Identification of the client-binding site on the Golgi membrane protein adaptor Vps74/yGOLPH3

Subject areas: Biochemistry, molecular biology, cell biology

Vps74 localizes to multiple Golgi cisternae

Vps74 and its homologs (the GOLPH3) are widely distributed throughout eukaryotes. These proteins are peripheral Golgi membrane proteins that reportedly associate with membranes through their PI4P-binding site and/or through a conserved β hairpin.^{20 ,21} GOLPH3s function as COPI-coatomer adaptors that bind to the cytoplasmically exposed N-termini of numerous Golgi-resident integral membrane proteins.^{8 ,9 ,10} Hereafter we refer to these integral membrane proteins as clients. In addition to its client-binding ability, Vps74 reportedly stimulates the PI4P phosphatase activity of Sac1.^{21 ,22} Presumably, where Vps74 and Sac1 distributions overlap the prevalence of PI4P is likely to be reduced.²¹ Given that PI4P is found predominantly in the trans Golgi,^{23 ,24 ,25} but Vps74’s clients are distributed throughout the Golgi, how does Vps74 bind to Golgi membranes with comparatively little PI4P?

To begin to address this question, we generated Vps74 fusions to mNeon to visualize the protein in cells. Although others have shown that N-terminally GFP-tagged Vps74 localizes to Golgi-like puncta, a more robust test of the functional consequences of fusing a fluorescent protein to Vps74 is needed.^{8 ,11} For this purpose, we generated three mNeon-tagged forms of Vps74 (Figure 1A) and used a yeast strain lacking VPS74 and TED1 balanced with a counter-selectable copy of TED1 on a plasmid (pTED1(URA3) to assess whether they were functional. TED1 encodes a remodelase that removes phosphoethanolamine from mannose 2 of glycosylphosphatidylinositol-anchored proteins (GPI-AP) in the ER, an essential function that is shared by Dcr2, which acts in the Golgi.²⁶ Dcr2 is a Vps74 client, and thus cells lacking VPS74 and TED1 are not viable because Dcr2 is mislocalized and degraded in the vacuole.²⁶ Therefore, whether a particular variant can support the growth of cells lacking VPS74 and TED1, reflects the variant’s impact on the Golgi retention of Dcr2.²⁶ All three mNeon/Vps74 fusion proteins were functional, as assessed by their ability to complement the synthetic lethal growth phenotype of cells lacking the VPS74 and TED1 genes²⁶ (Figure 1B). Similarly, vps74Δ ted1Δ cells expressing the mNeon/Vps74 fusion proteins grew like control cells at 37°C (Figure 1C). Although the expression levels of Vps74, Vps74-mNeon, and mNeon-Vps74 were similar (Figure 1D), none of the Vps74/mNeon constructs were as effective as WT Vps74 in maintaining the steady-state levels of Dcr2 (Figure 1E). However, in the case of Mnn5 and Kre2, mNeon-Vps74 was more effective in the retention of these clients than Vps74-mNeon (Figure 1F).^{9 ,27} Interestingly, the tandem chimera (Vps74-mNeon-Vps74) was no more effective than its monomeric counterparts in the retention of Dcr2, Mnn5, and Kre2 (Figures 1E and 1F) suggesting that tethering two molecules of Vps74 did not increase the effectiveness of the adaptor.

To establish the extent to which the three tagged forms of Vps74 could bind to different Golgi cisternae, we integrated the various mNeon-tagged variants at the VPS74 locus in cells expressing mCherry-Sed5 (a cis Golgi resident), Sec21-mScarlet (a cis, medial, and trans Golgi resident) or Sec7-mScarlet (a trans Golgi resident) (Figure 2A). Vps74-mNeon showed the fewest puncta ∼5 puncta per 20 cells, whereas mNeon-Vps74 and Vps74-mNeon-Vps74 showed roughly the same number of puncta per cell (∼35–40 per 20 cells) (Figure 2B). Despite the reduction in total puncta, the distribution of Vps74-mNeon across Golgi cisternae, as judged by colocalization with Sed5, Sec21, and Sec7, was like that seen with mNeon-Vps74 and Vps74-mNeon-Vps74 (Figure 2C). These fluorescence data indicate that Vps74 localizes to cis, medial, and trans cisternae. The relative decrease in Vps74-mNeon Golgi puncta is in accord with our findings that cells expressing this version of the adaptor show a reduction in the steady-state levels of Dcr2, Mnn5, and Kre2 (Figures 1E and 1F). We conclude that Vps74-mNeon is functional but hypomorphic due to the protein’s reduced capacity to bind to Golgi membranes. Importantly, the recruitment of Vps74 to multiple cisternae correlates well with the distribution of the adaptor’s clients in cells.^{27 ,28}

The co-localization data presented in Figures 2A–2C showed that the three Vps74 mNeon fusion proteins colocalize with ∼40%–45% of Sed5 (cis Golgi) positive puncta, with ∼30%–35% of Sec7 (trans Golgi) positive puncta and with ∼75%–90% of Sec21 (cis, medial, and trans). These findings reveal that irrespective of the positioning of mNeon on Vps74 all three variants display similar Golgi distributions. Apparently, Vps74 can bind to membranes with lower steady-state levels of PI4P than found in the trans Golgi, implying that an additional/alternative mechanism may account for the protein’s localization to cis and medial cisternae.^{20 ,21}

Client over-expression is sufficient to recruit Vps74 to Golgi membranes

To explore the prospect of an additional membrane-binding mechanism, we asked if Vps74 could be recruited to the Golgi directly by binding to its clients. To address this, we introduced high copy number plasmids containing various client genes into cells expressing Vps74-mNeon as their sole source of the adaptor. We chose Vps74-mNeon for these experiments as the fusion protein, while being functional (Figure 1), was found on fewer Golgi puncta, making any increase more readily apparent. We selected a subset of Vps74 clients to test as well as the non-Vps74 client OCH1.²⁷ The data presented in Figure 3A reveal that when transformed with high copy number plasmids containing the client genes MNN5, KRE2,^{9 ,27} and GRX7, cells had a larger number of Vps74-mNeon puncta than those containing empty vector or the non-Vps74 client OCH1 plasmid. On average, the presence of high copy number plasmids expressing client genes increased puncta to ∼20 per 20 cells (or 4-fold) compared to the non-client containing plasmid or vector only which showed ∼5 puncta per 20 cells (Figure 3A). The steady-state levels of Grx7, Mnn5, Kre2, and Och1 were indeed increased (∼5- to 10-fold) relative to wildtype cells as judged by immunoblots of whole cell extracts from cells transformed with the high copy number plasmids in which clients (excluding Kre2) were expressed as C-terminal fusion proteins to 9 copies of the myc epitope tag (Figure S1A). The addition of an epitope tag to these clients did not affect the capacity of over-expressed clients to increase the number of Vps74-mNeon puncta (Figure S1B). By contrast, overexpression of clients in cells expressing mNeon-Vps74 did not show any statistically significant increase in the number of puncta observed (Figure S1C). Nor did client overexpression alter the steady-state levels of Vps74 (Figure 3B).

We recovered GRX7 as a dosage suppressor of the temperature-sensitivity of vps74-1 ted1Δ cells (Figure 3C), and like other clients, in the absence of VPS74 the steady-state levels of Grx7 were reduced (∼20% of endogenous levels) (Figure 3D). Grx7 is a type II integral membrane protein with a short cytoplasmic N-terminal region and reportedly functions as a Golgi-localized monothiol glutaredoxin.²⁹

Possible explanations for the observed increase in puncta in cells overexpressing Grx7, Mnn5, and Kre2 include alterations to the steady-state distributions of these proteins, whereby they are found in more distal and PI4P-enriched cisternae. Alternatively, Vps74 can be recruited to the Golgi by binding directly to the N-termini of its clients. To address these scenarios, we introduced the high copy number GRX7, MNN5, and KRE2 containing plasmids into cells expressing Vps74-mNeon, and either mCherry-Sed5, Sec21-mScarlet, or Sec7-mScarlet, and quantified the number of co-localizing puncta (Figure 4A). These experiments revealed that over-expression of individual clients did not alter the overall percentage of Vps74-mNeon puncta in cisternae fluorescently labeled with Sed5, Sec21, or Sec7 when compared to cells transformed with empty vector only (Figures 4A and 4B). To conclude, clients can play a direct role in the recruitment of Vps74 to Golgi cisternae.

Client over-expression is not sufficient to recruit the PI4P-binding deficient and β hairpin variants of Vps74 to the Golgi

Previous studies identified a role for PI4P and the β hairpin in the recruitment of Vps74 to Golgi membranes.²⁰ Considering the findings presented in Figures 3 and 4, we re-examined the role of these features in the recruitment of Vps74 to the Golgi by generating amino acid substitutions to the PI4P binding site at W88, R97 and K178, R181, and deleting the amino acids comprising the β hairpin (amino acids 197–208 (Δ197-208)).²⁰ The functional consequences of these variants were assessed in cells lacking VPS74 and TED1 (Figure 5A). Only the R97A variant of Vps74 was incapable of supporting the growth of vps74Δ ted1Δ cells (Figure 5A). Surprisingly, substitutions at three of the four amino acids previously reported to be important for Vps74 binding to PI4P (W88, K178, and R181) did not result in loss of function of the protein, nor were vps74Δ ted1Δ cells expressing these variants (W88A and K178A R181A) temperature-sensitive for growth (Figure 5B). Although vps74Δ ted1Δ cells expressing the β hairpin deletion variant Vps74 (Δ197-208) were viable at 25°C, these cells were temperature-sensitive for growth at 37°C (Figure 5B). The β hairpin deletion variant also conferred calcofluor white sensitivity to cells (a proxy for glycosylation defects),³⁰ whereas vps74Δ ted1Δ cells expressing the W88A and K178A R181A amino acid substitution variants of Vps74 were no more sensitive to calcofluor white, than cells expressing the wildtype protein (Figure 5B). It appears that the β hairpin does not play a crucial role in the Golgi membrane recruitment of Vps74 when cells are grown at 25°C.

An important caveat in the interpretation of these findings is the genetic background (vps74Δ ted1Δ) used to assess the function of the various amino acid variants. Cells lacking VPS74 and TED1 are not viable because Dcr2 (a Vps74 client) is mislocalized and degraded in the vacuole.²⁶ In line with this reasoning, the steady-state levels of Dcr2 were significantly reduced in the R97A, K178A R181A and β hairpin (Δ197-208) variants, comparable to those observed in cells lacking VPS74 (Figure 5C). We note that the steady-state levels of the Vps74 variants are increased when TED1 is deleted (Figure S2A). However, the increased levels of the K178A, R181A, and β hairpin variant proteins only resulted in a modest increase in the steady-state levels of Dcr2 (Figure S2A). Presumably, the essential activity of Dcr2 can be provided by ∼ 30% of endogenous levels of the enzyme (Figure S2A). Nevertheless, these findings were recapitulated in in vitro mixing experiments using the Mnn4 N-terminus (Figure 5D) and revealed a direct role for the β hairpin and R97 (in the PI4P binding site) in client-binding. Thus, in addition to the reported importance of the β hairpin in Golgi membrane-binding²¹ it also appears to play a key role in Vps74 client-binding, at least in the case of Mnn4.

To further interrogate the impact of the W88, R97, K178, R181, and β hairpin variants on Vps74 membrane binding, we asked whether over-expression of Kre2 (Figures 3A and S2B) was sufficient to recruit these adaptor variants to Golgi membranes (Figure 5E). However, puncta were only apparent in cells expressing wildtype VPS74, and to a lesser extent in cells expressing the W88A substitution variant (Figure 5E). These findings were recapitulated when Grx7-9myc was overexpressed, except for the W88A variant, which had a greater effect on Grx7 retention than on Kre2 retention (Figures 5E, S2C, and S2D).

To explore the extent to which the Vps74 variants affected the retention of clients in the Golgi, we examined the steady-state levels of Kre2, Mnn4, Mnn2, Grx7, and Mnn5 in variant-expressing cells (Figures 5C and 5F). These experiments revealed that substitutions to W88 and K178, R181 had a comparatively minimal impact on the steady-state levels of clients, whereas the R97 and the β hairpin variants had the largest effect on client retention (Figures 5C and 5F). Although amino acid substitutions at R97, W88, and K178, R181 reportedly resulted in mislocalization of Kre2 to the vacuole,²⁰ these variants of Vps74 (with the exception of the R97 variant, which was not tested) were not sensitive to calcofluor white (Figure 5B) unlike cells lacking the KRE2,⁹ nor did cells expressing the W88 and K187, R181 variants show a precipitous reduction in the steady-state levels of Kre2 (Figures 5C and 5F).

Our results on substitutions to R97, W88, and K178, R181 are difficult to reconcile with a critical role for PI4P recruiting Vps74 to Golgi membranes. Except for Grx7, the W88 variant had a comparatively modest effect on Vps74 client retention (Figures 5C and 5F). Similarly, except for Dcr2 and Grx7, the K178A, R181A variant had a comparatively modest effect on all other clients tested (Figures 5C and 5F). However, as a structure of Vps74 bound to PI4P is not available, we cannot exclude the possibility that other amino acids in addition to R97 might also be critical for Vps74-PI4P binding in cells. Nevertheless, R97 and the β hairpin appear to play a role in adaptor-client-binding, as judged by in vitro binding studies with the N-terminus of Mnn4 (Figure 5D).

PI4P binding may be a co-requisite for recruitment of Vps74 to Golgi membranes

To further assess whether recruitment of Vps74 to PI4P-enriched Golgi cisternae was critical for client retention, we generated a dimer of Vps74 in which the PH domain of FAPP1, which has been shown to bind to PI4P in yeast cells and to co-localize with the trans Golgi protein Sec7,^{31 ,32} was inserted between the two monomers (Figure 6A). We opted for tethering two molecules of Vps74 as a covalent dimer is functional (Figures 1B and 1C). The W88A, R97A, and K178A, R181A substitutions and β hairpin deletion variants were introduced into both Vps74 monomers (Figure 6A) and were expressed from the VPS74 locus. In cells expressing Vps74-mNeon-Vps74 and the vps74-mNeon-vps74 variants, only WT Vps74-mNeon-Vps74 and the W88A vps74-mNeon-vps74 variant showed any puncta (Figure 6B), in agreement with our observations with Vps74-mNeon and these variants (Figure 5E). By contrast, WT Vps74-mNeon-PH-Vps74 and all the vps74-mNeon-PH-vps74 variants showed puncta, consistent with the ability of the FAPP1 PH domain to bind PI4P in yeast cells (Figure 6B).³¹ However, when we examined the steady-state levels of two clients (Mnn5 and Kre2) in cells expressing the Vps74-mNeon-Vps74 or Vps74-mNeon-PH-Vps74 variants R97A, K178A, R181A, and Δ197-208 (the β hairpin deletion), the Vps74-mNeon-PH-Vps74 variants were no better than the Vps74-mNeon-Vps74 variants in restoring the steady-state levels of Mnn5 or Kre2 (Figure 6C). Interestingly, both the WT and W88 variant of Vps74-mNeon-PH-Vps74 were less effective in maintaining the steady-state levels of Mnn5 than their Vps74-mNeon-Vps74 counterparts, suggesting that when the adaptor was localized to more trans cisternae PI4P enriched cisternae, it was less efficient retaining this client in the Golgi (Figure 6C). Similar findings were also obtained for Mnn4 and Mnn2 (Figure S3).

To obtain more insight into the role of PI4P in Vps74 membrane recruitment we examined the fate of mNeon-Vps74 membrane association in cells in which Pik1 (the sole yeast Golgi PI4P kinase³³) was subject to auxin-inducible proteosomal degradation.³⁴ We generated a yeast strain in which the mAID2 degron³⁴ coding region appended with two HA tags was fused in frame to the N-terminal coding sequence of the PIK1 gene at the PIK1 locus and OsTIR1(F74G) was integrated at the GAL1 locus. This strain (in which mNeon-Vps74 was expressed at the VPS74 locus, and mCherry-SED5 was expressed from the SED5 locus) was then transformed with a high copy number plasmid with or without KRE2 (KRE2 2μ). We chose mNeon-Vps74 in this instance because we wished to observe a decrease in puncta overtime rather than an increase. When these strains were grown in media containing galactose and in the presence of the TIR1(F74G) ligand 5-Ph-IAA, mAID-HA-Pik1 is targeted for degradation by the proteosome.³⁴ These strains allowed us to monitor the steady-state levels of Pik1 and mNeon-Vps74 over the time course of experiments (by immunoblotting) as well as visualize mCherry-Sed5 (as a proxy for Golgi morphology) and visualize and quantify the number of mNeon-Vps74 puncta present (as a proxy for PI4P levels). We examined 3 time points under galactose induction conditions (0, 90, and 120 min). Representative fluorescence microscopy images are given in Figure S4A. Quantification and bar graphs of Vps74 puncta number, protein levels of mAID-HA-Pik1 and mNeon-Vps74 and a representative immunoblot (from 1 of the three replicates) are presented in the Figures S4B and S4C, respectively, while three graphs summarizing data from these experiments are presented in Figure 6D.

The steady-state levels of mNeon-Vps74 were not statistically different across the duration of the experimental time course. In contrast, the steady-state levels of mAID-HA-Pik1 declined over the 120 min duration of 5-Ph-IAA treatment reaching ∼50% of time 0 levels at 90 min and ∼30% of time 0 levels at 120 min. The rate of decline in the strains carrying the empty vector (empty vector + 1 mM 5-Ph-IAA) or KRE2 (KRE2 2μ + 1 mM 5-Ph-IAA) were indistinguishable. This system is a little leaky as evidenced by the modest decline in steady-state levels of mAID-HA-Pik1 in controls (Figures 6D, S4B, and S4C).

At t = 0 min the number of mNeon-Vps74 puncta present in the empty vector + 1 mM 5-Ph-IAA and KRE2 2μ + 1 mM 5-Ph-IAA strains was not significantly different. However, over the course of the first 90 min the rate of decline in puncta number was greater for empty vector + 1 mM 5-Ph-IAA as indicated by the slope. In addition, the absolute number of puncta present in empty vector + 1 mM 5-Ph-IAA cells (∼5/per 20 cells) was significantly lower than in KRE2 2μ + 1 mM 5-Ph-IAA cells (∼20/per 20 cells). However, between 90 and 120 min the rate of loss of puncta is dramatically increased in KRE2 2μ + 1 mM 5-Ph-IAA cells, as was the absolute number of puncta—which was now not statistically significantly different from those seen in empty vector + 1 mM 5-Ph-IAA cells. These data confirm an important role of PI4P in the recruitment of Vps74 to Golgi membranes but are also consistent with a role for client N-termini in Vps74 membrane binding in agreement with the data presented in Figures 3A, 4A, 4B, S2B, and S2C.

To conclude, our findings are consistent with the role for clients and PI4P in the association of Vps74 with Golgi membranes, while also revealing that one or more of the amino acids previously identified as being required for PI4P binding (R97 and K178, R181) are in fact part of the client-binding site.

The client-binding site on Vps74 includes two evolutionarily conserved adjacent loops

The prospect that the amino acid residues important for PI4P binding and the β hairpin might constitute part of the client-binding site on Vps74, prompted us to search for evolutionarily conserved, surface exposed and membrane opposing regions on Vps74 and human GOLPH3s. This analysis identified two regions on Vps74/GOLPH3s defined by amino acids 78–90 and 165–177 (Vps74 numbering) (Figure 7A). These residues define two loops on Vps74 (hereafter referred to as Loop1 and Loop2) that flank the PI4P binding site and the β hairpin (Figures 7A and 7C). To identify those amino acids critical for client-binding, we used AlphaFold2. For these analyses we used the amino acid sequences of 9 client N-termini (Figure S5A) and evaluated possible interactions of these clients with Vps74 from the AlphaFold2 predictions. This in silico analysis identified W88 and D90 in Loop1 and E166, W168 and Q176 in Loop2 as the most frequent presumptive contact sites between clients and Vps74 (Figure 7B).

To examine the impact of substitutions in Loop1 and Loop2 on Vps74’s adaptor function, we generated three substitution variants that encompassed F87, W88, N89, D90 and E166, T167, W168 either singularly or in combination. Variant function was assessed in cells lacking VPS74 and TED1 and carrying a counter-selectable copy of the TED1 gene on a plasmid (Figures S5B and S5C). Of the three Vps74 amino acid substitution variants examined (⁸⁷AAAA⁹⁰, ¹⁶⁶AAA¹⁶⁸, ⁸⁷AAAA⁹⁰ + ¹⁶⁶AAA¹⁶⁸), cells expressing ¹⁶⁶AAA¹⁶⁸ or ⁸⁷AAAA⁹⁰ + ¹⁶⁶AAA¹⁶⁸ were unable to support the growth of vps74Δ ted1Δ cells (Figure S5B). Cells expressing ⁸⁷AAAA⁹⁰ grew indistinguishably from wildtype and were not temperature-sensitive for growth or sensitive to calcofluor white treatment (Figure S5C). We next expressed these three variants, as well as the W88A variant, in cells lacking VPS74, and examined the steady-state levels of Mnn5, Kre2, Mnn4, Mnn2, Dcr2, and Grx7 by immunoblotting (Figure 7D). Quantification of these data revealed that the combined substitutions to Loop1 and Loop2 (⁸⁷AAAA⁹⁰ + ¹⁶⁶AAA¹⁶⁸) had the most significant impact on the steady-state levels of all clients tested, while W88A was the least impactful (Figures 5C, 5F, 7D, and 7E). The identification of W88 as a presumptive client-binding residue is puzzling but likely reflects the relative robustness of our in-silico method. Based on the experiments conducted herein, W88 is unlikely to be imperative for PI4P- or client-binding.

In agreement with our genetic assessment, in cells expressing the loss of function Loop2 variant ¹⁶⁶AAA¹⁶⁸ (Figure S5B), the steady-state level of Dcr2 was substantially reduced (Figure 7D), whereas for the other clients examined the steady-state levels of Mnn4 and Mnn2 were affected to the greatest extent by these amino acid substitutions (Figures 7D and 7E). In contrast, the steady-state levels of Mnn5 were largely unaffected by amino acid substitutions to Loop2 whereas Kre2 retention was most significantly impacted by substitutions to Loop2 (Figures 7D and 7E). The effect of the combined Loop1+Loop2 substitutions appears to be additive, as cells expressing this variant showed a greater reduction in client steady-state levels than cells expressing either the Loop1 or Loop2 variants singularly (Figures 7D and 7E). The findings on Mnn4 steady-state levels in the various Vps74 variants could be replicated in an in vitro binding assay, using bacterially expressed Vps74 and the N-terminus of Mnn4 (Figure 7F).

In sum, these studies define the client-binding site on Vps74 and reveal that not all clients are similarly affected by amino acid substitutions at Loop1 and Loop2 (Figure 7E). The variable effect of Loop1 and Loop2 amino acid substitutions on client retention may explain in part how Vps74 (and other GOLPH3s) can accommodate client N-termini of differing amino acid composition and lengths (Figure S5A).^{7 ,9 ,10 ,27}

A nanobody against Vps74 corroborates identification of the client-binding site

In an orthogonal approach to our use of amino acid substitutions to define the client-binding site, we isolated a suite of Vps74-binding nanobodies (Table S4).³⁵ Plasmids expressing N-terminally ubiquitin-tagged nanobodies were introduced into yeast cells carrying a deletion of TED1 and a counter-selectable copy of TED1 linked to the URA3 gene (Figures S6A and S6B). Fusion to ubiquitin was necessary to increase the expression of nanobodies in the cytoplasm, however, endogenous cytoplasmic deubiquitinases cleaved ubiquitin, preserving the N-terminal sequence of the nanobody (Figure S6C). In this assay, cells expressing a nanobody that blocked an essential property on Vps74 would not be able to grow in ted1Δ cells as neutralization of Vps74 activity would phenocopy ted1Δ vps74Δ cells (Figure S6B). We chose one such inhibitory nanobody termed Vps74InNb#4 for further characterization (Figure 8A), as it also bound to human GOLPH3 and GOLPH3L, reasoning that it might bind to an evolutionarily conserved and therefore functionally significant region on the GOLPH3s (Figure 8B).

In agreement with its dominant negative effect in ted1Δ cells, when Vps74InNb#4 was expressed in wildtype cells, the steady-state levels of clients were significantly reduced as judged by immunoblots for Dcr2, Mnn5, and Kre2 (Figure 8C). This was not a result of changes in the steady-state levels of Vps74 (Figure 8C). The reduction of steady-state levels of Dcr2, Mnn5, and Kre2 was a direct result of Vps74InNb#4 binding to Vps74 (Figure S6C). Vps74InNb#4 also reduced the association of Mnn4 with Vps74 in an in vitro mixing assay (Figure 8D). Thus, Vps74InNb#4 likely binds to a membrane opposing surface on Vps74, a property that also blocked the binding of Vps74 to Golgi membranes (Figure 8E) and resulted in the mislocalization of clients to the vacuole (Figure S6D).

To validate that Vps74InNb#4 occupied the client-binding site, we docked the nanobody with Vps74 using AlphaFold3 (Figure 8F) and verified the interaction interface on Vps74 using site-directed variants in an in vitro mixing assay (Figures 8G and S6E). To conclude, Vps74InNb#4 bound to the β hairpin and to Loop1, but not to Loop2. Nevertheless, the interaction of Vps74InNb#4 with Vps74 was sufficient to prevent the adaptor from binding to Golgi membranes, and directly to Mnn4 in in vitro mixing experiments (Figures 8D and 8E). These data corroborate our interrogation of Loop1 and Loop2 by site directed mutagenesis (Figure 7).

The Arf1-GTP and Sac1 binding sites overlap with and exclude client-binding to Vps74

Vps74 has been shown to bind to the PI4P phosphatase Sac1, whereupon Vps74 reportedly stimulates the enzyme’s activity.^{21 ,22} Critically, the binding site on Vps74 for Sac1 encompasses Loop2²¹ coinciding with part of the client-binding site we identify in this study (Figure 7). We have previously shown that Vps74 binds to the small GTPase Arf1 in its GTP-bound state.¹⁵ Unlike the Vps74-Sac1 association, the location of the Arf1-GTP binding site on Vps74 and the functional significance of this interaction has remained elusive. Nonetheless, Vps74/GOLPH3 binds to the Golgi vesicle coat complex coatomer^{14 ,15} and Arf1-GTP is critical for COPI vesicle formation.³⁶ These findings are consistent with a role for Arf-GTP in the incorporation of Vps74 and its clients into COPI coated vesicles.

We took advantage of the availability of Vps74InNb#4 to investigate the relationship between Vps74, and its binding partners—Sac1 and Arf1-GTP. In in vitro mixing experiments Sac1 reduced Mnn4 binding to Vps74 by ∼ 85% (Figure 9A). These results concur with the data presented in Figure S7A, where we show that substitutions in Loop2, but not Loop1, reduce the binding of Sac1 to Vps74 in an in vitro mixing assay. These findings were corroborated with Vps74InNb#4, which preferentially binds to the β hairpin and Loop1 residues (Figures 8F, 8G, and 9B). Thus, when Vps74 is bound to Sac1, the adaptors capacity to bind its clients is reduced.

Interestingly, the binding site on Vps74 for Arf1-GTP and Sac1 also overlap. Sac1 binding to Vps74 resulted in a ∼80% reduction in the binding of Arf1-GTP (Figure 9C). Overlapping binding sites for Sac1 and Arf1-GTP were validated using site directed mutants of Vps74 in in vitro mixing experiments (Figures S7A and S7B). In contrast to Sac1, Vps74InNb#4 blocked Arf1-GTP binding to Vps74 (Figure 9D). This finding agrees with the AlphaFold3 predicted structure of Vps74InNb#4 bound to Vps74, which revealed that the nanobody binds to Loop1 and the β hairpin (Figures 8F and 8G), and with the in vitro mixing experiments where increasing amounts of the Mnn4 N-terminus reduced Arf1-GTP binding to Vps74 (Figure 9E).

To conclude, the data presented in Figures 9 and S7 indicate that the binding sites for clients, Sac1 and Arf1 overlap with one another (Figure 9F). While we have not measured the K_D for Sac1, Arf1-GTP or clients on Vps74, the in vitro mixing data we present here is consistent with Sac1 binding more robustly to Vps74 in cells than either Arf1-GTP or client N-termini. However, the extent to which Sac1 interferes with client and Arf1-GTP binding to Vps74 may be negligible under normal growth conditions, as Sac1 is predominantly an ER resident,^{31 ,32} whereas Vps74 is found on the Golgi membranes and in the cytoplasm—as is also the case of Arf1. Reportedly, Sac1 is found in the Golgi in cells deprived of glucose, rather than the ER.³¹ It is therefore possible that interactions between Sac1 and Vps74 are primarily restricted to changes in the cell’s growth parameters, such as when environmental glucose levels become limiting. The finding that Arf1-GTP and client-binding are exclusive of one another is difficult to reconcile with a role for Arf1-GFP in facilitating client selection by Vps74. Rather, Arf1-GTP may function to prevent client-free adaptors from being incorporated into nascent COPI vesicles upon hydrolysis of GTP.

Limitations of the study

Because of the low binding affinities, we could only perform in vitro binding studies with one client—Mnn4, an atypical Vps74 client with a much longer N-terminal sequence. Although mutations in the PI4P-binding site disrupted client binding in vivo and in vitro, we could not determine whether PI4P plays a role in the coincident detection of Vps74 clients.

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, David K. Banfield (bodkb@ust.hk).

Materials availability

All strains, plasmids and reagents used in this study are available from the lead contact upon request.

Data and code availability

• •

  Data reported in this paper will be shared by the lead contact upon request.

• •

  This paper does not report novel codes.

• •

  Any additional information required to reanalyze the data reported in this paper is available from the lead contact.

Key resources table

Experimental model and study participant details

All experiments were performed by using Saccharomyces cerevisiae BY4741 strains. Yeast cells were cultured at 25°C in YEPD or SD medium / plates for maintenance, unless specific temperature or medium was desired (e.g., 37°C , CFW medium and YEPGal medium).

Method details

Quantification and statistical analysis

• 1.Banfield D.K. Mechanisms of protein retention in the Golgi. Cold Spring Harb. Perspect. Biol. 2011;3 doi: 10.1101/cshperspect.a005264. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 2.Tu L., Banfield D.K. Localization of Golgi-resident glycosyltransferases. Cell. Mol. Life Sci. 2010;67:29–41. doi: 10.1007/s00018-009-0126-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 3.Glick B.S., Nakano A. Membrane traffic within the Golgi apparatus. Annu. Rev. Cell Dev. Biol. 2009;25:113–132. doi: 10.1146/annurev.cellbio.24.110707.175421. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 4.Glick B.S., Luini A. Models for Golgi traffic: a critical assessment. Cold Spring Harb. Perspect. Biol. 2011;3 doi: 10.1101/cshperspect.a005215. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 5.Pantazopoulou A., Glick B.S. A Kinetic View of Membrane Traffic Pathways Can Transcend the Classical View of Golgi Compartments. Front. Cell Dev. Biol. 2019;7:153. doi: 10.3389/fcell.2019.00153. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 6.Li J., Ahat E., Wang Y. Golgi Structure and Function in Health, Stress, and Diseases. Results Probl. Cell Differ. 2019;67:441–485. doi: 10.1007/978-3-030-23173-6_19. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 7.Rizzo R., Russo D., Kurokawa K., Sahu P., Lombardi B., Supino D., Zhukovsky M.A., Vocat A., Pothukuchi P., Kunnathully V., et al. Golgi maturation-dependent glycoenzyme recycling controls glycosphingolipid biosynthesis and cell growth via GOLPH3. EMBO J. 2021;40 doi: 10.15252/embj.2020107238. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 8.Schmitz K.R., Liu J., Li S., Setty T.G., Wood C.S., Burd C.G., Ferguson K.M. Golgi localization of glycosyltransferases requires a Vps74p oligomer. Dev. Cell. 2008;14:523–534. doi: 10.1016/j.devcel.2008年02月01日6. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 9.Tu L., Tai W.C.S., Chen L., Banfield D.K. Signal-mediated dynamic retention of glycosyltransferases in the Golgi. Science. 2008;321:404–407. doi: 10.1126/science.1159411. [DOI] [PubMed] [Google Scholar]
• 10.Welch L.G., Peak-Chew S.-Y., Begum F., Stevens T.J., Munro S. GOLPH3 and GOLPH3L are broad-spectrum COPI adaptors for sorting into intra-Golgi transport vesicles. J. Cell Biol. 2021;220 doi: 10.1083/jcb.202106115. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 11.Sardana R., Highland C.M., Straight B.E., Chavez C.F., Fromme J.C., Emr S.D. Golgi membrane protein Erd1 Is essential for recycling a subset of Golgi glycosyltransferases. eLife. 2021;10 doi: 10.7554/eLife.70774. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 12.Welch L.G., Muschalik N., Munro S. The FAM114A proteins are adaptors for the recycling of Golgi enzymes. J. Cell Sci. 2024;137:jcs262160. doi: 10.1242/jcs.262160. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 13.Brauer B.K., Chen Z., Beirow F., Li J., Meisinger D., Capriotti E., Schweizer M., Wagner L., Wienberg J., Hobohm L., et al. GOLPH3 and GOLPH3L maintain Golgi localization of LYSET and a functional mannose 6-phosphate transport pathway. EMBO J. 2024;43:6264–6290. doi: 10.1038/s44318-024-00305-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 14.Eckert E.S.P., Reckmann I., Hellwig A., Röhling S., El-Battari A., Wieland F.T., Popoff V. Golgi phosphoprotein 3 triggers signal-mediated incorporation of glycosyltransferases into coatomer-coated (COPI) vesicles. J. Biol. Chem. 2014;289:31319–31329. doi: 10.1074/jbc.M114.608182. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 15.Tu L., Chen L., Banfield D.K. A conserved N-terminal arginine-motif in GOLPH3-family proteins mediates binding to coatomer. Traffic. 2012;13:1496–1507. doi: 10.1111/j.1600-0854.2012.01403.x. [DOI] [PubMed] [Google Scholar]
• 16.Scott K.L., Kabbarah O., Liang M.-C., Ivanova E., Anagnostou V., Wu J., Dhakal S., Wu M., Chen S., Feinberg T., et al. GOLPH3 modulates mTOR signaling and rapamycin sensitivity in cancer. Nature. 2009;459:1085–1090. doi: 10.1038/nature08109. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 17.Rizzo R., Parashuraman S., D’Angelo G., Luini A. GOLPH3 and oncogenesis: What is the molecular link? Tissue Cell. 2017;49:170–174. doi: 10.1016/j.tice.201606008. [DOI] [PubMed] [Google Scholar]
• 18.Sechi S., Frappaolo A., Karimpour-Ghahnavieh A., Piergentili R., Giansanti M.G. Oncogenic Roles of GOLPH3 in the Physiopathology of Cancer. Int. J. Mol. Sci. 2020;21:933. doi: 10.3390/ijms21030933. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 19.Sechi S., Frappaolo A., Belloni G., Colotti G., Giansanti M.G. The multiple cellular functions of the oncoprotein Golgi phosphoprotein 3. Oncotarget. 2015;6:3493–3506. doi: 10.18632/oncotarget.3051. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 20.Wood C.S., Schmitz K.R., Bessman N.J., Setty T.G., Ferguson K.M., Burd C.G. PtdIns4P recognition by Vps74/GOLPH3 links PtdIns 4-kinase signaling to retrograde Golgi trafficking. J. Cell Biol. 2009;187:967–975. doi: 10.1083/jcb.200909063. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 21.Cai Y., Deng Y., Horenkamp F., Reinisch K.M., Burd C.G. Sac1-Vps74 structure reveals a mechanism to terminate phosphoinositide signaling in the Golgi apparatus. J. Cell Biol. 2014;206:485–491. doi: 10.1083/jcb.201404041. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 22.Wood C.S., Hung C.-S., Huoh Y.-S., Mousley C.J., Stefan C.J., Bankaitis V., Ferguson K.M., Burd C.G. Local control of phosphatidylinositol 4-phosphate signaling in the Golgi apparatus by Vps74 and Sac1 phosphoinositide phosphatase. Mol. Biol. Cell. 2012;23:2527–2536. doi: 10.1091/mbc.E12-01-0077. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 23.Audhya A., Foti M., Emr S.D. Distinct roles for the yeast phosphatidylinositol 4-kinases, Stt4p and Pik1p, in secretion, cell growth, and organelle membrane dynamics. Mol. Biol. Cell. 2000;11:2673–2689. doi: 10.1091/mbc.11.8.2673. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 24.Hama H., Schnieders E.A., Thorner J., Takemoto J.Y., DeWald D.B. Direct involvement of phosphatidylinositol 4-phosphate in secretion in the yeast Saccharomyces cerevisiae. J. Biol. Chem. 1999;274:34294–34300. doi: 10.1074/jbc.274.48.34294. [DOI] [PubMed] [Google Scholar]
• 25.Walch-Solimena C., Novick P. The yeast phosphatidylinositol-4-OH kinase pik1 regulates secretion at the Golgi. Nat. Cell Biol. 1999;1:523–525. doi: 10.1038/70319. [DOI] [PubMed] [Google Scholar]
• 26.Chen L., Tu L., Yang G., Banfield D.K. Remodeling-defective GPI-anchored proteins on the plasma membrane activate the spindle assembly checkpoint. Cell Rep. 2021;37 doi: 10.1016/j.celrep.2021.110120. [DOI] [PubMed] [Google Scholar]
• 27.Wang P., Ye Z., Banfield D.K. A novel mechanism for the retention of Golgi membrane proteins mediated by the Bre5p/Ubp3p deubiquitinase complex. Mol. Biol. Cell. 2020;31:2139–2155. doi: 10.1091/mbc.E20-03-0168. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 28.Tojima T., Suda Y., Jin N., Kurokawa K., Nakano A. Spatiotemporal dissection of the Golgi apparatus and the ER-Golgi intermediate compartment in budding yeast. eLife. 2024;13 doi: 10.7554/eLife.92900. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 29.Mesecke N., Spang A., Deponte M., Herrmann J.M. A Novel Group of Glutaredoxins in the cis-Golgi Critical for Oxidative Stress Resistance. Mol. Biol. Cell. 2008;19:2673–2680. doi: 10.1091/mbc.E07-09-0896. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 30.Ram A.F., Wolters A., Ten Hoopen R., Klis F.M. A new approach for isolating cell wall mutants in Saccharomyces cerevisiae by screening for hypersensitivity to calcofluor white. Yeast. 1994;10:1019–1030. doi: 10.1002/yea.320100804. [DOI] [PubMed] [Google Scholar]
• 31.Faulhammer F., Konrad G., Brankatschk B., Tahirovic S., Knödler A., Mayinger P. Cell growth-dependent coordination of lipid signaling and glycosylation is mediated by interactions between Sac1p and Dpm1p. J. Cell Biol. 2005;168:185–191. doi: 10.1083/jcb.200407118. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 32.Faulhammer F., Kanjilal-Kolar S., Knödler A., Lo J., Lee Y., Konrad G., Mayinger P. Growth control of Golgi phosphoinositides by reciprocal localization of sac1 lipid phosphatase and pik1 4-kinase. Traffic. 2007;8:1554–1567. doi: 10.1111/j.1600-0854.2007.00632.x. [DOI] [PubMed] [Google Scholar]
• 33.Flanagan C.A., Schnieders E.A., Emerick A.W., Kunisawa R., Admon A., Thorner J. Phosphatidylinositol 4-kinase: gene structure and requirement for yeast cell viability. Science. 1993;262:1444–1448. doi: 10.1126/science.8248783. [DOI] [PubMed] [Google Scholar]
• 34.Yesbolatova A., Saito Y., Kitamoto N., Makino-Itou H., Ajima R., Nakano R., Nakaoka H., Fukui K., Gamo K., Tominari Y., et al. The auxin-inducible degron 2 technology provides sharp degradation control in yeast, mammalian cells, and mice. Nat. Commun. 2020;11:5701. doi: 10.1038/s41467-020-19532-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 35.McMahon C., Baier A.S., Pascolutti R., Wegrecki M., Zheng S., Ong J.X., Erlandson S.C., Hilger D., Rasmussen S.G.F., Ring A.M., et al. Yeast surface display platform for rapid discovery of conformationally selective nanobodies. Nat. Struct. Mol. Biol. 2018;25:289–296. doi: 10.1038/s41594-018-0028-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 36.Orci L., Palmer D.J., Ravazzola M., Perrelet A., Amherdt M., Rothman J.E. Budding from Golgi membranes requires the coatomer complex of non-clathrin coat proteins. Nature. 1993;362:648–652. doi: 10.1038/362648a0. [DOI] [PubMed] [Google Scholar]
• 37.Gibson D.G., Young L., Chuang R.-Y., Venter J.C., Hutchison C.A., Smith H.O. Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat. Methods. 2009;6:343–345. doi: 10.1038/nmeth.1318. [DOI] [PubMed] [Google Scholar]
• 38.Schindelin J., Arganda-Carreras I., Frise E., Kaynig V., Longair M., Pietzsch T., Preibisch S., Rueden C., Saalfeld S., Schmid B., et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods. 2012;9:676–682. doi: 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 39.Schneider C.A., Rasband W.S., Eliceiri K.W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods. 2012;9:671–675. doi: 10.1038/nmeth.2089. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 40.Bolte S., Cordelières F.P. A guided tour into subcellular colocalization analysis in light microscopy. J. Microsc. 2006;224:213–232. doi: 10.1111/j.1365-2818.2006.01706.x. [DOI] [PubMed] [Google Scholar]

Supplementary Materials

Data Availability Statement

• •

  Data reported in this paper will be shared by the lead contact upon request.

• •

  This paper does not report novel codes.

• •

  Any additional information required to reanalyze the data reported in this paper is available from the lead contact.