|Home | About | Journals | Submit | Contact Us | Français|
Yeast cells contain two Bro1 domain proteins: Bro1, which is required for endosomal trafficking, and Rim20, which is required for the response to the external pH via the Rim101 pathway. Rim20 associates with endosomal structures under alkaline growth conditions, when it promotes activation of Rim101 through proteolytic cleavage. We report here that the pH-dependent localization of Rim20 is contingent on the amount of Bro1 in the cell. Cells that lack Bro1 have increased endosomal Rim20-green fluorescent protein (GFP) under acidic conditions; cells that overexpress Bro1 have reduced endosomal Rim20-GFP under acidic or alkaline conditions. The novel endosomal association of Rim20-GFP in the absence of Bro1 requires ESCRT components including Vps27 but not specific Rim101 pathway components such as Dfg16. Vps27 influences the localization of Bro1 but is not required for RIM101 pathway activation in wild-type cells, thus suggesting that Rim20 enters the Bro1 localization pathway when a vacancy exists. Despite altered localization of Rim20, the lack of Bro1 does not bypass the need for signaling protein Dfg16 to activate Rim101, as evidenced by the expression levels of the Rim101 target genes RIM8 and SMP1. Therefore, endosomal association of Rim20 is not sufficient to promote Rim101 activation.
The Bro1 domain family consists of endosome-associated proteins involved in membrane trafficking and signal transduction (4, 12, 15). The Bro1 domain directs endosomal association through interaction with Snf7 (10), a subunit of ESCRT-III (endosomal sorting complex required for transport). All eukaryotic genomes encode at least two Bro1 domain protein family members. Prior studies with Saccharomyces cerevisiae indicate that its two Bro1 domain proteins, membrane trafficking component Bro1 and alkaline pH signaling protein Rim20, require Snf7 interaction and endosomal localization for function (4, 13, 25). In both cases, mutations that abolish interaction with Snf7 result in nonfunctional protein (10, 25).
Bro1 acts as a modular scaffold: the N-terminal Bro1 domain interacts with Snf7 and tethers Bro1 to the endosome, where its C-terminal region recruits the ubiquitin protease Doa4 in proximity to ubiquitinated multivesicular body cargo (1, 10, 14, 19). Molecular and sequence analyses indicate that Rim20 also functions as a modular scaffold (4, 24, 25). Rim20, like its Aspergillus nidulans ortholog PalA, facilitates the proteolytic activation of the transcription factor Rim101 (whose A. nidulans ortholog is PacC), a pH-responsive zinc finger transcription factor that is highly conserved across fungal species (4, 5, 16, 24, 25). Rim101 proteolysis is dependent upon the cysteine protease Rim13, putative membrane proteins Rim9, Rim21, and Dfg16, and a soluble arrestin-like molecule, Rim8 (2, 7, 20). Rim101 proteolytic processing also requires Snf7, as well as subunits of the ESCRT-I (Vps23, Vps28, Vps37), ESCRT-II (Vps36, Snf8, Vps25), and ESCRT-IIIA complex (Vps20, Snf7). The ESCRT subunits required for Rim101 processing are also required for Snf7 recruitment to endosomes, while ESCRT subunits that do not affect Snf7 localization are not required for Rim101 processing (4, 25).
Despite their broad similarity, functional analysis has not found a significant link between Rim20 and Bro1 (3, 15, 24). While the localization of both Bro1 and Rim20 depends on ESCRT, distinct upstream inputs also exist. Bro1 robustly localizes to endosomes under acidic conditions, while very little Rim20 localizes to endosomes under similar conditions (4). Proper Rim20 localization, but not Bro1 localization, requires the Rim101 putative signaling complex (Rim8, Rim9, Rim21, and Dfg16), as well as an alkaline environment, suggesting the existence of distinct endosome compartments for each Bro1 domain protein. Colocalization analysis supports this view: some endosomes are associated with only Bro1 or Rim20 (4). Interestingly, environmental pH levels may influence not only Rim20 but also Bro1 localization, as absolute levels of Bro1 foci fall under alkaline conditions. Thus, as levels of one Bro1 domain protein associated with an ESCRT-endosome population rises, the levels of the homologous protein associated with an ESCRT-endosome population falls (4). Here we have tested the specific model that Bro1 and Rim20 compete for association with ESCRT-endosomes. Our findings indicate that Bro1-Rim20 competition occurs in vivo, but functional analysis argues that competition is not a major pH response-regulatory step.
The S. cerevisiae strains used here are listed in Table 1, and the primers used are listed in Table 2. The bro1Δ::URA3 mutation was created by PCR-directed gene disruption using primers BRO1.URA3 F and BRO1.URA3 R against a pRS306 template and transformed and URA+ cells were selected. Disruptions were confirmed by PCR using primers flanking the target region. Overexpression plasmids encoding BRO11-367 (pJB31) and BRO1 (pJB30) were made by cloning a PCR product encompassing 500 bp of 5′ regulatory sequence and relevant sequence into the pYES2.1 vector (Invitrogen) in accordance with the manufacturer's instructions. The Bro1 domain was defined as residues 1 to 367 (bp 1 to 1101), consistent with structural information (10).
Yeast growth media (YPD and SC) were of standard composition (8). For pH exposure assays with S. cerevisiae, SC medium containing 0.1 M HEPES was freshly titrated to pH 4.0 with HCl or to pH 8.3 with NaOH and used immediately as described previously (4). All cultures and plates were incubated at 30°C.
Imaging was performed at room temperature with a Nikon Eclipse E800 wide-field fluorescence microscope, a Nikon Plan Apochromat 100× 1.4 objective (Nikon, Melville, NY), and a Hamamatsu Orca100 digital charge-coupled device camera (Hamamatsu, Bridgewater, NJ). Images were acquired with OpenLab Improvision software and processed in Adobe Photoshop CS3 10.01 (Adobe, San Jose, CA) and ImageJ (NIH, Bethesda, MD).
Yeast strain BY4741 and rim101Δ, bro1Δ, dfg16Δ, bro1Δ dfg16Δ, and vps4Δ mutants were grown overnight, diluted into fresh YPD to an optical density at 600 nm (OD600) of ~0.2, and grown at 30°C with shaking to an OD600 of ~1.0. Equal amounts of cells were shifted for 120 min to YPD plus 0.1 M HEPES (pH 4.0 or 8.3). Cells were isolated by filtration, flash-frozen, and stored at −80°C. RNA isolation and subsequent qPCR analysis were done as described previously (18). Briefly, total RNA was isolated by using the RiboPure Yeast kit (Ambion) in accordance with the manufacturer's instructions. cDNA was prepared using an AffinityScript Multiple Temperature cDNA synthesis kit (Stratagene) by following the manufacturer's instructions. Primers located in the 3′ end of the gene were designed using Primer3 (http://frodo.wi.mit.edu/primer3/). Reverse transcription (RT) reaction mixtures were prepared in triplicate using iQ SYBR supermix (Bio-Rad), and RT-PCR was performed using a Bio-Rad iCycler (Bio-Rad). Data analysis was conducted using Bio-Rad iQ5 software 2.0. Transcript levels were normalized against TDH3 expression, and gene expression changes were calculated by the ΔΔCT method.
Bro1 and Rim20 require ESCRT-III subunit Snf7 for localization but show a reciprocal dependence on the environmental pH (4). A simple model is that one Bro1 domain protein displaces the other from Snf7 and hence allows endosomal localization. Therefore, we reasoned that loss of Bro1 may permit Rim20-endosome association under otherwise restrictive conditions. We examined the localization of a functional Rim20-green fluorescent protein (GFP) fusion, expressed from the RIM20 genomic locus, in a bro1Δ mutant background. Under acidic conditions, fewer than 10% of wild-type (WT) control cells exhibited a single Rim20-GFP focus (Fig. 1A and D). In contrast, all bro1Δ mutant cells exhibited one or more Rim20p-GFP foci (Fig. 1C and D). The positive control vps4Δ mutant cells (4) also exhibited Rim20-GFP on endosomes under acidic conditions (Fig. 1B). These findings indicate that Bro1 prevents Rim20-endosome association under acidic growth conditions.
A second prediction of the Bro1-Rim20 competition model is that overexpression of Bro1 will inhibit Rim20-endosome association under alkaline growth conditions. We quantified Rim20-GFP foci in cells carrying plasmids that overexpress the full-length BRO1 gene or a truncated BRO1 sequence encoding only the Bro1 domain. In control cells carrying the plasmid vector, Rim20-GFP foci were abundant under alkaline conditions (Fig. 2A and D). In contrast, strains that overexpressed Bro1 or just the Bro1 domain had reduced frequencies of Rim20-GFP foci (Fig. 2B, C, and D). These results support the model that Bro1 and Rim20 compete for association with endosomes and that competition is mediated by the Bro1 domain.
If Rim20 associates with ESCRT in the absence of Bro1, then ESCRT subunits should be required for the novel Rim20-GFP foci that form in the bro1Δ mutant background under acidic growth conditions. Thus, we measured the accumulation of Rim20-GFP foci in a panel of bro1Δ mutants lacking ESCRT subunits. Mutations that eliminated ESCRT-I, -II and -III subunits abolished most Rim20-GFP foci (Fig. 3A) compared to the WT (Fig. 3C). Although foci were largely abolished under acidic conditions (Fig. 3A and D), a few cells formed foci under alkaline conditions (Fig. 3B). These results are consistent with previous observations in BRO1 mutant strains (4). The dependence of Rim20-GFP foci on ESCRT subunits argues that the foci are endosomal compartments.
In addition to a shared requirement for ESCRT subunits, localizations of Bro1 and Rim20 have distinct genetic requirements. If Rim20 enters the Bro1 localization pathway in the bro1Δ mutant background, the formation of Rim20-GFP foci in this background may have Bro1-specific genetic requirements. The VPS27 gene product promotes the association of ESCRT-I with endosomes but is not required for Rim101 processing (4, 9, 25). Comparison of a vps27Δ bro1Δ mutant strain to a bro1Δ mutant strain revealed comparable amounts of Rim20 foci formed under alkaline conditions, when the Rim101 pathway is active (Fig. 3B). However, fewer foci were observed under acidic conditions in a vps27Δ bro1Δ mutant strain than in a bro1Δ mutant strain (Fig. 3A and D). Analysis of four Rim101 pathway genes (RIM8, RIM9, RIM21, DFG16), positive regulators of alkaline pH-induced Rim20 localization, showed that Rim20-GFP foci occurred in the bro1Δ mutant background independently of these genes (Fig. 4). In addition, loss of either the protease Rim13 or Rim101 itself had little effect on Rim20-GFP foci in the bro1Δ mutant background (Fig. 4A and D), whereas their loss causes increased numbers of Rim20-GFP foci in WT cells (4). We conclude that the novel Rim20-GFP foci that arise in bro1Δ mutant strains have the genetic requirements for Bro1-endosome association.
Previous studies in a vps4Δ mutant background suggested that Rim20-endosome association allowed some activation of Rim101 under otherwise restrictive conditions. In vps4Δ mutant strains, there is considerable Rim20-endosome association and partial Rim101 activation in the absence of many Rim101 pathway members and ESCRT subunits (4, 6). If endosomal Rim20 in a bro1Δ mutant background is sufficient for Rim101 processing, then mutant strains that form Rim20-GFP foci should promote Rim101 activity, as measured by repression of Rim101 target genes. Thus, we examined the expression of Rim101 target genes RIM8 and SMP1 under acidic and alkaline growth conditions. Rim101 represses both genes and is found associated with the promoter sequences of both genes (11). As expected, expression of both genes was low in the WT strain and elevated in a rim101Δ mutant and in the Rim101 pathway mutant dfg16Δ deletion-carrying strain (Fig. 5A and B). Loss of BRO1 did not restore repression in the dfg16Δ mutant (Fig. 5A and B), despite the presence of Rim20-GFP foci (Fig. 4). We conclude that association of Rim20 with endosomes due to lack of Bro1 is not sufficient to promote Rim101 activity.
Rim20 localizes to endosomes under alkaline pH conditions and is essential for proteolytic activation of the Rim101 transcription factor. Prior studies with S. cerevisiae vps4Δ mutants suggested that Rim20-endosome association may be the limiting step for activation of Rim101 (4, 6). This model was based on the finding that the vps4Δ defect promoted both Rim20-endosome association and partial Rim101 activation under a variety of conditions that were restrictive in VPS4 mutant strains. We report here that a bro1Δ mutant strain separates the two outcomes, permitting Rim20-endosome association under otherwise restrictive conditions but not Rim101 activation. Our observations together indicate that there must be an additional level of pH- and/or Rim101 pathway-dependent control over the Rim20-Rim13-Rim101 complex that leads to Rim101 cleavage.
Our experiments suggest that Bro1 domain proteins, expressed at their natural levels, compete for endosome association. These findings are consistent with the overall similarity of the Bro1 domains of Bro1 and Rim20 and with the finding that many alanine scan mutations in SNF7 impair both Bro1- and Rim20-related functions (22). In fact, Rim20-Bro1 competition may help to explain why snf7 mutants whose protein products have only a twofold reduced ability to bind Bro1 in vitro have endocytic trafficking defects as severe as those of a snf7Δ null mutant strain (21). Similarly, competition by Bro1 may contribute to the Rim101 pathway-specific phenotypes of the alanine scan snf7 alleles (22). However, we did not see functional consequences of such competition with WT Snf7 in experiments reported here. It is likely that simple replacement of Bro1 with Rim20 does not open up sites on the endosomes for other essential upstream components of the Rim101 pathway to assemble. In fact, it has been shown recently that Snf7 binding to both ESCRT proteins and Rim proteins is much stronger under alkaline growth conditions (21), which might explain why artificial Rim20 localization does not recapitulate other changes in the cell upon a shift to alkaline pH. From these results, it seems possible that the consequences of Bro1 domain competition may have been minimized through evolutionary selection.
Interestingly, when Rim20 substitutes for Bro1 at endosomal sites, it assumes the regulatory requirements for Bro1-endosome association. Specifically, under acidic conditions in the bro1Δ mutant background, Rim20-GFP foci are independent of upstream Rim101 pathway components and become dependent upon Vps27. This observation fits well with our mechanistic understanding of ESCRT-I recruitment to endosomes. Vps27 promotes ESCRT-I recruitment, presumably in response to the monoubiquitination signal that marks plasma membrane proteins for endocytosis (17, 23). Recent results from Vincent and colleagues suggest that the Rim101 pathway upstream components, via Rim8, may function analogously in alkaline pH-dependent ESCRT-I recruitment (7). Thus, it seems reasonable that most endosomal ESCRT under acidic growth conditions depends upon Vps27, because Rim8 and other upstream Rim101 pathway components are inactive.
Using double-mutant analysis in the absence of BRO1, we find upregulation of Rim20 localization but no suppression of Rim101 processing defects, which is unlike that seen in vps4 mutants. One idea to reconcile these findings is that trafficking of Rim101 pathway components necessary for processing complex formation may differ between the two mutant backgrounds. This would imply that a trafficking defect exists in the absence of Bro1 but not in the absence of Vps4, yet no trafficking evidence supports this model. We favor a simpler kinetic model, based on a functional differentiation between Vps4 and Bro1. Vps4 is an enzyme critical for endosomal protein disassociation, while Bro1 is a structural protein whose endosome disassociation is controlled by Vps4. In the absence of Vps4, significant endosome-protein capture and sequestration occur, including Rim20 and Rim13—the protease that cleaves Rim101—along with Rim101 itself and possibly other components. In contrast, Bro1 is a structural protein, and loss of Bro1 likely only affects the endosome association status of Rim20, with little effect on any additional proteins. Consistent with a carefully calibrated kinetic mechanism, as mentioned above, Rim20 foci are most abundant immediately following pH shock and recede within 10 min, suggesting an acute regulated time period of processing complex formation.
Previously, it was shown that in the absence of Bro1, Rim101 is processed efficiently but there is still a defect in transcriptional repression by Rim101 (20). Repression was assayed with a site that required the activity of both Rim101 and the repressor Nrg1. In our study, we quantitatively recorded the transcription of two native Rim101 targets, RIM8 and SMP1, and assayed RNA accumulation from the native genomic loci. Both were repressed by Rim101 in WT and bro1Δ mutant cells while being derepressed in dfg16Δ and bro1Δ dfg16Δ mutant cells. Our findings indicate that a bro1Δ mutation does not alleviate Rim101-dependent repression at all target genes.
Our observations reported here argue that association of Rim20 with endosomes has to be accompanied by an additional signal(s) mediated by alkaline pH that may change the conformation of the ESCRT complex or its interaction with Rim101 pathway components. This change mediated by a pH shift may specialize some ESCRT-endosomes as hubs for downstream signaling and open new sites to facilitate recruitment of the protease Rim13 or other factors.
We are grateful to all members of our lab for helpful discussions and for comments on the manuscript and to Miguel Peñalva and members of his lab for comments on this project.
This work was supported by National Institutes of Health grant 5R01AI070272 to A.P.M.
Published ahead of print on 26 February 2010.