Wbm0152, an outer membrane lipoprotein of the Wolbachia endosymbiont of Brugia malayi, inhibits yeast ESCRT complex activity

Abstract

Human pathogenic filarial nematodes of the family Onchocercidae, including Brugia malayi and Onchocerca volvulus, cause debilitating filarial diseases such as lymphatic filariasis and river blindness. These mosquito-borne pathogens are obligately colonized by the gram-negative intracellular alphaproteobacterium, Wolbachia pipientis, which is essential for nematode sexual reproduction, long-term survival, and pathogenicity in the mammalian host. Like many intracellular bacteria, Wolbachia likely uses numerous surface-exposed and secreted effector proteins to regulate its ability to persist and replicate within nematode host cells. However, due to the inability to cultivate Wolbachia in the laboratory and the genetic intractability of both filarial nematodes and the bacterium, the molecular underpinnings that define the bacterium:nematode relationship are almost completely unknown. In this work, we show that the expression of a Wolbachia outer membrane lipoprotein, wBm0152, in Saccharomyces cerevisiae inhibits the activity of the conserved Endosomal Sorting Complex Required for Transport (ESCRT) complex and strongly disrupts endosomal maturation, leading to defects in ubiquitylated protein turnover. Using in vivo bimolecular fluorescence complementation, we find that Wbm0152 interacts with the Vps2p subunit of the ESCRT-III subcomplex as well as the Vps2p ortholog (BmVps2, Bm6583b) from a Wolbachia host nematode, Brugia malayi. These data suggest a novel role of ESCRT in Wolbachia persistence providing insight into the elusive relationship between these two organisms.

AUTHOR SUMMARY

Filarial diseases of mammals, including lymphatic filariasis and canine heartworm, are caused by vector-borne filarial nematodes of the family Onchocercidae. Many of the nematodes in this family are obligately colonized by an intracellular bacterium, Wolbachia pipientis, which is essential for the nematode’s long-term survival, reproduction, and pathogenicity. Therefore, understanding the mechanisms used by Wolbachia to persist and replicate within host cells could provide new molecular targets for treating filarial infections. Due to the genetic intractability of both nematode and bacterium, however, significant progress on characterizing these interactions have proven difficult. In this work, we show that a predicted outer membrane lipoprotein, Wbm0152, of the Wolbachia endosymbiont of Brugia malayi inhibits yeast Endosomal Sorting Complex Required for Transport (ESCRT) complex activity in vivo. Wbm0152 interacts with a core subunit of the yeast ESCRT-III complex, as well as with the orthologous ESCRT-III protein from Brugia. ESCRTs are conserved across eukaryotes and are important for diverse cellular processes such as endosomal maturation, autophagy, and cellular division. As Wolbachia persists within a membrane-bound compartment within Brugia and must avoid host autophagic pathways, this study presents a potential mechanism by which Wolbachia may regulate Brugia membrane trafficking pathways to ensure its intracellular survival.

INTRODUCTION

Pathogenic filarial nematodes, such as Brugia malayi, Wuchereria bancrofti, and Dirofilaria immitis, are a group of parasitic, mosquito-borne nematodes that are known to cause debilitating and disfiguring illness in millions of humans and animals worldwide. Typically, humans infected with such nematodes are treated with a regimen of anthelmintic drugs such as ivermectin [1–3]. However, due to ivermectin’s inability to effectively eradicate adult stage worms in vivo [4, 5] combined with increasing observations of anthelmintic resistance in the nematode population [6–8], the demand for identifying novel drug targets to support the elimination of these nematodes has increased rapidly. Interestingly, filarial nematodes of the family Onchocercidae – which include the genera Brugia, Wuchereria, and Dirofilaria – are colonized by Wolbachia pipientis, a gram-negative, obligately intracellular alphaproteobacterium, which is essential for proper nematode reproduction and survival [9, 10]. Elimination of this bacterium from the nematode using doxycycline or tetracycline treatments leads to the sterilization of adult worms, killing of microfilaria, and suppression of infection symptoms in humans [11–14]. Despite the well documented requirement of Wolbachia for filarial nematode survival, little is understood about the molecular underpinnings of the complex bacterial:nematode essential endosymbiosis. This lack of knowledge predominantly stems from the inability to cultivate Wolbachia in the laboratory and the poor genetic tractability of both Wolbachia and these filarial nematodes. Therefore, our knowledge about the Wolbachia:nematode relationship stems largely from high-resolution microscopy techniques and heterologous model systems to hypothesize how this bacterium persists within its host.

Wolbachia, like many other intracellular bacterial pathogens, including Legionella pneumophila, Chlamydia trachomatis, and the closely related alphaproteobacterium Anaplasma phagocytophilium, persists and replicates within a host-derived, vacuolar-like compartment [15, 16]. The composition of these bacteria-laden compartments is distinct from normal host organelles and they do not fuse with lysosomes, thereby protecting the bacterium from host degradative pathways. Previous research on both nematode-derived and Drosophila-derived Wolbachia has suggested that this Wolbachia-containing compartment is likely formed from ER or Golgi membranes [17, 18], although the precise lipid and protein composition of this compartment remains unknown. Many intracellular bacteria employ dedicated secretion systems to deploy numerous secreted and surface-exposed proteins – termed effectors – that actively modulate host membrane trafficking and lipid synthesis pathways to create these unique compartments and to inhibit phagosome:lysosome fusion to prevent bacterial degradation [15]. As Wolbachia contains a Type IV secretion system [19–22] and survives within an intracellular membrane-bound compartment, it is likely that Wolbachia also uses secreted effectors to modulate host membrane dynamics and to ensure the creation of its intracellular replicative niche.

Previously, our laboratory used the unicellular eukaryote Saccharomyces cerevisiae as a model system to screen several predicted Type IV-secreted proteins from the Wolbachia endosymbiont of Brugia malayi for the ability of these candidate effectors to manipulate conserved eukaryotic processes [19, 23]. Expression of one such protein, wBm0152, lead to the aberrant accumulation of an enlarged prevacuolar compartment and the failure to deliver representative endosomal membrane-bound cargo proteins (CPS or Sna3) to the lumen of the degradative vacuole [19]; these phenotypes are similar to those observed in yeast strains defective in endosomal sorting complexes required for transport (ESCRT) activity [24]. ESCRTs are highly conserved, multisubunit complexes that are responsible for the invagination and scissioning of cellular membranes away from the cytosol. In yeast, ESCRTs are essential for intralumenal vesicle formation (ILV) during endosome maturation and for microautophagy [25–28]. In mammalian cells, ESCRTs are important for these same functions as well as membrane repair [29], nuclear envelope remodeling [30, 31], and viral budding [32–34].

ESCRT subcomplexes assemble in an ordered, stepwise fashion in which the first three complexes (ESCRT-0, -I, -II) function to bind and cluster ubiquitylated proteins on endosomal membranes, as well as recruiting and binding protein subunits of the downstream ESCRT complex. The fourth and most highly conserved complex (ESCRT-III) has the unique job of inducing membrane deformation and scissioning to allow for the creation of the endosomal ILVs. ESCRT-III assembly on endosomes is initiated when myristolated Vps20p is recruited to the endosomal membrane by the assembled ESCRT-II complex (consisting of Vps22p, Vps25p, and Vps36p), which then recruits and initiates the homopolymerization of Snf7p. The Vps2p:Vps24p module of ESCRT-III binds to the Snf7p homopolymer, extending its polymerization laterally, thus creating a three-dimensional helical structure that induces membrane curvature towards the lumen of the endosome [35–40]. The ESCRT-III accessory protein, Bro1p, also binds directly to Snf7p, promoting the recruitment of the ubiquitin hydrolase, Doa4p, which removes and recycles the ubiquitin moiety from target proteins prior to ILV internalization. Finally, the AAA+ ATPase, Vps4p, is recruited to disassemble the ESCRT complexes and completes the scission of the ILV vesicle into the lumen of the endosome [41–43]. Disruption of most of the core ESCRT proteins results in the aberrant accumulation of flattened stacks of endosomes that contain few ILVs and fail to properly fuse with the vacuole/lysosome; these abnormal endosomes have been termed ‘class E compartments’ [44].

In this study, we now show that wBm0152 expression inhibits the ILV-formation activity of ESCRT in yeast. By using the rapamycin-induced degradation (RapIDeg) system [45] to visualize ESCRT activity in vivo, we show that wBm0152 expression prevents the ESCRT-dependent translocation of an artificially-ubiquitylated vacuole membrane protein into the lumen of the degradative vacuole. We also demonstrate that expression of wBm0152 inhibits the formation of ILVs in yeast endosomes using cryo-fixation tomography. Using bimolecular fluorescence complementation assays, we find that Wbm0152 binds in vivo to the yeast ESCRT-III protein Vps2p – as well as the B. malayi Vps2p ortholog – and we show that Wbm0152 alters the recruitment of the downstream ESCRT accessory proteins through colocalization studies and electron microscopy. These findings not only identify a bacterial protein capable of inhibiting ESCRT activity but also illuminate an intricate relationship between Wolbachia and its nematode host to support its intracellular survival.

RESULTS

Wbm0152 inhibits ESCRT-dependent protein degradation in yeast.

Our previous research showed that expression of wBm0152 in yeast inhibited the normal delivery of the endosomal cargo proteins carboxypeptidase S (CPS) and Sna3p to the degradative vacuole lumen; these phenotypes are also observed in ‘class E’ protein sorting mutants of yeast caused by defects in Endosomal Sorting Complex Required for Transport (ESCRT) subunits [19, 46, 47]. Therefore, we hypothesized that wBm0152 expression may be inhibiting ESCRT either directly or indirectly. To test this hypothesis, we utilized the Rapamycin Dependent Degradation Assay (RapIDeg) [45] to determine the impact of wBm0152 expression on the ESCRT-dependent vacuolar degradation of an artificially-ubiquitylated protein. In this assay, cells express the vacuolar membrane iron transporter, Fth1p, are fused to a GFP and FK506 Binding Protein (FKBP) domain. These strains also harbor the FKBP-rapamycin binding domain (FRB) fused to 3x ubiquitin, which allows the dimerization of the FRB and FKBP domains in the presence of rapamycin, causing the artificial ubiquitylation of the Fth1-GFP protein. This ubiquitylation recruits ESCRT complexes to the vacuole membrane, thus concentrating the Fth1-GFP-3xUb cargo and delivering Fth1-GFP-3xUb into the vacuolar lumen in an ESCRT-dependent manner, leading to the degradation of the Fth1-GFP cargo [45, 48].

In empty vector control RapIDeg strains, we observed clear vacuolar membrane localization of Fth1-GFP, as expected (Fig. 1A). After rapamycin addition, we observed approximately 90% of the cell population move Fth1-GFP into the vacuole lumen, with concomitant degradation of that protein (Figs. 1A and B). In contrast, strains expressing wBm0152, failed to degrade Fth1-GFP with only 2% of cells accumulating Fth1-GFP in the vacuole lumen 2h after treatment with rapamycin (Fig. 1A and B), showing that wBm0152 inhibits the ESCRT-dependent invagination of the vacuolar membrane in response to a ubiquitylated membrane protein. Therefore, wBm0152 expression prevents the ESCRT-dependent turnover of a ubiquitylated protein in yeast.

Figure 1. Wbm0152 expression inhibits ESCRT-dependent ubiquitylated protein turnover.

RapIDeg yeast strains harboring either a vector control or the β-estradiol inducible wBm0152 expression vector were assayed for vacuolar turnover of Fth1-GFP-FKBPx2, as in Methods. A) Representative images showing localization of Fth1-GFP in response to 1 μg mL⁻¹ rapamycin; bar = 5 μ. Listed percentages indicate the population of cells displaying the presented Fth1-GFP localization; n > 400 cells per condition. B) Representative α-GFP immunoblot of whole cell lysates prepared from cells in A).

Expression of wBm0152 in yeast induces ESCRT mutant-like growth defects.

Previously, our lab observed a slight growth defect of wBm0152-expressing yeast strains in the presence of zinc and caffeine, suggesting expression of Wbm0152 may alter endolysosomal membrane compartment function [19]. Based on our RapIDeg results indicating that wBm0152 likely inhibits ESCRT activity, we sought to determine if wBm0152 expression induced specific growth phenotypes like those observed in yeast strains lacking individual ESCRT subunit proteins. Previous research identified that the 1,3-β-glucan-binding azo dye, Congo red [49], inhibited the growth of several individual ESCRT subunit mutants [50]. This same study also revealed that many of the ESCRT-III or ESCRT-III related subunits have distinct levels of growth inhibition when grown on media containing Congo red. Particularly, ESCRT-III mutant strains such as vps2Δ and vps24Δ are exquisitely sensitive to the presence of Congo red, while other strains – like snf7Δ and vps20Δ – are only moderately sensitive to these conditions, allowing us to compare growth patterns between ESCRT subunit mutants and wBm0152 expressing strains.

Vector control strains are insensitive to the presence of 15 μg ml⁻¹ Congo red on minimal medium (Fig. 2). Strains constitutively expressing wBm0152, however, are strongly inhibited for growth under these conditions (Fig. 2). When comparing the growth of this strain to representative deletion mutants of ESCRT complex subunits, we noted that our wBm0152-expressing strain has similar levels of sensitivity to Congo red as vps2Δ, vps24Δ, and bro1Δ strains; no other ESCRT subunit deletion strain showed comparable growth sensitivities under these conditions. These results indicate that expression of wBm0152 likely impacts the function of the yeast ESCRT-III subcomplex, as well as the downstream ESCRT accessory proteins (Bro1p) required for efficient ESCRT function. Vps2p and Vps24p are known to interact in a bipartite module, binding to polymerized Snf7 to promote the creation of the 3-dimensional helix to induce endosomal membrane deformation and ILV formation [35]. Bro1p, known to be activated by the polymerization of Snf7, is a key accessory protein required for the recruitment of the ubiquitin hydrolase, Doa4p [51]. In addition to its role in recruiting Doa4p to ESCRT-III, Bro1p:Snf7p interactions also inhibit the association of Vps4p with the ESCRT complex, allowing time for the deubiquitylation of protein cargo by Doa4p [52]. Therefore, Wbm0152 may be disrupting the recruitment of the downstream ESCRT accessory proteins to the assembled complex. As ESCRT-III is primarily responsible for the physical deformation of the endosomal membrane to generate ILVs, we next chose to directly observe the impact of wBm0152 expression on yeast endosomal maturation via electron microscopy.

Figure 2. Wbm0152 expression phenocopies ESCRT mutant growth defects.

BY4742 yeast strains harboring the indicated ESCRT subunit deletion, an empty vector control, or the constitutively expressing pYES_TDH3-wBm0152 (Methods) were grown overnight at 30 °C in CSM-URA media. Cultures were diluted to a final OD₆₀₀ = 1.0, serially diluted 10-fold four times into sterile water, and 10 μL of each dilution was spotted to CSM media lacking uracil either lacking or containing 15 μg mL⁻¹ Congo red. Plates were incubated at 30°C and imaged after 72h.

wBm0152 expression results in defective ILV formation and endosomal morphology.

During endosomal maturation, ESCRT activity is required to create the ILVs indicative of late endosomes (also termed multivesicular bodies, or MVBs). To visualize to impact of wBm0152 expression on endosomal maturation, we used cryo-fixation for electron tomography to generate high-resolution images of endolysosomal membrane compartments and associated organelles. We observed spherical MVBs in both the control strain and the wBm0152-expressing strain, although the strain harboring wBm0152 appeared to have a slight reduction in the numbers of MVBs per 100 cell profiles (Figs. 3A–C and D–F, quantified in Fig. 3G). Notably, we also found tubular MVBs in strains expressing wBm0152 (Figs. 3E and 3F), similar to those found in other ESCRT-impaired yeast strains like vta1Δ, did2Δ, vps60Δ, or BRO1-overexpressing strains [42, 43, 52–58]. Furthermore, we noted numerous aberrant membrane structures in wBm0152-expressing cells when compared to control strains, including tubular endoplasmic reticulum and increased frequency of lipid droplets (Figs. 3E and 3F).

Figure 3. wBm0152 expression inhibits intralumenal vesicle formation in vivo.

Yeast strains harboring either the vector control (A-C) or wBm0152 expression vector (D-F) were visualized via cryo-fixation electron tomography (Methods, bar = 100 nm) and membrane structures were modeled as indicated: MVBs, yellow; degradative vacuole, red; endosomal ILVs, red spheres; ER, green; lipid droplets; white. G) Quantification of total MVBs (left) or ILVs per MVB (right) observed across 100 cell profiles for the indicated yeast strain. H) Calculated diameters of all ILVs observed in cell profiles from (G). I) The total surface area of the limiting membranes of observed MVBs or ILVs from the tomogram profiles of the indicated strains were quantified, as outlined in Methods.

To better observe the morphology of the smaller and less frequent MVBs found in wBm0152-expressing strains, we modeled the membrane compartments found in each tomographic stack and calculated the physical dimensions of the relevant structures. Compared to control strains, the endosomes in strains harboring wBm0152 contained fewer ILVs per MVB (Fig. 3G), although those infrequent ILVs tended to be larger in diameter than those found in control strain MVBs (33.38 nm vs 26.51 nm, P<0.0001; Fig. 3H). By calculating the surface area of the limiting membranes of the MVBs and ILVs in both strains, we found that the cumulative surface area of the MVBs in wBm0152-expressing strains was approximately twice that of the control strains with a concomitant reduction in ILV membrane surface area (Fig. 3I). Therefore, wBm0152 expression strongly inhibits the formation of endosomal ILVs in vivo, providing additional evidence of its ability to inhibit ESCRT complex activity.

Wbm0152 colocalizes with representative ESCRT complex subunits.

To determine if Wbm0152 localizes with ESCRT protein subunits in vivo, we performed a colocalization analysis with GFP-tagged candidate ESCRT protein subunits and Wbm0152-mRuby. Each ESCRT-GFP subunit (Vps27-GFP, ESCRT-0; Vps36-GFP, ESCRT-II; Snf7-GFP, ESCRT-III; Bro1-GFP, accessory) was localized to multiple punctate structures indicative of endosomal compartments, as expected [36, 51, 59, 60] (Fig. 4A). Analysis of these strains co-expressing wBm0152-mRuby showed strong colocalization of Wbm0152-mRuby with Vps27-GFP, Vps36-GFP, and Snf7-GFP, as measured by Pearson Correlation Coefficient (Figs. 4A and 3B). Interestingly, however, Wbm0152 failed to strongly colocalize with the ESCRT accessory protein Bro1-GFP (Figs. 4A and 4B). As it is known that Bro1p normally binds to – and colocalizes with – Snf7p to direct the recruitment of Doa4p to the assembled ESCRT complex [51, 52], it was surprising to note that Bro1p did not strongly colocalize with Wbm0152, despite the observation that Snf7p colocalized with Wbm0152. These results show a reduction in the recruitment of Bro1p to assembled ESCRT complexes in the presence of Wbm0152, which may result from Wbm0152-dependent ESCRT-III complex assembly or disassembly defects.

Figure 4. wBm0152 colocalizes with core ESCRT subunits in vivo.

A) SEY6210 yeast strains harboring the indicated chromosomal ESCRT subunit-GFP and the pYES-wBm0152-mRuby expression vector were grown to saturation at 30°C in selective media. Cells were diluted 1:10 into fresh selective media containing 1 μM β-estradiol, outgrown for 6h at 30°C, and imaged. Bar = 5 μ. B) Truncated violin plot showing calculated Pearson’s Correlation Coefficient of the ESCRT:wBm0152 colocalization.

Wbm0152 interacts with ESCRT-III protein Vps2p.

As it appears that wBm0152 expression either inhibits the activity of ESCRT-III or disrupts ESCRT complex assembly dynamics, we hypothesized that Wbm0152 would interact with at least one protein subunit of the ESCRT-III complex. By utilizing a split-Venus bimolecular fluorescence complementation assay (BiFC) [61–63], we sought to identify potential ESCRT binding partners of Wbm0152 through the reconstitution of fluorescence upon protein:protein interactions. Therefore, yeast strains harboring individual ESCRT protein fusions with either the non-fluorescent Venus N-terminal (VN) or Venus C-terminal (VC) domains were transformed with a copper-inducible expression construct containing wBm0152 fused to the complementary Venus fragment. Should Wbm0152 interact with any of these ESCRT subunits and bring the VN and VC domains into close proximity, the fluorescent Venus chromophore will be reconstituted and detectable via fluorescence.

In yeast strains expressing Snf7-VC and Vps2-VN – proteins from the ESCRT-III complex known to directly interact in vivo [42] – we detected numerous fluorescent punctae, as expected (Fig. 5); these punctae were not observed in strains harboring the Snf7-VC and empty vector control constructs (Fig. 5). Using the inducible wBm0152-VN (or-VC) expression construct from above and a number of ESCRT subunit strains expressing the reciprocal Venus fusion, we identified fluorescent punctae under wBm0152-VC induction conditions only in Vps2-VN harboring strains (Fig. 5); no other subunit tested showed fluorescence under similar conditions (Fig. S1; tested Vps25p (ESCRT-II), Vps20p (ESCRT-III), Vps24p (ESCRT-III), Snf7p (ESCRT-III), Bro1p (accessory), the Snf7p-binding domain of Bro1p (BOD [42]), and Doa4p). Interestingly, we did not see reconstitution of Venus fluorescence in Vps2-VC/wBm0152-VN strains (Fig. S1). This observation could be due to the C-terminal -VN fusion forcing Vps2 into a constitutively active, ‘open’ conformation to bind Wbm0152, which is known to occur with yeast and mammalian ESCRT-III GFP fusion proteins [40, 64, 65]. Alternatively, the VC/VN components in this putative Vps2p-Wbm0152 interaction may simply be in the wrong orientation to form the fluorescent chromophore upon interaction. Regardless, these results show that Wolbachia Wbm0152 likely interacts specifically with the Vps2p ESCRT-III subunit in vivo.

Figure 5. wBm0152 binds ESCRT-III subunit Vps2 in vivo.

SEY6210 strains harboring either the N-terminus (VN) or C-terminus (VC) of a Venus-YFP molecule on the C terminus of the indicated ESCRT subunit were transformed with the corresponding copper-inducible pYES_CUP1-wBm0152-VN or pYES_CUP1-wBm0152-VC plasmid. Strains were grown in CSM media lacking uracil for 18h at 30°C with shaking, diluted 1:10 into fresh selective media lacking or supplemented with 0.5 mM CuSO₄, and outgrown for 6 hours before imaging. Bar = 5 μ; images are representative of three separate experiments.

Wbm0152 requires normal ESCRT assembly for activity.

Taking advantage of the growth defect we observed on media containing Congo red upon wBm0152 expression in a wildtype strain background, we sought to determine which ESCRT components were required for the Wbm0152-mediated toxicity on this medium. Therefore, candidate ESCRT deletion strains representing one protein from each of the ESCRT-0,-I, and -II complexes – as well as each subunit of the ESCRT-III and downstream accessory complexes – were grown on media containing Congo red with and without wBm0152 expression. As we observed previously, strains missing components of the ESCRT-0, -I, and -II complexes harboring a vector control did not show strong growth defects in the presence of Congo red, when compared to the vector control wild type strain (Fig. 6). Strikingly, wBm0152 expression in these mutant backgrounds did not cause additional growth defects on Congo red, in contrast to the wild type (Fig. 6). Similarly, wBm0152 expression did not induce Congo red growth defects in vps20Δ or snf7Δ strains (Fig. 6). As each ESCRT subcomplex generally does not assemble and accumulate on the endosomal membrane without the proper assembly and recruitment of the previous ESCRT subcomplex [40, 59, 60, 66, 67], these results show that normal, ordered ESCRT complex assembly is required for wBm0152 activity under these growth conditions. ESCRT disassembly was not important for the toxicity of wBm0152 expression, however, as vps4Δ strains remained sensitive to wBm0152 expression on Congo red (Fig. 6). Due to the strong growth defect observed on Congo red in vps2Δ, vps24Δ, bro1Δ, and doa4Δ strains alone, the necessity of these subunits for Wbm0152 toxicity could not be determined. Taken together, these data suggest the action of Wbm0152 requires fully assembled ESCRT complex and acts upstream of the vps4Δ disassembly apparatus.

Figure 6. Wbm0152 activity requires ESCRT assembly.

BY4742 yeast strains harboring the indicated ESCRT subunit deletion, an empty vector control, or the constitutively expressing pYES_TDH3-wBm0152 (Methods) were grown overnight at 30 °C in CSM-URA media. Cultures were diluted to a final OD₆₀₀ = 1.0, serially diluted 10-fold four times into sterile water, and 10 μL of each dilution was spotted to CSM media lacking uracil either lacking or containing 15 μg mL⁻¹ Congo red. Plates were incubated at 30°C and imaged after 72h.

Wbm0152 interacts with the Brugia Vps2 ortholog in yeast.

We initiated this work to identify an interaction between a Wolbachia candidate effector protein, Wbm0152, with a conserved eukaryotic protein from yeast, which would help elucidate the activity of Wbm0152 in the Wolbachia:Brugia malayi endosymbiosis. As ESCRT complexes are broadly conserved across eukaryotes, we identified the VPS2 ortholog from the B. malayi genome (BM6583, isoform b) [68] and constructed a yeast codon-optimized expression vector to assess Wbm0152:Bm6583b interactions in vivo.

As it is known that Vps2p interacts directly with Snf7p during ESCRT-III complex formation and that interaction can be observed using BiFC, we first ensured that Bm6583b (hereafter, BmVps2) would interact with yeast Snf7p in vivo. As observed previously, a yeast strain expressing Snf7-VC and Vps2-VN reconstitutes Venus fluorescence (Fig. 7A), confirming the expected Vps2p:Snf7p interaction in vivo. Expression of BmVps2-VN from the endogenous yeast VPS2 promoter in the Snf7-VC background also produced fluorescent punctae (Fig. 7A), showing that BmVps2 also interacts with yeast Snf7p and likely engages in ESCRT-III complexes with the yeast protein subunits. Next, we introduced the inducible wBm0152-VC vector previously used into strains expressing BmVps2-VN in both wild type and vps2Δ backgrounds. In both backgrounds, we observed the reconstitution of Venus fluorescence only upon induction of wBm0152-VC expression (Fig. 7B), showing that wBm0152 interacts with BmVps2 in a similar manner to yeast Vps2p in vivo, thus strengthening the possibility that Wbm0152 interacts with ESCRT-III subunits in the nematode.

Figure 7. wBm0152 binds Brugia malayi Vps2 ortholog (Bm6583b) in yeast.

A) Yeast SEY6210 harboring yeast wild type SEY6210 (left) or vps2Δ (right) strains harboring the plasmids pRS414-pVPS2-Bm6583b-VN (Methods) and pYES_CUP1-wBm0152-VC were grown in CSM media lacking uracil and tryptophan for 18h at 30°C with shaking, diluted 1:10 into fresh selective media lacking or supplemented with 0.5 mM CuSO₄, and outgrown for 6 hours before imaging. Bar = 5 μ; images are representative of three separate experiments.

DISCUSSION

In this study, we have shown that expression of Wbm0152 in yeast inhibits the eukaryotic ESCRT complex via interactions with the core ESCRT-III protein, Vps2p. This inhibitory activity requires the formation of a properly assembled ESCRT complex, as in the absence of ESCRT-0, -I, -II, and most -III subunits, wBm0152 expression failed to induce growth defects on media containing Congo red. Importantly, Wbm0152 was also found to interact with the Brugia Vps2 ortholog in yeast (Bm6583b), providing evidence of a physiologically-relevant interaction within the nematode that cannot be easily confirmed in the natural host. These data highlight a potential role for Wbm0152 in the regulation of Brugia ESCRT activity during endosymbiosis.

Despite the many advances made in the analysis of Wolbachia:host interactions, much of the current research focuses on understanding Wolbachia survival and transmission in arthropod hosts, in which the relationship between the bacterium and host is usually parasitic in nature. Through a combination of mutant insect lines and cell culture systems, researchers have shown the importance of host actin dynamics for the uptake and maternal transmission of Wolbachia during host development [69, 70], although information regarding Wolbachia’s ability to actively manipulate these host pathways remains lacking. Understanding of the Wolbachia:nematode relationship, however, remains even less clear due to the genetic intractability of Brugia and Wolbachia, the general difficulty to rear such nematodes, the lack of primary or immortalized nematode cell lines, and the fact that nematode-derived Wolbachia (strain types C and D) is known to express evolutionarily-divergent Type IV-secreted effectors than the arthropod-derived Wolbachia (strain types A and B) [71]. Therefore, the development of an alternative biological model system for the study of these Wolbachia effector proteins found in filarial nematodes is demanded. Our lab, as well as several others, have had great success utilizing Saccharomyces cerevisiae as a model eukaryotic cell to study the molecular underpinnings of many host:bacterium interactions, especially when the target of these bacterial effectors is conserved in eukaryotes [72–74].

Many intracellular bacteria such as Salmonella enterica and Mycobacterium tuberculosis are known to directly modulate host ESCRT activity, likely to support intracellular survival and transmission. Although there are well documented ESCRT-interacting proteins, like the Mycobacterium secreted effectors EsxG and EsxH [75], ESCRT-interacting domains amongst intracellular bacteria are not well conserved, nor are their molecular activities on ESCRT well known. Despite the lack of understanding on how intracellular bacteria interact with ESCRT, there is a wealth of knowledge on how many viruses, including HIV-1, interact with and manipulate ESCRT to promote viral budding within mammalian cells. Most of these viral effector proteins are known to contain P(S/T)AP, PPxY, or YPx(n)L motifs that mimic motifs present on ESCRT [76]. The P(S/T)AP domain is known to interact with TSG101, the human ortholog to the yeast ESCRT-I protein Vps23p [77–79]. The PPxY motif is noted to interact with NEDD4, the human ortholog to the yeast Rsp5p protein [80–82], which is a ubiquitin ligase that interacts with Vps27p and Hse1p to regulate the biogenesis of MVBs [83, 84]. Lastly, the YPX(n)L motif is known to interact with ALIX ( yeast Bro1p), [64] which will also be found in complexes containing the ESCRT-III subunits CHMP2, (Vps2p ortholog), CHMP4 (Snf7p ortholog), and Vps4p (reviewed in [85]). Some of these domains have been found on ESCRT-interacting proteins from other intracellular pathogens, such as GRA14 and RON4 from Toxoplasma gondii, which are known to contain the P(T/S)AP and YPX(n)L domains, respectively [86, 87]. However, it is important to note that Toxoplasma has other known ESCRT-interacting proteins, such as GRA64, which is noted to interact with ESCRT subunits TSG101, VPS37, VPS28, UMAD1, ALG-2, and CHMP4, but does not contain any of the known ESCRT-binding motifs previously mentioned [88]. Therefore, it is difficult to predict the ESCRT-binding activity of a protein based on sequence information alone.

Previous studies have noted the knockdown of ESCRT components like the Vps2 homolog, CHMP2, and the SNF7 homolog, SHRB, in JW18 Drosophila melanogaster embryonic cell lines caused an increase in Wolbachia populations in vivo [89], indicating ESCRT activity likely plays an important role in Wolbachia intracellular persistence. Additionally, activation of autophagic pathways via exogenous addition of rapamycin to either Wolbachia-infected Drosophila cell lines or Brugia malayi was shown to reduce intracellular Wolbachia populations, presumably through lysosomal degradation of the bacterium [90]. As ESCRT activity is required for macro and microautophagy [91–93], it is possible that Wbm0152 (and orthologs) dampens host ESCRT activity to prevent autophagic degradation of the bacterium. Furthermore, RNA-Seq data obtained from 6-week Brugia malayi microfilaria isolated from tetracycline-treated animals to eliminate Wolbachia from the nematode, found that the transcription of Brugia genes producing proteins involved in multivesicular body formation were upregulated in the absence of Wolbachia [94], indicating that the presence of Wolbachia does appear to alter Brugia/filarial nematode endolysosomal membrane dynamics.

Interestingly, Wbm0152 belongs to a conserved family of peptidoglycan-associated lipoproteins (Pal) found extensively throughout gram negative bacteria, where it plays a role in bacterial outer membrane stability and in cell division [95, 96]. In Gram-negative bacteria, Pal proteins interact with the peptidoglycan cell wall, outer membrane proteins, and the periplasmic TolB protein – which is part of the larger inner membrane Tol complex (consisting of TolQ, TolR, and TolA) – and concentrates at the site of cell division to modulate peptidoglycan processing during cell division via the recruitment of cell separation amidases [97, 98]. While it is difficult to hypothesize how a periplasmic-facing outer membrane lipoprotein could interact with ‘external’ host cell proteins in the context of Wolbachia physiology, it is important to note that the Wolbachia genome does not appear to contain any known Tol protein homologs – including TolB [99] – suggesting the role of this Pal-like protein in Wolbachia may be quite different than typically observed in most Gram-negative bacteria. Furthermore, Pal-family proteins have been shown in some bacteria to have a ‘dual confirmation’ in which the population of Pal can have an ‘inward’ facing (periplasmic) C-terminus, or an ‘outward’ facing, exposed C-terminus on the outer membrane surface. Despite this well documented phenotype, the physiological role of this ‘outward’ facing Pal population remains unknown [100]. In support of the hypothesis that Wbm0152 is outward facing in Wolbachia, Wbm0152 has been previously localized to both the surface of the Wolbachia bacterium – as well as on the membrane of the Wolbachia-containing vacuole – via immunofluorescence and immuno-EM in Brugia tissues [101]. The fact that Wbm0152 is present on the surface of the Wolbachia-containing vacuole suggests that this protein may be positioned to interact directly with Brugia cytoplasmic proteins, like ESCRT.

Why might Wolbachia modulate ESCRT activity in Brugia? As Wolbachia lives inside of a membrane-bound compartment within nematode cells, we could hypothesize Wolbachia may be modulating Brugia ESCRT function to initiate ILV formation into the Wolbachia-containing compartment, perhaps to deliver nutrients from the cytosol of the host cell. Furthermore, this activity could also be utilized either for the creation of the Wolbachia containing vacuole from post-Golgi membranes, or to prevent entry into lysosomal degradation pathways that require ESCRT activity, such as autophagy. Lastly, it is known that Wolbachia is required to both maintain a quiescent pool of germline stem cells in the female Brugia germline and stimulates mitotic cell division during embryonic development [102]. As ESCRT-III activity was also found to be critical for abscission during cytokinesis and nuclear envelope repair during cell division [30, 103, 104], it is even possible that Wolbachia directly controls the cell cycle of the Brugia germline cells through the manipulation of ESCRT-III activity. Unfortunately, direct experiments to test these hypotheses are difficult without greatly expanding the availability of research reagents for both Brugia and Wolbachia. Nevertheless, this study marks an important milestone in the analysis of the molecular relationship between Wolbachia and the filarial nematode host, opening doors for new drug development for the eradication of Wolbachia – and subsequently – filarial nematodes.

METHODS

Yeast strain and plasmid constructions

All microscopy-based experiments were performed with derivatives of the yeast strain SEY6210 (MATa ura3–52 leu2–3, 112 his3-Δ100 trp1-Δ901 lys2–801 suc2-Δ9). For the growth experiments performed in Figs 2 and 7, yeast strains were derivatives of BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0). Any expression studies utilizing β-estradiol for the induction of GAL1 promoters required transforming relevant yeast strains with linearized pAGL (a gift from Dr. Daniel Gottschling, University of Washington), thus introducing the gene encoding for the Gal4-estrogen receptor-VP16 (GEV) chimeric protein into the leu2 locus [105].

To create a high-copy, constitutive yeast expression vector for wBm0152, we replaced the GAL1 promoter contained in pYES2/NT A (ThermoFisher Scientific) with the strong, constitutive TDH3 promoter. The TDH3 promoter was amplified from the p413-GPD plasmid [106] using the primer pair YTDH F and YTDHR (Table S1), each containing 30-bp of homology upstream and downstream of the GAL1 promoter in pYES2/NT A. The resultant amplicon was co-transformed into BY4742 with pYES2/NT A that had previously been linearized with PvuII, using standard lithium acetate techniques [107]. Gap-repaired plasmids were selected on CSM medium lacking uracil. To generate the pYES_TDH3-wBm0152 plasmid, the pYES2-wBm0152 plasmid [19] was digested with PmeI and HindIII, the wBm0152-containing insert was gel purified and co-transformed with pYES-TDH3 previously linearized with BamHI, as above.

To create a high-copy, inducible yeast expression vector for wBm0152 that does not rely on activation of the GAL1 promoter, we replaced the GAL1 promoter from pYES2/NT A with the copper-inducible CUP1 promoter [108]. To avoid repetitive DNA sequences contained in the duplicated S. cerevisiae CUP1–1 CUP1–2 locus, we first amplified the upstream sequence of CUP1–1 with the primer pair CUP F and CUP R. The resultant amplicon was used as template with the primer pair YCUP F and YCUP R; the amplicon from this second reaction was then co-transformed into yeast with pYES2/NT A previously linearized with PvuII. To create the wBm0152 expression construct in this background, pYES_CUP1 was linearized with BamHI and co-transformed with the PmeI-HindIII wBm0152-containing insert from above.

The copper-inducible pYES_CUP1-wBm0152 split Venus yeast expression plasmids used for the biomolecular fluorescence complementation assays were created by amplifying either the VN or VC domain from the relevant plasmid template using the indicated primer pairs containing 30-bp homology to the C-terminus of wBm0152 and the pYES plasmid backbone (VN: pFA6a-VN-HIS3 with 0152VCN F and 0152VN R; VC: pFA6a-VC-TRP1 with 0152VCN F and 0152VC R) [62]. The resultant amplicons were separately co-transformed into yeast with PmeI-linearized pYES_CUP1-wBm0152 to create the final constructs via gap repair. To generate the pYES_CUP1-VPS2-VN split Venus yeast expression plasmid, the VPS2-VN fusion was amplified off the genome of the SEY6210 VPS2-VN yeast strain[REF], using the primer pair VPS2VN F and VPS2VN R. The resultant amplicon was co-transformed into yeast with BamHI-linearized pYES_CUP1-wBm0152 to create the final construct via gap repair.

The Brugia malayi VPS2 homolog was identified via blastp [109] using the yeast Did4p/Vps2p sequence and restricting the search to Brugia malayi. The closest match was Bm6583b (GenBank accession: VIO93588) and the transcript cDNA encoding for this ORF was retrieved from the WormBase web site, http://www.wormbase.org, release WS296, date Apr 30 2025 [110]. This sequence was codon-optimized for Saccharomyces cerevisiae expression using the Codon Optimization Tool (IDT^™ Technologies). When generating the Bm6583b gBlock^®(IDT^™ Technologies), we also appended the 300 base pairs immediately upstream of the yeast VPS2 ORF, containing the endogenous promoter region. Prior to the Bm6583b stop codon, we added sequences encoding for the c-Myc epitope for detection via immunoblot, a BamHI restriction site immediately after the stop codon, and the 300 bases immediately downstream of the yeast VPS2 ORF, containing the endogenous terminator region. Finally, we appended 30-bp sequences to the 5’ and 3’ ends of this sequence, providing homology flanking the MCS of the yeast plasmid pRS414 for cloning via gap repair in yeast (Table S1). This gBlock^® was co-transformed into yeast with pRS414 previously linearized with BamHI, creating pRS414-Bm6583-myc. To generate the Bm6583 expression plasmid used for the split Venus bimolecular fluorescence complementation study, the Venus VN domain was amplified from pFA6a-VN-HIS3 with the primer pair BM6583 VN F and BM6583 VN R (Table S1) and co-transformed into yeast with pRS414-Bm6583-myc plasmid previously digested with BamHI.

All generated plasmid constructs were confirmed through whole plasmid sequencing by Plasmidsaurus using Oxford Nanopore Technology with custom analysis and annotation.

RapIDeg protein turnover assay

To visualize the ESCRT-dependent turnover of Fth1p in response to rapamycin, we modified the RapIDeg yeast strain Fth1-GFP-FKBP 3x Ub (SEY6210.1 tor1–1 fpr1Δ::NATMX6 pRS305-pGPD-FRB-3xUb::LEU2 FTH1-GFP-2xFKBP::HIS3, a gift from Dr. Ming Li, University of Michigan) [48] to express GAL promoters with β-estradiol. First, we amplified the HPHMX6 gene from pAG32 [111] using primer pair pr29 and pr32 (Table S1); this amplicon contains homology to the NATMX6 cassette inserted into the FPR1 gene. The RapIDeg strain was transformed with this amplicon and hygromycin-resistant colonies were selected. This strain was then screened for nourseothricin sensitivity, confirming the replacement of the NATMX6 cassette with HPHMX6. The resultant strain was then transformed with the linear pAGL, as above, creating the β-estradiol-responsive RapIDeg strain.

RapIDeg strains harboring either the galactose-inducible pYES-wBm0152 plasmid or empty vector control were grown to saturation in selective medium at 30° C for 16h. Strains were subcultured into 10 mL fresh selective medium and grown to mid-log for 4h at 30 °C. To induce the expression of wBm0152, 1 μM β-estradiol was added to each culture and incubated at 30° C with shaking for another 2h. After 2h, each culture was evenly split and either rapamycin or the DMSO vehicle control was added to 1 μg mL⁻¹ and cultures were outgrown for an additional 2h at 30°. 1 mL aliquots were removed from each condition, yeast cells were harvested by centrifugation (7000 × g, 1 min), and visualized via fluorescence microscopy. From the remaining culture, 3.0 OD₆₀₀ units were harvested via centrifugation and total cellular proteins were extracted from cell pellets [112] for immunoblotting experiments.

Microscopy

Cells to be visualized were grown overnight at 30° C in selective media, subcultured in fresh media with or without induction agent(s), and grown for an additional 5 hours. Cells were harvested via centrifugation, washed with sterile water, and suspended in 50 μL water. Cell suspensions were mounted to slides pre-treated with a 1:1 mixture of polylysine (10% w/v):concanavalin A (2 mg mL⁻¹) solution. Cells were visualized using a Nikon Ti-U fluorescence microscope, and images were processed using the Fiji software package (ImageJ2, v2.14.0/1.54f) [113, 114]. Colocalization analyses were performed using the Coloc2 plugin contained within the Fiji package.

Thin section electron microscopy

Saccharomyces cerevisiae strain SEY6210 GAL⁺ harboring either a vector control or the pYES-wBm0152 expression plasmid were grown in selective media (CSM lacking uracil) for 18h at 30° C with shaking. Cells were harvested via centrifugation, washed with sterile water, then diluted 1:10 into fresh CSM-ura containing 2% galactose to induce wBm0152 expression. Cells were harvested at log phase, collected on 0.45μ filter discs (Millipore) via vacuum, loaded into 0.25 mm aluminum hats, and high-pressure frozen in the Wolwend high-pressure freezing machine (HPF) as previously described [56]. Freeze substitution was carried out with media containing 0.1% uranyl acetate and 0.25% gluteraldehyde in anhydrous acetone [115] in a Leica AFS (Automated Freeze Substitution, Vienna, Austria). Samples were then embedded in Lowicryl HM20, and UV-polymerized at −60C over 2 weeks (Polysciences, Warrington, PA, A. Staehelin, personal communication, Wemmer et al JCB 2011). A Leica Ultra-Microtome was used to cut 80 nm serial thin sections and 200 nm serial semi-thick sections from polymerized blocks and sections were collected onto 1% formvar films adhered to rhodium-plated copper grids (Leica Biosystems, Nussloch, Germany, and Electron Microscopy Sciences). Thin sections were imaged using a FEI Tecnai T12 “Spirit” electron microscope (120 kV, AMT 2 × 2 k CCD).

Semi-thick section electron tomography

Semi-thick sections of polymerized blocks (~200 nm) were applied to grids labeled on both sides with fiduciary 15 nm colloidal gold (British Biocell International). Dual-axis tilt series were collected of the samples from +/− 60 degrees with 1-degree increments at 300kV using SerialEM [116] on a Tecnai 30 FEG electron microscope (FEI-Company, Eindhoven, the Netherlands). Tilt series were regularly imaged at 23,000X using SerialEM [117], with 2X binning when recording on a 4×4K CCD camera (Gatan, Inc., Abingdon, UK) creating a 2×2K image with a pixel size of 1.02 nm. Tomograms were constructed and modeled with the IMOD software package (3dmod 4.0.11, [118]). All membrane structures were identified with areas of interest modeled (ER-LD compartments, small MVBs, tubular MVBs). IMODINFO provided surface area and volume data of contour models, diameters and distances were measured from the outer membrane leaflets at optimal X Y orientations in the tomograms at 50 nm intervals using 3DMOD.