23 and pex29 strains, plasmid pHIPN18 Tom70(full)-mGFP-Ubc6 was constructed. In order to get these plasmids, pHIPH18 VPS39, pHS6A P_aoxPmp47-mGFP-Ubc6 and pHIPN Tom70 were constructed.

To construct pHIPH18 VPS39 plasmid, a PCR fragment containing VPS39 was amplified with primers Vps39 over fw and Vps39 over rev using genomic DNA as the template. The obtained PCR product was digested with HindIII and SalI and inserted between the HindIII and SalI sites of pHIPH4 plasmid, resulting in plasmid pHIPH4 VPS39. The ADH1 fragment was digested with NotI and HindIII from pHIPN18 PEX37 to replace the AOX fragment in pHIPH4 VPS39 to obtain pHIPH18 VPS39 plasmid.

To construct pHS6A P_aoxPmp47-mGFP-Ubc6, four PCR experiments were performed. To construct plasmid pH6SA-P_aox, a PCR fragment containing AOX promoter was amplified with primers AK-1 and AK-2 using genomic DNA as the template. The obtained PCR product was digested with SalI and PspXI and inserted between the SalI sites of pHS6A plasmid. The PMP47 fragment was obtained by performing PCR over the primers AK-3 and AK-4 using genomic DNA as the template. Plasmid pHIPZ-mGFP fusionator was restricted by HindIII and was performed Klenow fill in. The restricted plasmid pHIPZ-mGFP fusionator was digested by BglII, and PMP47 fragment was digested by BamHI. Two fragments were ligated to obtain plasmid pHIPZ-Pmp47-mGFP. The Pmp47mGFP fragment was amplified with primers AK-5 and AK-6 using pHIPZ-Pmp47mGFP as the template. The obtained PCR product was digested with PspXI and BamHI and inserted between the SalI and BamHI sites of plasmid pH6SA-P_aox, resulting in pHS6A-P_aoxPmp47-mGFP. The UBC6 fragment was obtained by performing PCR over the primers AK-7 and AK-8 using genomic DNA as the template. The obtained PCR product was digested with BamHI and inserted between the BamHI and SmaI sites of plasmid pHS6A-P_aoxPmp47-mGFP, resulting in pHS6A P_aoxPmp47-mGFP-Ubc6.

To construct pHIPN Tom70, a PCR fragment contains TOM70 was amplified with primers Fw Tom70 and Rv Tom70 using genomic DNA as the template. Use NotI and XbaI to restrict PCR fragment and plasmid pHIPN7 GFP-SKL and ligate two fragments to get pHIPN Tom70.

Plasmid pHIPN18 Tom70(full)-mGFP-Ubc6 was construct as follows. First a PCR fragment containing ADH1 promoter was amplified with primers F-Padh-primer-2 and R-Padh-primer using plasmid pHIPH18 VPS39 as a template. The PCR fragment was digested by enzyme PsiI and NotI, then inserted into plasmid pHIPN Tom70 to get plasmid pHIPN18-Tom70. A PCR fragment containing mGFP-UBC6 was amplified with primers F-GFP-UBC6-primer and R-GFP-UBC6-primer using plasmid pHS6A P_aoxPmp47-mGFP-Ubc6 as a template. Two XbaI/XhoI digested fragments from the obtained PCR fragment and pHIPN18-Tom70 were ligated to get plasmid pHIPN18 Tom70(full)-mGFP-Ubc6.

Then BstXI-linearized pHIPN18 Tom70(full)-mGFP-Ubc6 was transformed into WT, pex23, and pex29 strains to construct P_ADH1Tom70-mGFP-Ubc6 (ERMIT tether) expressing strains.

A plasmid encoding Erg6-GFP was constructed as follows: A PCR fragment encoding the C-terminus of ERG6 was amplified by using primers ERG6-fw and ERG6-rev with WT genomic DNA as a template; the obtained PCR fragment was digested with HindII/BamHI and ligated with the HindII/BglII digested pHIPN-Pex14-mGFP to get plasmid pHIPN-Erg6-mGFP. Subsequently, to construct the Erg6-mGFP expressing strains, BglII-linearized pHIPN-Erg6-mGFP was transformed into yku80, pex23, pex24, pex29 and pex32 strains.

A plasmid encoding Fzo1-GFP was constructed as follows: A PCR fragment encoding the C-terminus of FZO1 was amplified by using primers F-FZO1 and R-FZO1 with WT genomic DNA as a template; two HindII/BglII digested fragments from the obtained PCR fragment and pHIPN-Pex14-mGFP were ligated to get plasmid pHIPN-Fzo1-mGFP. Subsequently, to construct the Fzo1-mGFP expressing strains, BbvCI-linearized pHIPN-Fzo1-mGFP was transformed into yku80, pex23, pex24, pex29 and pex32 strains.

A plasmid encoding Dnm1-GFP was constructed as follows: two HindII/BglII digested fragments from the plasmid pHIPZ-Dnm1-GFP and pHIPN-Pex14-mGFP were ligated to get plasmid pHIPN-Dnm1-mGFP. Subsequently, to construct the Dnm1-GFP expressing strains, BstBI-linearized pHIPN-Dnm1-mGFP was transformed into yku80, pex23, pex24, pex29 and pex32 strains.

Rhodamine 123 was used to monitor the mitochondrial membrane potential. Cells were grown on 0.5% glucose and harvested in the log-phase and incubated with 50 nM Rhodamine123 (Invitrogen) for 30 min at 37°C. The fluorescence intensity of the cells stained with Rhodamine123 was analysed using a flow cytometer (BD Accuri™ C6 Plus) as described previously (Drakulic et al., 2005).

Cell extracts of TCA-treated cells were prepared for SDS-PAGE as described previously (Baerends et al., 2000). Equal amounts of protein were loaded per lane and blots were probed with anti-mGFP antibodies (Santa Cruz Biotechnology, #sc-9996; 1:2000 dilution), anti-elongation factor 1-α (EF1A) antibodies (1:10,000 dilution) (Kiel et al., 2007) or anti-pyruvate carboxylase 1 (Pyc1) antibodies (1:10,000 dilution) (Ozimek et al., 2007). Secondary goat anti-rabbit (Thermo Fisher Scientific 31,460, 1:5000 dilution) or goat anti-mouse (Thermo Fisher Scientific 31430, 1:5000 dilution) antibodies conjugated to horseradish peroxidase (HRP) were used for detection. EF1A or Pyc1 were used as loading controls. Enhanced chemiluminescence (Amersham ECL Prime, #RPN2232) was used to visualize proteins of interest following the manufacturer's instructions.

Blots were scanned using a densitometer (Bio-Rad, XRS+) and protein levels were quantified using ImageJ software. The density of each band measured was standardized by dividing the density of the corresponding loading control band. Relative density values were calculated by using the same standard sample (WT sample) as a common reference.

For vacuole staining, 1 ml of cell culture was supplemented with 1 μl FM4-64 (Invitrogen, #T13320), incubated for 60 min at 37°C, and analyzed. The lipid droplet dye BODIPY 493/503 (Invitrogen, #D3922) was used at a concentration of 1 µg/ml, incubated for 10 min at 37°C. Mitochondria were stained by adding 0.1 µl Mitotracker Orange (Invitrogen, #M7510; 1 mM) and Mitotracker Red (Invitrogen, #M7512; 1 mM) to 1 ml of cells. These cells were incubated for 5 min at 37°C and subsequently spotted on agar containing glucose. DAPI (Sigma-Aldrich, #32670; 1 µg/ml) was used for DNA staining (de Boer and van der Klei, 2023).

Images were obtained from the cells in growth media or from 200 nm thick cryosections for CLEM using a fluorescence microscope (Axioscope A1; Carl Zeiss) using Micro-Manager software and a digital camera (Coolsnap HQ²; Photometrics). The GFP and BODIPY fluorescence were visualized with a 470/40 nm band-pass excitation filter, a 495 nm dichromatic mirror, and a 525/50 nm band-pass emission filter. Mitotracker and FM4-64 fluorescence were visualized with a 546/12 nm band-pass excitation filter, a 560 nm dichromatic mirror, and a 575-640 nm band-pass emission filter. DAPI fluorescence was visualized with a 380/30 nm band-pass excitation filter, a 420 nm dichromatic mirror, and a 460/50 nm band-pass emission.

Airyscan images were captured with a confocal microscope (LSM800; Carl Zeiss) equipped with a 32-channel gallium arsenide phosphide photomultiplier tube (GaAsP-PMT), Zen2009 software (Cal Zeiss) and a 63×1.40 NA objective (Carl Zeiss).

Image analysis was performed using ImageJ and figures were prepared using Adobe Illustrator software. Mitochondrial morphology was analysed from 3D airyscan images using the Mitochondria analyzer plug-in in ImageJ (Chaudhry et al., 2020). Thresholded images from the mitochondria analyzer plug-in were used for 3D segmentation in the IMOD software package.

The number of LDs was quantified from Z-stacks of CLSM images. Quantification was performed automatically using a custom-made plugin from ImageJ (Thomas et al., 2015).

For morphological analysis, cells were harvested by centrifugation and washed three times with distilled water. Cells were fixed for 20 min with 1.5% potassium permanganate. After washing three times with distilled water cells were post-stained for 16 h with 0.5% uranyl acetate in distilled water. Cells were dehydrated in series of ethanol and embedded in Epon. 70 nm sections are collected on 100 mesh formvar coated copper grids and analysed with a CM12 transmission electron microscope (TEM) (Philips) running at 100 kV. Mitochondrial area and the distance between mitochondria are measured using ImageJ.

Serial sectioning was performed on cryo-fixed cells. These cells were cryo-fixed and freeze substituted as described previously using the self-pressurized rapid freezing and rapid freeze substitution method (Leunissen and Yi, 2009; Mcdonald and Webb, 2011; Wu et al., 2019). Cells were concentrated by centrifugation and inserted into copper specimen tubes (Leica, 16706871). Both ends were closed using pliers and plunged into liquid propane. The copper capillaries were sliced open longitudinally and placed on freeze-substitution medium containing 1% osmium tetroxide, 0.5% uranyl acetate and 5% water in acetone. After rapid freeze substitution samples were embedded in Epon and 60 nm serial sections were collected on formvar coated single slot copper grids and inspected with a CM12 TEM (Philips). Image alignment and 3D rendering was performed using the IMOD software package.

CLEM was performed as described previously (de Boer and van der Klei, 2023). After fluorescence imaging, the grid was post-stained and embedded in a mixture of 0.5% uranyl acetate and 0.5% methylcellulose. Acquisition of the double-tilt tomography series was performed manually in a CM12 TEM (Philips). The CM12 TEM ran at 100 kV and a tilt range of 45° to −45° with 2.5° increments was included. To construct the CLEM images, pictures taken with FM and EM were aligned using the eC-CLEM plugin in Icy (Paul-Gilloteaux et al., 2017). Reconstruction of the tomograms and alignment of the serial sections was performed using the IMOD software package.

We would like to thank Silke Vedelaar (Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, the Netherlands) for assistance with the membrane potential measurements.