Identification of distinct active pools of yeast serine palmitoyltransferase in sub-compartments of the ER

Sphingolipids (SPs) are an essential class of lipids mainly found in the outer leaflet of the plasma membrane. They can act as signaling molecules as well as structural components of the plasma membrane (Cartier and Hla, 2019; van Meer et al., 2008). In order to allow cells and membranes to adapt, SP levels respond to different environmental conditions. For example, SP levels in budding yeast (Saccharomyces cerevisiae) are changed in different carbon sources (Klose et al., 2012) and exposure of cells to heat rapidly elevates SP levels (Dickson et al., 1997). In mammalian cells, changes in SP composition in response to external stimuli, for example the amount of carbon source (Mondal et al., 2022), are also known. In particular, under conditions where increased amounts of SPs are required, as in the formation of the myelin sheath, increased SP biosynthesis is observed (Davis et al., 2020).

SP biosynthesis begins in the endoplasmic reticulum (ER) with the enzyme serine palmitoyltransferase (SPT) (Buede et al., 1991; Gable et al., 2000; Nagiec et al., 1994). SPT catalyzes the first and rate-limiting step of SP biosynthesis by condensing L-serine and palmitoyl-CoA to form 3-ketosphinganine (3-KS). This short-lived intermediate is reduced to dihydrosphingosine (DHS), which in yeast is further hydroxylated to form phytosphingosine (PHS). DHS and PHS together are also known as long chain bases (LCBs). Very long chain fatty acids (VLCFAs) with 24 or 26 carbon atoms are then amide-linked to LCBs to form ceramides (D'mello et al., 1994; Guillas, 2001; Vallée and Riezman, 2005). Ceramides are transported to the Golgi through vesicular and non-vesicular transport (Ikeda et al., 2020; Kajiwara et al., 2013; Limar et al., 2023; Liu et al., 2017). In the Golgi, inositol-containing head groups are added to form complex SPs, which are then transported to the plasma membrane (Klemm et al., 2009).

SP biosynthesis is highly regulated with multiple input signals. The SPT in yeast forms the SPOTS complex, which comprises SPT (itself a heterodimer of Lcb1 and Lcb2) with the Orm proteins Orm1 and Orm2, the small subunit Tsc3 and the phosphatidylinositol-4-phosphate (PI4P) phosphatase Sac1 (Breslow et al., 2010; Han et al., 2010). Although the mammalian SPT acts as a dimer, we have recently postulated that the yeast complex forms a monomer with either Orm1 or Orm2 bound to it (Schäfer et al., 2023). In yeast, SPT activity is regulated by phosphorylation of the Orm proteins, which relieves their inhibitory activity. Orm phosphorylation originates either from the plasma membrane-localized target of rapamycin complex 2 (TORC2) signaling pathway via the Ypk kinases 1 and 2 (Ypk1/2) or from the TORC1 signaling pathway starting at the vacuole (Berchtold et al., 2012; Niles and Powers, 2012; Roelants et al., 2011; Shimobayashi et al., 2013; Tabuchi et al., 2006). In addition, the SPT is regulated via the downstream metabolite ceramide itself. This evolutionary conserved mechanism relies on the direct interaction of an Orm protein and the SPT with a ceramide molecule (Davis et al., 2019; Liu et al., 2023; Schäfer et al., 2023; Xie et al., 2023). The two Orm proteins also underlie different regulatory mechanisms. Only the Orm2 protein is a substrate for the endosome/Golgi-associated degradation (EGAD) pathway, which depends on its phosphorylation by Ypk1/2, export from the ER, ubiquitylation and finally degradation via the proteasome (Bhaduri et al., 2023; Schmidt et al., 2019). In addition, it has been suggested that upregulation of SPT activity is directly linked to the uptake of exogenous serine via the Gnp1 amino acid permease (Esch et al., 2020). Other regulatory mechanisms controlling SP biogenesis involve the regulation of VLCFA biosynthesis (Olson et al., 2015; Zimmermann et al., 2013) and the post-translational regulation of the ceramide synthase (Muir et al., 2014).

Together, this highlights a complex regulatory network of SP biosynthesis. However, the molecular mechanism underlying phosphorylation-mediated regulation of the Orm proteins still remains enigmatic. It is not clear how the rapid increase in LCB biosynthesis upon heat shock can be aligned with the relatively slow EGAD degradation pathway acting on Orm2. In addition to the temporal regulation, spatial components of SPT regulation have not been studied at all.

To shed light on these processes, we use a nanobody-based tool allowing us to recruit the SPOTS complex to different sub-compartments of the ER, namely the nuclear and the peripheral ER. By combining this re-localization tool with SP biosynthesis flux analysis, we are able to determine differentially active pools of the SPT in subcellular compartments. We can show that the nuclear ER-localized SPT pool is less active than the peripheral ER-localized SPT pool. We believe that our study will be a starting point for the analysis of SP metabolism with subcellular resolution.

Despite multiple efforts, the exact mechanism of Orm-dependent regulation of SPT remains enigmatic. Most experiments regarding Orm protein phosphorylation and its effect on SPT activity have been conducted in ORM deletion strains or in the presence of SP metabolism inhibitors, such as myriocin. However, it is problematic to directly inhibit the enzyme whose Orm-dependent regulation is to be investigated. In addition to the complicated regulatory mechanism, we previously proposed that spatially separated SPT pools, specifically the nuclear envelope-resident pool and the peripheral ER-resident pool, are differentially regulated (Esch et al., 2020). We now aimed to study SP biosynthesis regulation under heat shock, a physiological condition known to rapidly upregulate SPT activity, with subcellular resolution.

First, we compared the levels of LCBs under control conditions and after a 5-min heat shock in WT cells using targeted lipidomics. As previously reported (Dickson et al., 1997), we measured a rapid four-fold increase in LCBs after the short treatment (Fig. 1A). We had previously demonstrated that increased de novo LCB biosynthesis is directly dependent on the uptake of exogenous serine through the general amino acid permease Gnp1 (Esch et al., 2020). Consistent with our previous findings, deletion of GNP1 resulted in a reduced heat shock response, whereas deletion of the endogenous serine biosynthesis pathway (ser2Δ) had no effect on LCB biosynthesis. Deletion of the Gnp1 homolog AGP1 did not affect the LCB increase after heat shock. The double deletion of GNP1 and AGP1 phenocopied the effect seen in the gnp1Δ strain (Fig. 1A). The incomplete block of heat shock-induced LCB biosynthesis in a gnp1Δ strain might be due to a cytoplasmic pool of serine that can be utilized under these conditions. In summary, our experiments demonstrate that a short heat shock enables the investigation of SPT regulation in yeast cells.

Next, we assessed the individual contribution of the Orm proteins to heat shock-dependent LCB upregulation. Deletion of ORM2 resulted in a significant increase in steady-state LCB levels compared to in wild-type (WT) control cells. Consistent with previous findings, we detected a small increase in LCB levels after deletion of ORM1, which was less pronounced than the effect of orm2Δ (Han et al., 2010) (Fig. 1B). In contrast, deletion of both ORM genes led to a 25-fold increase in LCBs, as previously reported (Breslow et al., 2010) (Fig. 1C). Taken together, these results suggest that one of the two Orm proteins is sufficient to maintain LCB levels within a physiological range under basal conditions. To investigate whether each Orm protein is also capable of maintaining LCB levels under rapidly changing conditions, we exposed orm1Δ, orm2Δ and orm1Δ orm2Δ cells to heat shock, measured their LCB levels, and compared their heat shock-mediated increase of LCBs to those of a WT yeast strain (Fig. 1D). Both single deletion strains were still able to respond to the temperature change with a rapid increase in LCB biosynthesis. However, the heat shock-dependent LCB increase was significantly lower in orm2Δ cells, with a three-fold increase compared to a four-fold increase in orm1Δ and WT cells. In contrast, the orm1Δorm2Δ strain did not show any further increase in the already very high LCB levels after heat shock (Fig. 1D). Taken together, this suggests that both Orm proteins contribute to increased LCB biosynthesis after heat shock, with Orm2 making the greater contribution.

Subsequently, we evaluated whether the short 5-min heat shock resulted in increased phosphorylation of the negative SPT regulators Orm1 and Orm2. We observed increased phosphorylation for Orm2 but not for Orm1 after heat shock (Fig. 1E; quantification in Fig. S1A). This finding is consistent with previous observations where increased phosphorylation of Orm2 was observed after 2 min, followed by its decrease after 10 min of heat shock (Sun et al., 2012). Interestingly, basal phosphorylation levels were already high under normal growth conditions (Orm1∼90%; Orm2∼60%) (Fig. S1A).

To determine whether Orm phosphorylation is the key mediator in heat shock-dependent SPT regulation, we analyzed non-phosphorylatable ORM mutants lacking the three Ypk1-regulated phosphorylation sites [Orm1 S51A S52A S53A (Orm1-3A); Orm2 S46A S47A S48A (Orm2-3A)] (Breslow et al., 2010; Roelants et al., 2011). Indeed, the expression of either Orm1-3A or Orm2-3A resulted in reduced LCB levels compared to WT cells. Cells expressing both Orm1-3A and Orm2-3A had four-fold reduced LCB levels compared to WT cells (Fig. 1F). Each strain, WT, Orm1-3A, Orm2-3A and Orm1-3A Orm2-3A showed a three-fold increase in LCB levels after heat shock (Fig. S1B). However, the increase of LCBs in all mutant strains resulted in total levels resembling the LCB levels of a WT strain under normal growth conditions (Fig. 1F). Thus, the strong four-fold increase of LCB levels in WT cells after heat shock directly depends on the phosphorylation of Orm proteins. However, the Orm-3A mutants show a small regulatory potential that appears to be independent of Ypk phosphorylation. Taken together, in the short time frame of increased LCB biosynthesis after heat shock, the necessity of Ypk signaling and the importance of serine uptake suggest that a local upregulation of an SPT pool close to the plasma membrane is possible.

To further pursue this hypothesis, we tested whether peripheral ER-restricted TORC2 and Ypk1 signaling is sufficient for the Orm-dependent heat shock response of yeast cells. Previous studies have indicated that the SPT is differentially regulated in the peripheral ER, where a decrease in Orm2 is observable after myriocin treatment (Breslow et al., 2010). The degradation of Orm2 via the EGAD pathway is initiated by phosphorylation from the cytosolic Ypk1 kinase, which is activated in response to signals from the plasma membrane (Berchtold et al., 2012; Niles and Powers, 2012; Schmidt et al., 2019). To determine whether Orm-mediated regulation of SPT mainly occurs in the peripheral ER or whether regulation in the entire ER is necessary, we tethered Ypk1 to the plasma membrane using a CAAX box (Tang et al., 2009), thus restricting its signaling to the plasma membrane. In comparison to GFP-tagged Ypk1, a GFP-tagged Ypk1 harboring a CAAX motif after the GFP was localized at the plasma membrane (Fig. 2A). We assessed the functionality of Ypk1–CAAX using genetic interaction studies and tetrad analysis. As previously reported, the deletion of both YPK1 and YPK2 is lethal (Fig. 2B) (Roelants et al., 2002). Expression of the Ypk1–CAAX allele was not sufficient to rescue this lethal phenotype (Fig. 2C). The additional deletion of both ORM genes was sufficient to rescue both the YPK1 YPK2 double deletion (Fig. 2B) (Roelants et al., 2011) and the ypk2Δ Ypk1-CAAX strain (Fig. 2C). This suggests that the lethality induced by the deletion of YPK1/2 is mediated by the dysregulation of SP biosynthesis, although several other targets of Ypk1/2 are known (as previously described by Muir et al., 2014; Roelants et al., 2011).

We next added an additional 103 amino acid linker between Ypk1 and the CAAX box to bridge the ∼30 nm distance between the peripheral ER and the plasma membrane (Gatta et al., 2015; Stradalova et al., 2012; West et al., 2011). This would theoretically allow phosphorylation of the Orm proteins at the peripheral ER from Ypk1–linker–CAAX at the plasma membrane. We confirmed tethering to the plasma membrane with the additional linker using microscopy and tested its functionality using negative genetic interactions with ypk2Δ (Fig. 2D,E). YPK1–linker–CAAX showed no growth defects when combined with a YPK2 deletion, demonstrating the functionality of the construct. To exclude major differences induced by the tethering of Ypk1 to the plasma membrane, we also assessed the proteome of YPK1–linker–CAAX cells. First, we compared ypk2Δ cells to WT cells using label-free proteomics. The YPK2 deletion induced no difference in the proteome, and this was also the case when Ypk1 was tethered to the plasma membrane. The only difference we detected was the decrease of Ypk1 in ypk2Δ YPK1-linker-CAAX cells compared to ypk2Δ cells or WT cells (Fig. S2A–C). Indeed, cells expressing Ypk1–linker–CAAX had lower basal phosphorylation levels of both Orm1 and Orm2 as expected if only a fraction of the Orm proteins is available at the peripheral ER for phosphorylation (Fig. 2F,G; Fig. S2C,D). We were unable to detect the previously observed small increase in phosphorylation of Orm2 after heat shock under these conditions (Fig. 2G; Figs S2D, S1A). This might reflect the lack of sensitivity in the used assays. However, when we determined the increase in LCB levels after heat shock, we could not detect differences between a WT strain, a ypk2Δ strain and a ypk2Δ Ypk1–linker–CAAX-expressing strain. Together, these results suggest that small changes in the phosphorylation state of the Orm proteins can have significant impact on SPT activity and that plasma membrane-restricted Ypk1 signaling is sufficient for Orm regulation. To further dissect the spatial and temporal coordination of SP biosynthesis, we aimed to develop a novel tool to analyze SPT regulation with subcellular resolution.

As mentioned above, we have previously suggested that SPT activity is directly coupled to Gnp1-dependent serine uptake at the plasma membrane and Orm2 levels appear to be especially sensitive to changes in the peripheral ER (Breslow et al., 2010; Esch et al., 2020). We reasoned that we need tools to isolate the different intracellular pools of the SPT from both the peripheral ER and the nuclear ER (Fig. 3A). We developed a nanobody (NB)-based recruiting system that depends on a short peptide tag on the SPT as well as a corresponding NB that either targets it to the peripheral ER or to the nuclear ER. The latter is achieved by fusing the NB to either Rtn1 (for the peripheral ER) or the membrane anchor (amino acids 1–121) of Nvj1 (for the nuclear ER) (Fig. 3A) (Craene et al., 2006; Kvam and Goldfarb, 2006; Millen et al., 2008). As a proof-of-principal experiment, we generated a diploid yeast strain expressing one copy of Lcb1 fused to a GFP and the other copy of Lcb1 fused to mKate and a short ALFA tag (Götzke et al., 2019). The two proteins colocalized in both the peripheral and the nuclear ER. When we co-expressed a GFP-NB fused to Nvj1_1-121 and an ALFA-NB fused to Rtn1 we were able to completely separate both pools of Lcb1 (Fig. 3B). While the recruitment was successful, we carefully examined whether the tagged SPT subunits were functional. We therefore tagged Lcb1 and Lcb2 with either GFP or the short ALFA tag at either the N-terminus or the C-terminus and performed growth tests on control plates or plates containing 0.4 µM myriocin (Wadsworth et al., 2013). Although all strains grew normally under control conditions, only an N-terminally ALFA-tagged Lcb2 strain expressed under the control of its endogenous promotor grew to a similar level to a WT strain on myriocin (Fig. 3C). This suggests that most tags do interfere with the delicate regulation of SPT activity. Based on these results we decided to work in haploid strains recruiting the entire SPT population to either the peripheral or the nuclear ER via the intracellular expressed ALFA-NB (Götzke et al., 2019).

Hereafter, we will refer to cells harboring ALFA–Lcb2 as all ER SPT (SPT^allER) cells. Cells with ALFA–Lcb2 and the nuclear ER ALFA-NB will be referred to as nuclear ER SPT (SPT^nER) cells. Cells with ALFA–Lcb2 and a peripheral ER ALFA-NB will be referred to as peripheral ER SPT (SPT^pER) cells (Fig. 4A). First, we utilized the GFP-tagged form of Lcb1, the other catalytically active subunit of the SPT, to confirm recruitment to both parts of the ER. Microscopy confirmed the co-recruitment of the entire population of Lcb1–GFP and thereby also of the ALFA-tagged subunit Lcb2 (Fig. 4B). Next, we analyzed the co-recruitment of the other subunits of the SPOTS complex namely, GFP–Orm1, GFP–Orm2 and GFP–Sac1. Similar to Lcb1, we were also able to co-recruit GFP–Orm1 and GFP–Orm2 to the different ER sub-compartments (Fig. 4C,D). However, we noticed that a small population of GFP–Orm1 and GFP–Orm2 were not co-recruited, suggesting that a pool of Orm proteins exists in the cells that is not bound in the SPOTS complex. In contrast to the other tested proteins, Sac1 recruitment was only possible in the peripheral ER. We did not manage to recruit the Sac1–GFP protein to the nuclear ER via the Nvj1_1-121-NB fusion protein (Fig. 4E). This can be either explained by there being a free pool of Sac1 in the peripheral ER, or by stable protein–protein interactions of Sac1 at the ER–plasma membrane contact site. In line with this observation, Sac1 has been proposed to interact with the VAP proteins (Manford et al., 2012). Alternatively, the tag on Lcb2 could interfere with the protein–protein interaction between Lcb2 and Sac1, which is exclusively mediated via its N-terminus (Schäfer et al., 2023). To directly test whether recruitment interfered with protein–protein interactions we performed co-immunoprecipitations of the recruited SPOTS complex followed by label-free mass spectrometry (MS)-based proteomics. Importantly, the affinities of the ALFA-NB with the ALFA tag are so high that it cannot be used for recruitment and co-immunoprecipitations in parallel. We therefore fused another short peptide tag, the SPOTS tag, in front of the ALFA tag. The free SPOTS tag was used for immuno-affinity purification of the SPT^allER, SPT^nER and SPT^pER. The proteomics analysis revealed that all subunits of the SPOTS complex were co-enriched even in the recruited conditions, suggesting that intracellular recruiting does not interfere with the protein–protein interactions within the SPOTS complex. Further, the co-recruitment of the small regulatory subunit Tsc3 was confirmed (Fig. 5A–C).

To further exclude that the tagging of Lcb2 or SPT recruitment did affect the expression of other proteins, we also analyzed the proteome of the three strains used for recruitment, including SPT^allER, SPT^nER, SPT^pER and WT cells. No significant changes in protein abundance were observed among the 2945 identified proteins, and there were only minor changes in the expression levels of components of the SPOTS complex and Rtn1 (Fig. S3A–C). To control whether tagging of Rtn1 affected its function, we utilized the negative genetic interactions of RTN1 with SPO7 to assess its functionality after peripheral ER rewiring. Spo7 is a component of the Nem1–Spo7 protein phosphatase complex, which controls the function of Pah1 and is essential for triacylglycerol synthesis and nuclear ER morphology (Siniossoglou, 1998; Su et al., 2018). Our tetrad analysis showed negative genetic interactions between spo7Δ and rtn1Δ, whereas RTN1-ALFA-nB spo7Δ cells displayed normal growth, indicating that tagging Rtn1 for SPT rewiring did not affect its function (Fig. S3D,E). Additionally, we measured the lipidome in the rewired strains to detect possible differences due to SPT rewiring. We found only minor changes between the three tested strains (SPT^allER, SPT^nER and SPT^pER) (Fig. 5D–F). Thus, any potential side effects on SPT activity resulting from changes in the lipidome can be excluded. In summary, the NB-based recruitment system works for the SPOTS complex in yeast. As shown here, this system can be utilized to measure protein–protein interactions in vivo using fluorescence microscopy as a readout.

To test whether the NB recruitment system allows to detect changes in the activity of the different SPT pools, we measured LCB levels in the SPT^allER, SPT^nER and SPT^pER strains using targeted lipidomics under control and heat shock conditions. LCB increase after heat shock was not significantly changed in response to cellular SPT localization (Fig. 6A). We also analyzed phosphorylation of FLAG–Orm1 and FLAG–Orm2 in the strains with differentially recruited SPTs. Similar to our previous experiments, we were not able to detect significant phosphorylation changes in the background of the high overall phosphorylation levels. (Fig. 6B,C, quantification in Fig. S4A,B). Next, we investigated whether the deletion of either of the ORM genes had different effects in the SPT recruited strains in control and heat shock conditions. Targeted lipidomics revealed that the deletion of ORM1 had only a minimal effect on LCB levels in the SPT^allER strain, the SPT^pER strain and the SPT^nER strain (Fig. 6D). In contrast, deletion of ORM2 alone led to a nearly four-fold increase in LCB levels in the SPT^allER strain and the SPT^pER strains and to a smaller increase in the SPT^nER strain (Fig. 6E). Exposing the cells to 5 min of heat shock led to a five-fold increase in LCB levels in all orm1Δ strains (Fig. 6F). In contrast, exposing orm2Δ strains to heat shock conditions only resulted in a small increase in LCB levels under all conditions (Fig. 6G). Together, we find small differences in the activities of the recruited SPT strains depending on the overall localization of the SPT together with their inhibition by Orm1 and Orm2.

To test whether the lack of differences in the LCB levels in the analyzed strains is based on the lack of sensitivity of our detection method, we used pulse labeling of LCBs with ¹³C₃¹⁵N₁-serine (Esch et al., 2020; Martínez-Montañés et al., 2020). During the SPT-catalyzed condensation of serine and palmitoyl-CoA, carbon dioxide from the serine is lost. Therefore, we expect a mass difference of +3 for all ¹³C₃¹⁵N₁-serine-labeled SPs (Fig. 7A). We added labeled serine and followed its incorporation into LCBs and ceramide over a time course of 30 mins in SPT^allER, SPT^nER and SPT^pER strains. This analysis showed that the incorporation of serine was slower in the SPT^nER strains compared to the SPT^allER and SPT^pER strains, suggesting that the activity of the SPT is lower when recruited to the nuclear ER (Fig. 7B). The same effect was observed with a 30 min delay in the levels of labeled ceramides (Fig. 7C). Interestingly, the overall levels of LCBs and ceramides remained largely unchanged under the conditions used (Fig. 7B,C). When we exposed the cells to heat shock and measured the incorporation of labeled serine into LCBs and ceramides, we observed comparable incorporation rates in the SPT^nER strain (Fig. 7D,E). However, incorporation of labeled LCBs into ceramides was reduced in the SPT^nER cells after heat shock (Fig. 7F). Thus, decreased SPT^nER activity seems to result in an overall decreased flux through the SP biosynthesis pathway. In summary, combining the NB recruitment system with pulse labeling experiments allows the detection of altered activities of the SPT pools in different subcellular compartments.

The condensation of serine and palmitoyl-CoA catalyzed by the SPT is the rate limiting step in SP biosynthesis. It is clear that multiple input signals control this step, for example phosphorylation of the Orm proteins by the TORC2 signaling pathway as well as the levels of the downstream metabolite ceramide (Berchtold et al., 2012; Breslow et al., 2010; Davis et al., 2019; Niles and Powers, 2012; Roelants et al., 2011). Besides this known complexity of the regulatory network, we previously suggested that there are differentially regulated pools of the SPT in the nuclear and the peripheral ER (Esch et al., 2020). However, determining the activities of the same enzyme in different subcellular compartments is extremely challenging. Here, we used SPT as a model to develop a system that is able to determine lipid metabolic enzyme activities in different parts of the ER. We combined a NB-based recruitment system with a pulse labeling approach to analyze the activity and regulation of SPT in different ER sub-compartments. Together, the system we developed here is a first step towards studying lipid metabolism with subcellular resolution.

Our study has revealed additional insights into the complex SPT regulation network. We indeed find two different active SPT pools in cells, with the less active one at the nuclear ER. This would be in line with our previous hypothesis that serine taken up by the cells is preferentially incorporated into LCBs at the peripheral ER (Esch et al., 2020). This would argue for substrate availability as the major regulatory factor and argue against there being SPT pools that are differentially regulated by the YPK kinases. Our data also show that it is indeed challenging to directly correlate the phosphorylation levels of the Orm proteins with the activity of the SPT. Even under standard laboratory growth conditions, the majority of both Orm1 and Orm2 are phosphorylated at the Ypk-dependent serine residues. This is in line with previous studies that have also observed high basal phosphorylation levels (Breslow et al., 2010). Whether phosphorylated Orm proteins exit the SPOTS complex or whether phosphorylation leads to a structural re-orientation of the complex as well as the exact mechanism of Orm-mediated SPT inhibition remain a matter of debate. In a recent study, we showed that ceramide is stably bound to the SPOTS complex via interactions with both Lcb2 and Orm1 (Schäfer et al., 2023). Similar results have been obtained from studies of the human ORMDL–SPT complex and the Arabidopsis thaliana ORM–SPT complex (Liu et al., 2023; Xie et al., 2023). Ceramide has previously been shown to be a major regulator of SPT activity (Davis et al., 2019). This is further supported by data showing that mutations rendering the mammalian ceramide transfer protein (CERT) hyperactive also result in drastically increased SPT activity (Gehin et al., 2023). However, the exact mode of Orm-mediated inhibition appears to be different in yeast and humans. Whereas the human ORMDL N-terminus directly blocks the active site of the SPT, the yeast Orm1 N-terminus extends away from the active site and two tyrosine residues of Lcb2 appear to keep the bound ceramide molecule at the substrate entry tunnel. Importantly, the structure of the yeast SPOTS complex was solved in the presence of the Orm1-3A mutant, most likely reflecting an inhibited confirmation. In line with our results, even the Orm1-3A-bound complex maintains some residual SPT activity. Similarly, we now show, that yeast cells expressing both, Orm1-3A and Orm2-3A remain some residual SPT activity. In addition, both the human SPT–ORMDL complex and the Arabidopsis thaliana ORM–SPT complex also remain active in the presence of the Orm protein in vitro (Liu et al., 2023; Xie et al., 2023). One additional factor besides the phosphorylation of the Orm proteins after heat shock, could be that the increase in temperature also weakens the physical interaction between the SPT Orm complex and the ceramide. This would be similar to the concept of temperature-sensitive alleles in cells.

Another major difference in the regulation of the SPT between yeast and humans is the lack of the phosphorylated ORMDL N-terminus in humans. The phosphorylation of the yeast Orm proteins potentially results in a conformational change that allows the bound ceramide to dissociate from the complex. Our recruitment data show that the majority of the Orm proteins are co-recruited with the SPT, even if the Orm phosphorylation levels are already very high supporting a stable interaction between the SPT and phosphorylated Orm proteins. However, it was previously suggested that Orm2-3D does not associate with the SPT anymore (Han et al., 2019). The structural analysis of the SPOTS complex also revealed the presence of its monomeric form. This could explain why the two Orm proteins are differentially regulated by the EGAD pathway (Bhaduri et al., 2023; Schmidt et al., 2019). It is also noteworthy that the timing of the increase in LCB biosynthesis after 5 min of heat shock compared with that of the regulatory mechanism by the EGAD pathway is difficult to align. In summary, our data add novel insights into the complex regulatory network of SPT regulation and show that new analysis methods are required to dissect the regulatory mechanisms spatially and temporally. Importantly, our analyses also reveal that it is crucial to carefully evaluate the functionality of tags added to enzymes of SP metabolism.

Regarding the technical parts of our NB recruitment system, intracellular expressed NBs combined with small complementing peptide tags have previously been used to alter the localization of proteins in the cell (Traenkle et al., 2020). Similarly, the rapamycin-induced targeting system represents an inducible re-localization approach (Chen et al., 1995). Our analysis of the recruitment system that allows SPT to be recruited to different ER sub-compartments shows that it can be used to recruit entire protein complexes. Recruiting one subunit of a complex allows the co-recruitment of the other subunits and therefore this system can also be used as a system to measure protein–protein interactions in vivo, using fluorescence microscopy as a readout. Similar assays have been established for example in mammalian cells (Sotolongo Bellón et al., 2022). Interestingly, it was only possible to completely target the Sac1 subunit of the SPOTS complex to the peripheral ER and not to the nuclear ER, suggesting that a non-SPT-bound Sac1 pool localizes only to the peripheral ER. In line with this observation, Sac1 was previously shown to interact with other proteins in the cell including the VAP proteins, which form pER-plasma membrane contacts (Manford et al., 2012).

Combining the SPT recruitment system with pulse labeling of SP intermediates allows the detection of different SPT pools in the cell. So far it has only been possible to measure bulk activity of lipid metabolic enzymes in the cell. Other complementary approaches are, for example, lipidomics MALDI imaging mass spectrometry (MALDI-IMS), which allows the detection of lipids in a certain cellular sub-compartment (Dreisewerd et al., 2022; Soltwisch et al., 2015). The resolution of this approach is limited and therefore will be difficult to adapt to yeast cells. Other assays that allow the detection of local lipid biosynthesis, modification and lipid transport have been recently demonstrated by the Kornmann laboratory (John Peter et al., 2022a,b). There, yeast cells expressed non-yeast lipid modifying enzymes in different sub-compartments to allow the visualization of lipid biosynthesis and transport.

Why is it important to measure lipid metabolism with subcellular resolution? A prominent example is the activities of the phosphatidylserine decarboxylases (Psd) Psd1 and Psd2 in yeast cells. Both enzymes catalyze the same reaction, the decarboxylation of phosphatidylserine to phosphatidylethanolamine (Voelker, 1997). Whereas Psd1 is described as a mitochondrial and ER-localized enzyme, Psd2 is described as both endosomal and Golgi localized (Friedman et al., 2018; Gulshan et al., 2010). Approaches to target Psd1 to either mitochondria or the ER based on different targeting sequences have already been developed (Friedman et al., 2018). Our approach would allow similar experiments for proteins that cannot be targeted just by organelle targeting motifs. Another prominent example for the compartmentation of a membrane is the yeast plasma membrane. Here, amino acid transporter activity depends on their localization in and out of a specialized domain formed by eisosomes (Busto et al., 2018; Gournas et al., 2018; Walther et al., 2006). Other examples are the differences between the tubular ER and ER sheets in mammalian cells. Are certain lipid metabolic enzymes only active in one of the compartments? Other possible applications could be the analysis of entire metabolic pathways and enzyme super complexes formation that form in an analogous manner to the mitochondrial super-complexes (Robinson and Srere, 1985). Using yeast SP metabolism as an example, one could imagine that SPT could be recruited to the nuclear ER while the subsequent enzyme Tsc10 is recruited to the peripheral ER. This would allow the determination of the need for substrate handover. These questions could be addressed using a similar approach as presented here.

The approach used also has its limitations that should be addressed in the future. The recruitment of the entire SPT population to one sub-compartment changes the overall enzyme amount at this location. This by itself could lead to changes in the activity. It also remains possible that oligomerization of the SPT complexes regulates their activity in the different ER sub-compartments (Han et al., 2019; Hornemann et al., 2007; Li et al., 2021; Wang et al., 2021), and this might be susceptible to local changes in enzyme abundance. In addition, the non-inducible recruitment could allow the cells to adapt to the changing conditions and therefore modulate SPT activity by homeostatic regulations. Thus, an inducible recruitment system would be even more preferable.

Yeast strains used in this study are shown in Table S1. All deletions and tagging of all proteins were performed as described in (Janke et al., 2004). All plasmids used in this study are shown in Table S2. All oligonucleotides used in this study are shown in Table S3. Sequences were cloned into plasmid vectors via fast cloning (Li et al., 2011) and the ALFA tag, SPOT tag, CAAX box and 103aa linker were inserted using Q5 mutagenesis (Gatta et al., 2015; Götzke et al., 2019; Tang et al., 2009). All experiments were performed in normal YPD medium [1% yeast extract (212730; Gibco), 2% peptone (211830; Gibco), 2% glucose (HN06.4; Roth)]). For microscopy experiments SDC −lysine medium was used [2% glucose, 6.75 g/l yeast nitrogen base without amino acids (291929; BD Difco), 1.92 g/l yeast synthetic drop-out media supplements without lysine (Y1896; Sigma-Aldrich)] supplemented with 30 mg/ml lysine. Sporulation plates were made with 1% potassium acetate and 3% agar.

To conduct tetrad analyses, diploid yeast cells were collected by centrifugation (16,873 g for 1 min) and placed onto 1% potassium acetate agar for sporulation at 30°C. After 3–5 days and microscopic inspection for ascus formation, a sample of each culture was suspended in 100 µl of sterile water. 5 µl of Zymolyase 20 T (10 mg/ml; MP Biomedicals, Eschwege, Germany) was added, and incubated at room temperature for 9 min. A small number of cells was streaked out on YPD plates and spores were segregated using a Singer MSM400 micromanipulator (Singer Instruments, UK). The plates were then incubated for 3 days at 30°C.

For spotting assays, cells from an overnight preculture were inoculated and grown to exponential growth phase in YPD. They were serial diluted and spotted to the YPD plates with and without addition of the indicated concentrations of myriocin (Sigma-Aldrich). Plates were incubated for 2 days at 30°C.

For fluorescence microscopy experiments, cells were inoculated from an overnight preculture and grown to exponential growth phase. For the experiments with diploid cells (Fig. 3B), cells were imaged with equipment as described in Eising et al. (2022). All other microscopy experiments were performed using an Axioscope 5 FL (Zeiss) microscope. It was equipped with an Axiocam 702 mono camera Plan-Apochromat 100× [1.4 numerical aperture (NA)] and an oil immersion objective using the ZEN 3.1 pro software. ImageJ was used for picture processing. Pictures were taken at same settings and were processed in the same way unless otherwise mentioned.

For proteomic analysis, cells were inoculated from an overnight preculture and grown in YPD at 30°C until they reached exponential growth phase in triplicates. Two OD₆₀₀ units of all cultures were taken and pelleted at 2272 g for 2 min. Cell pellets were further treated according to the ‘iST Sample Preparation Kit (Pelleted cells & precipitated protein)’ protocol with the iST Sample Preparation Kit (Preomics) for cells lysis and protein digestion. Dried peptides were resuspended in 50 µl LC-Load and 3 µl were loaded for LC-MS/MS measurement on a Thermo Ultimate 3000 RSLC nano system connected to a Q ExactivePlus mass spectrometer (Thermo Fisher Scientific) as described previously (Limar et al., 2023). Briefly, the resulting peptides were transferred to a glass vial and 3 µl was used to perform reversed-phase chromatography on a Thermo Ultimate 3000 RSLC nano system connected to a QExactivePLUS mass spectrometer (Thermo Fisher Scientific) through a nano-electrospray ion source. Peptides were separated on a PepMap RSLC C18 easy spray column (2 µm, 100 Å, 75 µm×50 cm, Thermo Fisher Scientific) with an inner diameter of 75 µm. The column temperature was kept at 40°C. The peptides were eluted from the column via a linear gradient of acetonitrile from 12–35% in 0.1% formic acid for 80 min at a constant flow rate of 200 nl/min followed by a 20 min increase to 60% and finally 10 min to reach 90% buffer B. Eluted peptides from the column were directly electro sprayed into the mass spectrometer. Mass spectra were acquired on the Q ExactivePlus in a data-dependent mode to automatically switch between full scan MS and up to ten data-dependent MS/MS scans. The maximum injection time for full scans was 50 ms, with a target value of 3,000,000 at a resolution of 70,000 at m/z 200. The ten most intense multiply charged ions (z≥2) from the survey scan were selected with an isolation width of 1.6 Th and fragment with higher energy collision dissociation with normalized collision energies of 27. Target values for MS/MS were set at 100,000 with a maximum injection time of 80 ms at a resolution of 17,500 at m/z 200. To avoid repetitive sequencing, the dynamic exclusion of sequenced peptides was set at 20 s. Resulting data were analyzed with MaxQuant (v2.1.4.0, www.maxquant.org) (Cox and Mann, 2008; Cox et al., 2011) and Perseus (v2.0.7.0, www.maxquant.org/perseus) (Tyanova et al., 2016). All proteomics data are provided in Table S4.

For pulldown experiments, cells were inoculated from an overnight preculture in 100 ml YPD in triplicates and grown to exponential growth phase at 30°C. The same amounts of cells were harvested from all cultures at 2272 g 4°C for 5 min and snap frozen as cell pellets in Eppendorf tubes. Cells were lysed in with glass beads in 500 µl SPOT PD buffer (20 mM HEPES pH 7.4, 150 mM KOAc, 5% Glycerol, 1% GDN, Roche Complete Protease Inhibitor Cocktail EDTA free) using a FastPrep machine (MP Biomedicals). Supernatant was cleared at 20,817 g for 10 min and incubated for 30 min rotating at 4°C together with 25 µl pre-equilibrated Spot-Trap beads (Chromotek). Beads were washed four times with SPOT PD buffer at 2500 g for 2 min at 4°C. Afterwards, they were washed two times with Wash buffer (20 mM HEPES pH 7.4, 150 mM KOAc and 5% glycerol) at 2500 g for 2 min at 4°C. Beads were further treated following the ‘iST Sample Preparation Kit (Agarose Immunoprecipitation Samples)’ protocol with the iST Sample Preparation Kit (Preomics) for protein digestion. Dried peptides were resuspended in 10 µl LC-Load and 5 µl were loaded for LC-MS/MS analyzes using the same settings and evaluation methods as described above. Resulting data were analyzed with MaxQuant (v2.0.3.0, www.maxquant.org) (Cox and Mann, 2008; Cox et al., 2011) and Perseus (v2.0.7.0, www.maxquant.org/perseus) (Tyanova et al., 2016). All proteomics data are provided in Table S4.

Cultures were inoculated from a logarithmic growing preculture in 50 ml YPD and grown to exponential growth phase at 30°C. Cells were collected at 2272 g 5 min 4°C and snap frozen in liquid nitrogen. Cells were lysed with glass beads in 500 µl 155 mM ammonium formate using a FastPrep machine (MP Biomedicals). Lipids were extracted corresponding to 400 µg protein by a two-step extraction as described previously (Ejsing et al., 2009). Internal standard (PG 17:0-14:1, PS 17:0-14:1, PE 17:0-20:4, PC 17:0-20:4, PI 17:0-20:4, LPE 17:1; Avanti Polar Lipids) was added before lipid extraction. First, lipids were extracted using 15:1 chloroform/methanol, which were later analyzed via LC-MS/MS using positive ion mode. The remaining hydrophilic phase was re-extracted using 2:1 chloroform/methanol, which were later analyzed via LC-MS/MS using negative ion mode. Dried lipids were dissolved in 50 µl 65:35 Buffer A (50:50 acetonitrile/H₂O, 10 mM ammonium formate and 0.1% formic acid)/Buffer B (88:10:2 2-propanol/acetonitrile/H₂O, 2 mM ammonium formate and 0.02% formic acid). A C18 reverse-phase column (Thermo Accucore RP-MS, C18, 2.1×150 mm, 2.6 µm; Thermo Fisher Scientific) was used with a Shimadzu Nexera HPLC system with a heated electrospray ionization (HESI) and a Exactive Plus Orbitrap mass spectrometer as described previously (Esch et al., 2020). The elution was performed with a 20-min gradient. At 0 to 1 min, elution starts with 30% solvent B and increases to 100% over 12 mins in a linear gradient. For 3 min 100% solvent B is maintained. Afterwards, solvent B was decreased to 30%. For 4 min, 30% solvent B is maintained for column re-equilibration. The flowrate was set to 0.3 ml/min. MS spectra of lipids were acquired in full-scan/data-dependent MS2 mode. The maximum injection time for full scans was 100 ms, with a target value of 3,000,000 at a resolution of 70,000 at m/z 200 with a mass range of 200–2000 m/z in both, positive and negative ion mode. The ten most intense ions from the survey scan were selected and fragmented with HCD with a normalized collision energy of 27. Target values for MS/MS were set at 100,000 with a maximum injection time of 50 ms at a resolution of 17,500 at m/z 200. To avoid repetitive sequencing, the dynamic exclusion of sequenced lipids was set at 10 s. Resulting spectra were analyzed using LipidSearch 5.0 (Thermo Fisher Scientific). Lipid species were identified by database (>1,500,000 entries) search of positive (+H⁺; +NH₄⁺) or negative (−H⁻; +HCOO⁻) adducts. Sample alignment was conducted with a retention time window of 0.5 min. Lipid standards were used for the calculation of lipid concentrations [PG 17:0-14:1, PS 17:0-14:1, PE 17:0-20:4, PC 17:0-20:4, PI 17:0-20:4, PA 15:0-18:1, LPS 17:1, LPE 17:1, LPC 17:1, LPI 17:1, TAG 15:0-18:1-15:0, DG 15:0-18:1 (Avanti Polar Lipids)]. A LPE (negative) or Cer (positive) standard was used for the normalization between samples. Values are depicted as fold change from control strain. All lipidomic raw data are provided in Table S5.

The flux analysis process was adapted from Esch et al. (2020) and Martínez-Montañés et al. (2020). Cultures were inoculated from a logarithmic growing preculture in 20 ml YPD and grown to exponential growth phase at 30°C until they reached an OD₆₀₀ of 0.8. Into 10 ml of the cultures [¹³C₃¹⁵N₁]serine (CCN3000P1; CortecNet) to a final concentration of 3.8 mM was added (t=0). Samples (2.5 OD units each) were collected after 5, 15 and 30 min (t=5, 15, 30) at 2272 g 2 min 4°C and cell pellets were directly snap frozen in liquid nitrogen. Lipids were extracted and analyzed as described in the targeted LCB and ceramide analysis paragraph.

Cultures were inoculated from a logarithmic growing preculture in 15 ml YPD and grown to exponential growth phase at 23°C until they reached an OD₆₀₀ of 0.8. The cell culture was then split; 2.5 OD units of the culture was incubated at 23°C for 5 min (no heat shock), and another 2.5 OD units of the culture was incubated at 39°C for 5 min (heat shock) in a water bath. If mentioned, [¹³C₃¹⁵N₁]serine (CCN3000P1; CortecNet) to a final concentration of 3.8 mM was added (t=0) before incubation. After heat shock, cells were harvested at 2272 g for 2 min, supernatant was poured out and cells were directly snap frozen in liquid nitrogen. Lipids were extracted and analyzed as described in the targeted LCB and ceramide analysis paragraph.

Cells were thawed on ice and washed with ice-cold 155 mM ammonium formate. Cell pellets were spiked with internal standard (Sphingosine d17:1, Ceramide d17:1/24:0; Avanti Polar Lipids) and lipid extraction with 2:1 chloroform/methanol was performed as described previously (Ejsing et al., 2009; Esch et al., 2020). Dried lipids were dissolved in 65:35 Buffer A (50:50 acetonitrile/H₂O, 10 mM ammonium formate and 0.1% formic acid)/Buffer B (88:10:2 2-propanol/acetonitrile/H₂O, 2 mM ammonium formate and 0.02% formic acid). An external standard curve was prepared using dihydrosphingosine (DHS; Avanti Polar Lipids) 18:0, phytosphingosine (PHS; Avanti Polar Lipids) 18:0 and ceramide t18:0/24:0 (Avanti Polar Lipids/Cayman). Samples were analyzed on a QTRAP 5500 LC-MS/MS (SCIEX) mass spectrometer connected to a Shimadzu Nexera HPLC system and an Accucore C30 LC column (150 mm×2.1 mm 2.6 µm Solid Core; Thermo Fisher Scientific) in positive mode. For the gradient, 40% B for 0.1 min was used gollowed by its increase from 40% to 50% over 1.4 min. Afterwards, buffer B was increased from 50% to 100% over 1.5 min. 100% B was used for 1 min and decreased to 40% B for 0.1 min. Then, 40% B was kept until the end of the gradient. A constant flow rate of 0.4 ml/min was used with a total analysis time of 6 min and an injection volume of 2 μl. The MS data were measured in positive ion mode, scheduled MRM mode without detection windows (Table S6). For evaluation the SciexOS software was used. The internal standard was used for normalization. Measured OD₆₀₀ units were used for correction of the used cell number. All lipidomic raw data are provided in Table S5.

Cultures were inoculated from a preculture in 20 ml YPD and grown to exponential growth phase at 23°C until they reached an OD₆₀₀ of around 1. Cell cultures were split into 10 ml each. On half was incubated at 23°C for 5 min (no heat shock), another half of the culture were incubated at 39°C for 5 min (heat shock) in a water bath. After heat shock, cells were harvested at 4000 rpm for 2 min, supernatant was poured out and cells were directly frozen in liquid nitrogen. Cells were lysed with glass beads in 250 µl RIPA buffer (25 mM Tris/HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, Roche complete protease inhibitor cocktail; Roche PhosStop tablet) using a FastPrep machines (MP Biomedicals). Supernatant was cleared at 1681 g for 5 min. Protein concentration was determined and similar amount of protein was heated at 60°C for 5 min in Laemmli buffer with DTT. Phos-tag blots were performed as described in Schmidt et al. (2019). FLAG-tagged proteins were detected with a mouse anti-FLAG (F1804; Sigma-Aldrich) antibody diluted 1:1000. Pgk1 was detected with a 1:20,000 diluted mouse anti-Pgk1 (RRID:AB_2532235, Thermo Fisher Scientific) antibody. FLAG and Pgk1 antibodies were detected using a DyLight 800 coupled anti-mouse IgG secondary antibody (SA535521; Invitrogen). ALFA-tagged proteins were detected using a 1:1000 diluted rabbit anti-ALFA antibody (N1505, Nanotag) using a DyLight 800-coupled mouse anti-rabbit IgG secondary antibody (SA535571; Invitrogen). For phos-tag blots anti-FLAG (RRID:AB_259529, Sigma) was used and detected using an HRP coupled anti-mouse IgG secondary antibody (RRID:AB_258167, Sigma) and a Vilber Fusion FX7 Imager. All raw data of western blots are provided in Fig. S5.

For statistical analysis a two-sided unpaired t-test or a one-way ANOVA with Tukey's multiple-comparison test were used (*P<0.05). For proteomic experiments statistics were performed as described previously (Cox and Mann, 2008; Cox et al., 2011; Tyanova et al., 2016).

We thank members of the Fröhlich lab for valuable discussions. We also thank Jacob Piehler (Osnabrück) for providing the nanobody expression plasmids.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261353.reviewer-comments.pdf