SLC26A9 is selected for endoplasmic reticulum associated degradation (ERAD) via Hsp70-dependent targeting of the soluble STAS domain

Abstract

SLC26A9, a member of the solute carrier protein family, transports chloride ions across various epithelia. SLC26A9 also associates with other ion channels and transporters linked to human health, and in some cases these heterotypic interactions are essential to support the biogenesis of both proteins. Therefore, understanding how this complex membrane protein is initially folded might provide new therapeutic strategies to overcome deficits in the function of SLC26A9 partners, one of which is associated with Cystic Fibrosis. To this end, we developed a novel yeast expression system for SLC26A9. This facile system has been used extensively with other ion channels and transporters to screen for factors that oversee protein folding checkpoints. As commonly observed for other channels and transporters, we first noted that a substantial fraction of SLC26A9 is targeted for endoplasmic reticulum associated degradation (ERAD), which destroys folding-compromised proteins in the early secretory pathway. We next discovered that ERAD selection requires the Hsp70 chaperone, which can play a vital role in ERAD substrate selection. We then created SLC26A9 mutants and found that the transmembrane-rich domain of SLC26A9 was quite stable, whereas the soluble cytosolic STAS domain was responsible for Hsp70-dependent ERAD. To support data obtained in the yeast model, we were able to recapitulate Hsp70-facilitated ERAD of the STAS domain in human tissue culture cells. These results indicate that a critical barrier to nascent membrane protein folding can reside within a specific soluble domain, one that is monitored by components associated with the ERAD machinery.

Introduction

The regulation of salt and fluid balance across membranes is an essential function of epithelial cells, and to this end numerous ion channels and transporters are required to maintain cellular homeostasis. Consequently, mutations that interfere with the folding, transport, or activity of these channels and transporters to the cell surface can disrupt electrolyte balance and lead to disease [1-3]. For example, the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR), is an epithelial chloride channel expressed most prominently in the lungs, the gastrointestinal tract, and sweat glands. To date, there are >2000 identified variants and ~300 disease-associated mutations which impact protein expression, channel activity, protein folding in the endoplasmic reticulum (ER), and the transport and stability of CFTR at the plasma membrane [4-6]. By far, the most common cause of CF arises from the F508del allele. The resulting mutant protein, which lacks a phenylalanine within the first nucleotide binding domain in CFTR, misfolds in the ER and fails to traffic beyond this organelle due to aggressive targeting to the ER associated degradation (ERAD) pathway [7-12]. While significant progress has led to the discovery of drugs that alleviate many CF symptoms [13,14], there remains a need for better and more widespread clinical treatments to correct defects associated with this allele as well as rare alleles [15,16]. Unfortunately, many of these disease-associated alleles are refractory to existing treatments.

CFTR directly influences chloride and water flux across epithelial membranes, but other ion transporting proteins similarly contribute to salt and fluid dynamics. Because these alternative chloride channels may function like CFTR, there is also a focus on identifying drugs that regulate the alternate transporters, thereby providing a means to rescue chloride secretion defects associated with mutant forms of CFTR, especially those that are refractory to current treatments [17,18]. Interestingly, some of these channels and transporters interact with CFTR.

One notable group of ion channels that associate with and co-regulate CFTR is the SLC26A proteins. The SLC26A family consists of 11 members that transport an array of mono- and divalent anions, such as Cl⁻, OH⁻, NO3⁻, HCO3⁻, and SO4⁻ [19-21]. SLC26A proteins have a modular structure with a region enriched for membrane spanning helices followed by a large intracellular regulatory region, known as the STAS (sulfate transporter and anti-sigma factor antagonist) domain [20,22,23]. Early models suggested that SLC26A members contain 12 transmembrane helices (see Supplementary Figure S1 for a schematic of this topology for one SLC26A family member). More recent evidence suggests the presence of 14 transmembrane spanning segments [24-26]. Other data indicate that the ~34 kDa STAS domain mediates the dimerization of SLC26A proteins, and serves as a point of interaction with other proteins [26]. The extreme C-terminus of some family members also contains a small region — a PDZ motif — which in these cases can bridge SLC26A with protein partners, facilitates interactions with the STAS domain, and regulates the trafficking and degradation of SLC26 [20,27-29]. Interestingly, two members of the SLC26A family with chloride-bicarbonate exchange activity, SLC26A3 and SLC26A6, bind the phosphorylated R-domain in CFTR through the STAS domain, which in turn increases both CFTR-dependent chloride flux as well as SLC26A anion exchange activity [30].

Yet another SLC26A family member, SLC26A9, is also emerging as an important CFTR regulator and potential drug target [30-38]. SLC26A9 acts as a constitutive chloride transporter; early studies suggested it could also exhibit anion exchange activity, like SLC26A3 and SLC26A6 [32,39-43], but recent structural analyses have failed to identify anion exchange behavior [26,44]. Moreover, polymorphisms in SLC26A9 are linked to several CF-associated phenotypes [45-48], and the response to drugs that potentiate the gating of a disease-causing allele in CFTR (G551D) is modulated by non-coding polymorphisms in SLC26A9 [49,50]. In addition, when SLC26A9 activity is enhanced in lung epithelia, this helps alleviate mucus build-up in response to inflammation [51], which is frequently seen in CF patients [52,53]. A significant body of work also indicates that both SLC26A9 and CFTR contribute to basal and stimulated chloride currents in human bronchial epithelia (HBE), but SLC26A9 currents are absent from CF patient tissue [4,30,31,39,54-57]. Yet, other data indicate SLC26A9 exhibits chloride transport activity even in the absence of CFTR expression [33,40,43]. Indeed, when SLC26A9 is co-expressed with F508del CFTR in model cell types, SLC26A9 currents are absent [32]. Moreover, when the biogenesis and activity of F508del CFTR are partially rescued by low temperature or by a protein folding corrector compound, cell surface expression of SLC26A9 also increases and SLC26A9 currents are again detected [35]. Although SLC26A9 is also turned over by the ERAD pathway when co-expressed with F508del CFTR, transfection of wild-type CFTR into F508del CFTR-expressing cells increases SLC26A9 trafficking to the plasma membrane [38]. Together, these results reveal that F508del CFTR prevents SLC26A9 trafficking to the plasma membrane and increases the degradation of ER-retained SLC26A9.

Although a significant body of work indicates that SLC26A9 might be modulated to overcome symptoms associated with CF, cellular chaperones that facilitate SLC26A9 maturation or degradation in the early secretory pathway are unknown. This undertaking is vital as drugs that target many of these chaperones are being developed [58-60]. To identify factors that contribute to the biogenesis of other ion channels and transporters, we have made extensive use of yeast expression systems. In all cases, the identified chaperones functioned similarly in higher cells [61-68]. We now report that a significant portion of SLC26A9 is targeted for ERAD in yeast, as in mammalian cells, and discovered that substrate ubiquitination and degradation are mediated by the cytoplasmic Hsp70 molecular chaperone. A dissection of the SLC26A9 protein next revealed that the STAS domain is highly unstable, and like the full-length protein, its turnover required Hsp70 activity. Protein stability measurements in HEK293 cells support an analogous requirement for Hsp70 during STAS domain degradation in mammals. Overall, this study defines a specific folding barrier during SLC26A9 maturation and provides a platform to dissect the contributions of SLC26A9 polymorphisms on protein stability and selection for chaperone-dependent ERAD.

Materials and methods

Yeast strains, plasmid construction, and protein expression

All yeast strains used in this study are listed in Supplementary Table S1. Plasmids containing the coding sequences of the two full length forms of SLC26A9 and the free STAS domain were obtained in a pcDNA3.1 vector [32,35]. Yeast expression vectors were constructed by PCR amplification of the SLC26A9 coding sequence, incorporating restriction sites for SpeI and Cla1 at the 5′ and 3′ ends of the gene, respectively, and with the addition of a c-myc tag (EQKLISEEDL) at the N-terminus of the protein. The amplified DNA was digested with the SpeI and Cla1 enzymes, gel purified, and ligated into the multicopy histidine-selectable yeast expression vector pRS423MET [69], which had been digested with the same enzymes. This created plasmids 423MET-A9-1 and 423MET-A9-2, in which SLC26A9 expression is controlled by the methionine-repressible MET25 promoter. A yeast expression vector for the STAS domain was similarly created by PCR amplifying an N-terminal c-myc tagged coding sequence starting at amino acid 505 of SLC26A9 isoform 1 and incorporating Spe1 and Cla1 restriction sites into the primers for ligation into pRS423MET. In turn, the SLC26A9 construct truncated after the transmembrane region (Δstas) was created by introducing a stop codon into the sequence of full length SLC26A9 isoform 1 at lysine 518 using the Quick Change II Site Directed Mutagenesis Kit (Agilent) and primer sequences 5′-gacatttatgtgaatccctagacctataatagggccc-3′ and 5′-gggccctattataggtctagggattcacataaatgtc-3′. All candidate plasmids were screened by restriction digest and verified by DNA sequencing (GeneWiz). To utilize methionine auxotrophic yeast strains, such as BY4741 and hrd1Δ asi1Δ with our SLC26A9 MET25 promoter expression system, we restored the MET15 locus in these strains to allow for growth in -Met media. To this end, the MET15 locus [70] was first PCR amplified from BY4742 genomic DNA using primers 5′-tagctctcattattttttgctttttctcttgaggtcaca-3′ and 5′-ttgttgaatgttgagcaagttaacatcttataggacatat-3′. The amplified DNA was introduced into the yeast genome by homologous recombination and the cells were selected on SC media lacking methionine (SC-Met). Colonies were re-streaked onto SC-Met plates two additional times to ensure MET15 integration.

To express each protein, the indicated vectors were introduced into yeast using lithium acetate transformation [71] and selected on synthetic complete media lacking histidine but supplemented with methionine at a final concentration of 0.5 mM (SC-His + Met). Yeast transformed with the indicated expression plasmid(s) were then grown in liquid media overnight from single colonies in the appropriate selective media at room temperature. The cells were washed in sterile water, resuspended at an optical density (OD₆₀₀) of 0.5 in SC-His-Met to induce protein expression from the Met-repressible promotor, and growth was continued for 2–3 h at 30°C with shaking. These cultures were then transferred to a shaking water bath at 30°C or shifted for the indicated times to 37°C for assays in temperature sensitive mutant strains.

Measurements of protein stability, ubiquitination, and membrane association in yeast

Protein translation was arrested by addition of cycloheximide to a final concentration of 100 μg/ml, and equal aliquots of cells were removed at the indicated times. The assay was performed at 30°C unless otherwise indicated (i.e. for the temperature sensitive yeast strains). In this case, cells were maintained at the elevated temperature throughout the chase. In all cases, cells were collected by centrifugation at 13 000 RPM in a Sorvall microcentrifuge for 1 min at 4°C, quick frozen in liquid nitrogen, and the cell pellets were stored at −80°C. To prepare protein fractions, the cell pellets were thawed, and total protein was extracted by trichloroacetic acid (TCA) precipitation [72]. Protein samples were suspended in sample buffer and resolved by SDS–PAGE.

Following SDS–PAGE, proteins were transferred to nitrocellulose membranes and a quantitative Western Blot analysis was performed. The quality of the load and transfer was assessed by Ponceau-S staining of the membrane for every blot before antibody incubation. SLC26A9 was detected with rabbit polyclonal anti-myc (Santa Cruz A-14) and anti-rabbit HRP-conjugated secondary antibody (CST). Rabbit anti–glucose-6-phosphate dehydrogenase (G6PD) (A9521; Sigma–Aldrich) was used as a loading control. The blots were then incubated with the ECL reagent (Prometheus ProSignal), visualized on a Bio-Rad imager, and quantified with Image-J software. All statistical analyses were performed using GraphPad Prism 8 (GraphPad Software, Inc., CA, US). The amount of protein at time 0 was set to 100%, and the remaining time points were expressed as a percent of the starting material. Only data with a P < 0.05 was considered significant.

To measure protein ubiquitination [73], cells engineered for STAS domain expression were transformed with pLHP462, an HA-tagged ubiquitin expression plasmid under the control of the copper inducible CUP1 promoter [74]. The expression of the c-myc tagged STAS domain was induced as described above, and after 1 h, Ub-HA expression was induced by the addition of CuSO4 to a final concentration of 100 μM. A 40 ml culture of the pdr5Δ yeast strain was then split and treated with MG-132 at a final concentration of 100 μM or with an equal volume of DMSO at 30°C in a shaking water bath for 1 h. SSA1 or ssa1–45 yeast were instead shifted to 37°C in a shaking water bath for 30 min before cells were treated with 10 mM (final concentration) NaN₃ and quickly cooled in an ice/water bath. In all cases, cells were centrifuged in a tabletop clinical centrifuge at 4000 RPM, and the resulting cell pellets were stored at −80°C. Next, the pellets were thawed and resuspended in 50 mM Tris, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton-X100, 1% SDS, 10 mM N-ethylmaleimide, 1 mM PMSF, 1 μg/ml leupeptin, and 0.5 μg/ml pepstatin A, and the cells were disrupted by agitation with glass beads and debris was removed by centrifugation at 10 000g for 1 min. Clarified lysate was diluted with resuspension buffer lacking SDS to a final concentration of 0.2% SDS. Anti-myc conjugated to agarose (Pierce) pre-equilibrated in the same buffer was then added to the clarified lysate, and the mixture was rotated overnight at 4°C. The beads were subsequently collected by centrifugation at 500g, the supernatant was removed by pipetting, and the beads were washed three times in the same buffer. The bound protein was removed from the beads by incubation in SDS–PAGE sample buffer at 42°C for 30 min. Following SDS–PAGE, an immunoblot analysis was performed using anti-c-myc antibody to detect the STAS domain (see above) or HRP-conjugated anti-HA antibody (Roche) to detect ubiquitinated protein.

To determine the degree to which a specific protein associated with organellar membranes, cells expressing each SLC26A9 construct were harvested by centrifugation. Cell lysate was prepared by resuspending cells in lysis buffer (0.1 M sorbitol, 50 mM potassium acetate (KOAc), 2 mM EDTA, 20 mM HEPES-NaOH, pH7.4) in the presence of 1 mM PMSF, 1 μg/ml leupeptin, and 0.5 μg/ml pepstatin A, and the cells were disruption by agitation with glass beads. Debris was removed by centrifugation for 2 min at 2000 RPM in a tabletop clinical centrifuge. The resulting supernatant was transferred to a clean tube and the membranes pelleted by centrifugation at 20 000g for 15 min. The pellet fraction (P1) was resuspended in lysis buffer, split into separate tubes and pelleted again at 20 000g for 15 min. Association was detected using specific treatments as described [75]. Briefly, the P1 fraction was suspended in either lysis buffer or lysis buffer containing 6 M urea and the tubes were rotated at 4°C for 1 h. After treatment, the solutions were centrifuged at 25 000g and the supernatant (S2) was removed from the pellet (P2). Each of these fractions was analyzed by SDS–PAGE and western blotting. Controls included Pma1 (an integral membrane protein), Pdi1(a peripheral membrane-associated protein in the ER), and G6PD (a soluble protein).

Mammalian cell protein stability assays

HEK293 cell cultures were maintained in DMEM (Sigma–Aldrich) supplemented with 10% FBS (Seradigm premium grade) at 37°C in a 5% CO₂ humidified atmosphere. At 24 h prior to transfection, 0.6 × 10⁶ cells were plated into wells of a six-well poly-l-lysine coated plate (Greiner BioOne) with 3 ml DMEM containing 10% FBS per well. Transfections with pcDNA3.1 engineered for the expression of the STAS domain [76] included 2 μg of plasmid DNA and 6 μl of Lipofectamine™ 2000 (Thermo Fisher Scientific #11668) per well, suspended in a total volume of 500 μl Opti-MEM (Thermo Fisher Scientific #31985) per well, according to the manufacturer’s protocol. The media was then changed after 6 h. Experiments were conducted 24 h after transfection.

To perform cycloheximide chase assays, the cells were pre-treated with 10 μM MG-132 (EMD Millipore) for 2 h, the media was removed and replaced with DMEM containing 10% FBS as well as 20 μg/ml cycloheximide (Sigma), and then either 10 μM MG-132, 25 μM VER155008 (Tocris), or an equivalent volume of DMSO was added. At the indicated times, the media was aspirated from the wells, and the cells were lysed in 300 μl of ice-cold TNT Buffer (50 mM Tris, pH 7.2, 150 mM NaCl, 1% Triton X-100) with a complete protease inhibitor cocktail (Roche). During this time, plates were incubated on ice for 30 min with gentle rocking, and the solution was removed and centrifuged at ~16 000g for 10 min at 4°C. The cleared lysate was diluted into SDS Sample Buffer. Total protein was resolved on 10% polyacrylamide gels by SDS–PAGE and transferred to nitrocellulose. To assess protein loading, blots were stained with the REVERT total protein staining kit (Licor) and imaged in the 700 channel of a Licor Odyssey CLX. The stain was then reversed, the blots were blocked in TBST-milk solution for 1 h, and the myc-tagged STAS protein was detected with clone 9E10 (Biolegend) at a 1 : 250 dilution. Bound antibodies were decorated for 2 h with goat anti-mouse IRdye800 antibody (Licor), diluted 1 : 15 000, and imaged in the 800 channel of a Licor Odyssey CLX.

For experiments in which the steady-state levels of the STAS domain were examined in HEK293 cells, the cells were transfected as above and then treated with VER155008 or an equivalent volume of DMSO at the final concentrations indicated above. The cells were then incubated for 2 h and processed for immunoblotting, as above, except that a 1 : 500 antibody dilution of the anti-myc antibody was used.

Results

To determine whether SLC26A9 is prone to misfolding in the ER — as observed for other polytopic membrane ion channels [77] — and to identify the factors that control its stability and degradation, we developed a yeast SLC26A9 expression system. The yeast model has been used extensively to identify chaperones and chaperone-like proteins that regulate the stability of human ion channels and transporters, and in nearly all cases the results uncovered from this organism have been recapitulated in human cells [61-68].

To this end, we first constructed a regulated expression system for two SLC26A9 isoforms in yeast. The isoforms differ by 96 amino acids [36] (Supplementary Figure S1), and to date their relative stabilities in the ER have not been investigated. The expression of these proteins was under the control of the methionine repressible MET25 promoter, which allowed for SLC26A9 expression to be repressed on media containing methionine and then quickly activated by resuspending cells in methionine-free media. Next, each plasmid was introduced into wild-type yeast, and a time course after incubation in methionine-free media indicated rapid expression of both isoforms after 1 h with peak levels reached by 4 h (Supplementary Figure S2A). To measure the stability of the SLC26A9 isoforms in yeast, we then conducted a cycloheximide chase assay after protein expression had been derepressed for 2 h. As shown in Supplementary Figure S2B, SLC26A9 — like wild-type and F508del CFTR in yeast [61-63] — was unstable and degraded over time. Furthermore, the degradation rates of isoforms 1 and 2 were indistinguishable in this strain or in other strains examined (Supplementary Figure S2B and data not shown). Therefore, we conducted all experiments in this study using the more commonly described isoform 1 [78].

Since SLC26A9 was rapidly degraded after a translation shut-off, we investigated if degradation was via ERAD or an alternative post-ER degradation pathway [79]. ERAD substrates are targeted for degradation via the ubiquitin-proteasome pathway [80-82], so we first asked whether SLC26A9 degradation was proteasome-dependent. To allow robust proteasome inhibition with MG-132, we utilized a PDR5-deficient yeast strain as it lacks a drug efflux pump [83]. As shown in Figure 1A, MG-132 treatment resulted in significant SLC26A9 stabilization over the time course of the assay. We next performed cycloheximide chase assays in a yeast strain deficient in the two major ER-associated E3 ubiquitin ligases, Hrd1 and Doa10 [84,85] (Figure 1B). The absence of the genes encoding HRD1 and DOA10 also resulted in significant SLC26A9 stabilization, strongly suggesting that the ERAD pathway eliminates SLC26A9. Interestingly, deletion of either E3 ligase only subtly stabilized the protein (data not shown), suggesting that both enzymes work together to most efficiently target this complex substrate for degradation.

Figure 1. SLC26A9 is an ERAD substrate in yeast.

The stability of SLC26A9 in the indicated wild-type and mutant strains was determined in cycloheximide chase assays. (A) SLC26A9 was expressed in a pdr5Δ strain, and cells were pre-treated for 30 min with either the vehicle (DMSO) or with 100 μM MG-132, as indicated. (B) SLC26A9 degradation was measured in wild-type and hrd1Δ doa10Δ yeast. (C) SLC26A9 degradation was measured in wild-type and cdc48-2 yeast after cells were pre-shifted to 38°C for 2 h. (D) The vacuolar dependence of SLC26A9 in the wild-type and a pep4Δ mutant strain was examined at 30°C. Please note that the background strains used in parts (A) and (D) were distinct, and thus the relative rates of ERAD differ. In all experiments closed circles represent the wild-type yeast and open circles represent the corresponding mutant yeast strain or cells treated with MG-132, as indicated. A representative immunoblot corresponding to the quantified data is also shown below each graph. The data represent the means ± SEM for 3–6 independent experiments. *P < 0.05.

Ubiquitinated ERAD substrates must be extracted or ‘retrotranslocated’ from the ER prior to or concomitant with proteasome-dependent degradation [86,87]. Therefore, we examined whether Cdc48, which drives the ATP-dependent retrotranslocation of ERAD substrates, was also required for SLC26A9 degradation. As anticipated, inactivation of a thermosensitive CDC48 allele led to profound stabilization of SLC26A9 in a cycloheximide chase assay (Figure 1C). Although these data further support the role of ERAD in SLC26A9 turnover, we also excluded another major location of protein degradation in yeast, the vacuole (which is the yeast lysosome equivalent). Yeast lacking Pep4, which express <5% of residual vacuolar protease activity [88], still efficiently degraded SLC26A9 (Figure 1D). This is in stark contrast with the dramatic stabilization of a known vacuolar substrate SZ* in pep4Δ yeast ([89] and see for example Supplementary Figure S4B). Taken together, we conclude that ER quality control accounts for the degradation of immature, ER-resident SLC26A9 species in yeast, as suggested by studies performed previously in higher cells [38].

Cytoplasmic Hsp70 molecular chaperones bind misfolded proteins and can contribute to the recognition of misfolded substrates in the ER, which are then delivered to ubiquitin ligases such as Doa10 [90,91]. For substrates that deposit misfolded domains in the ER lumen, a lumenal Hsp70 (BiP) similarly recognizes misfolded proteins in this compartment and prevents substrate aggregation, which favors retrotranslocation [92,93]. To determine if the cytoplasmic and/or lumenal Hsp70 chaperones facilitate SLC26A9 degradation, cycloheximide chases were performed in yeast expressing a temperature-sensitive mutant form of the major cytoplasmic Hsp70 that lacks the other primary cytoplasmic Hsp70s (ssa1–45, ssa2, ssa3, ssa4) [94], or that contain an ERAD-specific BiP (kar2-1) mutant [95]. Upon a shift to 37°C, SLC26A9 was significantly stabilized in the ssa1–45 strain (Figure 2A). Importantly, the magnitude of stabilization was similar to that seen in the ERAD-deficient strains (Figure 1A-C). In contrast, the degradation of SLC26A9 in kar2-1 yeast was unaffected compared with the wild-type strain, indicating that this ER lumenal Hsp70 most likely does not select SLC26A9 for degradation (Figure 2B). More generally, these data suggest that the misfolded domain that contributes to the selection of SLC26A9 for ERAD resides in the cytosol, i.e. the ~34 kDa STAS domain (Supplementary Figure S1).

Figure 2. The activity of the cytosolic Hsp70 chaperone is required for SLC26A9 degradation.

The stability of SLC26A9 in the indicated wild-type and mutant strains was determined in cycloheximide chase assays. (A) SLC26A9 degradation in SSA1 yeast and ssa1–45 (the Hsp70 mutant) strain was measured after cells were pre-shifted to 37°C for 30 min. (B) SLC26A9 degradation in wild-type and kar2-1 (BiP) mutant yeast was measured at 30°C. In all experiments, closed circles represent the wild-type yeast and open circles represent the corresponding mutant yeast strain. A representative immunoblot corresponding to the quantified data is also shown below each graph. The data represent the means ± SEM for 3–6 independent experiments. *P < 0.05.

To more definitively determine which region of the protein is recognized by the ERAD machinery, we constructed several versions of the protein. These included full length SLC26A9, a form of SLC26A9 that terminates at amino acid 517 which is before the start of the STAS domain at amino acid 520 (‘Δstas’), and a membrane domain-free version of the STAS domain beginning immediately after the transmembrane region at amino acid 505 (Supplementary Figure S1B). Each species also contained an N-terminal myc epitope tag for detection. We then performed cycloheximide chases to determine the relative stabilities of each protein.

We first noted that the transmembrane-only Δstas species was degraded much slower than full length SLC26A9 in wild-type yeast, although the modest level of measured degradation was reduced in a strain lacking Hrd1 and Doa10 E3 ligases (Figure 3A), as observed for the full-length protein (Figure 1B). Second, and as noted above, most ERAD substrates with misfolded cytosolic domains are degraded in an Hsp70-dependent manner (or in yeast, in an Ssa1-dependent manner) [90,91]. Therefore, we also examined whether disabling this chaperone through the use of the thermosensitive ssa1–45 mutant slowed Δstas degradation. As anticipated based on the lack of a misfolded cytoplasmic region, Δstas turnover was Ssa1-independent (Figure 3B), again suggesting that the STAS domain was responsible for Hsp70-dependent degradation. Third, and consistent with our hypothesis, we discovered that the turnover of the isolated STAS domain was almost completely blocked when the yeast Hsp70, Ssa1, was disabled (Figure 4A). Therefore, Hsp70 appears to recognize and target full-length SLC26A9 by virtue of the cytosolic STAS domain. Finally, as a control, we established that the STAS domain was also strongly stabilized by treating cells with MG-132 (Figure 4B).

Figure 3. The ERAD of the SLC26A9 transmembrane domain, Δstas, is Hsp70-independent.

The stability of the Δstas construct in the indicated wild-type and mutant strains was determined by cycloheximide chase assay in (A) wild-type and hrd1Δdoa10Δ mutant yeast at 30°C, and (B) SSA1 and ssa1–45 strains after cells had been pre-shifted to 37°C for 30 min. In all experiments, closed circles represent the wild-type yeast and open circles represent the corresponding mutant strain. A representative immunoblot corresponding to the quantified data is also shown below each graph. The data represent the means ± SEM for three independent experiments. *P < 0.05.

Figure 4. The degradation of the STAS domain is cytosolic Hsp70- and proteasome-dependent.

The stability of the STAS domain in the indicated wild-type and mutant strains was determined by cycloheximide chase assay in (A) SSA1 and ssa1–45 strains after pre-shift to 37°C for 30 min, and (B) pdr5Δ cells that were either pre-treated with DMSO or with 100 μM MG-132, as indicated, for 30 min. In all experiments closed circles represent the wild-type yeast (or the DMSO treated cells) and open circles represent the corresponding mutant yeast strain (or MG-132). A representative immunoblot corresponding to the quantified data is also shown below each graph. The data represent the means ± SEM for 3–6 independent experiments. *P < 0.05.

In contrast to full-length SLC26A9 (Figure 1B), the STAS domain was degraded at the same rate when Hrd1 and Doa10 were lacking (Figure 5A). These results implied that STAS proteolysis required an alternate E3 ubiquitin ligase, an outcome that is frequently observed when the degradation of misfolded cytoplasmic proteins is examined [75,96,97]. Therefore, we first examined STAS domain stability when two cytoplasmic quality control E3s, San1 and Ubr1 [96,97], were absent, but again the protein was degraded proficiently (Figure 5B). As a control, we confirmed that an established ERAD substrate, CPY*, as well as a cytoplasmic misfolded protein, NBD2* [75,98], were stabilized in hrd1Δ doa10Δ and san1Δ ubr1Δ yeast, respectively (Supplementary Figure S3A,B). STAS domain degradation was also unaffected in pep4Δ yeast (Supplementary Figure S4A).

Figure 5. Analysis of the E3 ubiquitin ligase requirements for the degradation of the STAS domain.

The stability of the STAS domain in the indicated wild-type and mutant strains was determined by cycloheximide chase assay in (A) wild-type and hrd1Δ doa10Δ mutant yeast, (B) wild-type and ubr1Δ san1Δ mutant yeast, (C) wild-type and asi1Δ hrd1Δ mutant yeast, and (D) wild-type and rsp5-1 mutant yeast at 38°C after cells were pre-shifted to 38°C for 30 min. In all experiments closed circles represent the wild-type yeast (or the DMSO treated cells) and open circles represent the corresponding mutant yeast strain (or those treated with MG-132). A representative immunoblot corresponding to the quantified data is also shown below each graph. The data represent the means ± SEM for 3–6 independent experiments. *P < 0.05.

Some misfolded proteins enter the nucleus and their degradation is facilitated by the Asi1-containing ubiquitin ligase complex, which resides in the inner nuclear membrane, and Hrd1 can aid in the turnover of these substrates [99,100]. Therefore, we assessed STAS domain stability in a hrd1Δ asi1Δ strain. As a control, we first established that a model ERAD substrate, CPY*, was stabilized in this mutant (Supplementary Figure S3C). While the overall degradation of STAS was rapid, there was a small but statistically significant stabilization of STAS when both HRD1 and ASI1 were deleted (Figure 5C). Finally, we examined STAS domain turnover in strains containing a thermosensitive version of RSP5, which encodes a cytosolic ubiquitin ligase that plays critical roles in plasma membrane protein endocytosis [101,102]. However, Rsp5 has also been reported to contribute to other quality control phenomena [103-105]. When STAS degradation was analyzed in the temperature sensitive rsp5-1 mutant after a shift to the non-permissive 38°C, we again saw modest but statistically significant stabilization (Figure 5D). Interestingly, Rsp5 was reported to contribute to the degradation of heat-damaged proteins in yeast [106], suggesting that STAS might assume a non-native structure mimicking that attained by a thermolabile protein. Based on these collective results, it is likely that a compendium of E3 ubiquitin ligases, which includes Hrd1, Asi, and Rsp5, contributes promiscuously to the quality control of STAS, an outcome that is not uncommonly seen [107]. Future efforts will seek to better define the ligase ensemble that mediates STAS degradation.

To confirm that STAS domain degradation was ubiquitin-dependent, we immunoprecipitated the domain from cell lysates prepared from pdr5Δ yeast treated with either DMSO or MG-132 (to demonstrate ubiquitinated protein targeting to the proteasome), and from wild-type (SSA1) and ssa1–45 yeast after a shift to the non-permissive temperature (Figure 6). Analysis of STAS ubiquitination was facilitated by co-expression with HA epitope-tagged ubiquitin, which can be readily detected and quantified. As shown in Figure 6A, the STAS domain was ubiquitinated, and as expected for a proteasome-targeted substrate, the relative amount of ubiquitination increased upon MG-132 treatment. Because STAS degradation also required Hsp70 (Ssa1) function (Figure 4A), we then asked if Ssa1 activity was required for STAS ubiquitination. Therefore, STAS ubiquitination was measured at 26°C and 38°C both in the wild-type and in the ssa1–45 mutant strain. Although total protein ubiquitination increased in both strains after a shift to the higher temperature (Figure 6B, top panel, input lanes) relative to the amount of STAS in the load (see ‘Myc-STAS’), there was a modest reduction in the amount of STAS ubiquitination when the precipitated protein was examined in lysates prepared from the ssa1–45 mutant (Figure 6B, bottom). These results are consistent with Hsp70 aiding in the delivery of the STAS domain to the ubiquitination machinery. However, because STAS domain ubiquitination was not entirely blocked in the mutant, it is possible that Ssa1 also contributes to the delivery of ubiquitinated STAS adducts to the proteasome (but see Discussion).

Figure 6. STAS domain ubiquitination is Hsp70-dependent.

The STAS domain was co-expressed with HA epitope-tagged ubiquitin and the domain was immunoprecipitated with anti-myc agarose from whole cell yeast lysates prepared from the indicated strains under denaturing conditions. The ubiquitination of the STAS was determined after immunoprecipitation by blotting for HA-ubiquitin and normalizing the signal to the amount of myc-tagged STAS protein. (A) pdr5Δ yeast were pre-treated with either DMSO (‘D’) or with 100 μM MG-132 (‘MG’) for 1 h prior to cell lysis and the denaturing immunoprecipitation. The graph below the image represents the averaged ubiquitin signal for the data obtained from this experiment performed with three independent cultures (numbered 1, 2, and 3) and for each treatment relative to the amount of STAS protein. Data were normalized to the amount of ubiquitination in the DMSO control. (B) Wild-type SSA1 or ssa1–45 mutant yeast were grown at 26°C or 39°C for 30 min prior to cell lysis and denaturing immunoprecipitation. The input and immunoprecipitated material (IP) for a representative experiment is shown. Both light and dark exposures of the HA-ubiquitin blot are also shown. The average ubiquitin signal relative to the amount of immunoprecipitated STAS protein was calculated and then normalized to the signal in wild-type yeast at 26°C. The graph below the blots represents the means ± SEM of four independent experiments for the wild-type (white) and ssa1–45 mutant (grey) at 26°C, or wild-type (hatched), and ssa1–45 mutant (grey hatched) after cells had been shifted to 39°C.

Since Hsp70 supports the targeting of substrates to both the ERAD and cytoplasmic quality control pathways [72,87], we asked whether the STAS domain is completely cytosolic, is membrane-associated, or is unexpectedly integrated into the membrane. When total lysate was subjected to high-speed centrifugation analysis to pellet the membranes (Figure 7A), full length SLC26A9 and the Δstas protein were found associated with the initial pellet (P1) fraction, as was the integral membrane protein Pma1 (Figure 7B,C). Further treatment of P1 with buffer or with 6 M urea failed to remove significant amounts of these proteins from the final membrane fraction (P2), in-line with their expected membrane integration. In contrast, and as expected, the peripheral membrane-associated Pdi1 protein was partially liberated from the P2 membranes with urea. Also as expected, the soluble G6PD protein was found in the S1 fraction.

Figure 7. The STAS domain is membrane-associated.

Full length SLC26A9, the transmembrane-rich region of SLC26A9 (Δstas), or the STAS domain were expressed in wild-type yeast and whole cell lysates were prepared and fractionated by centrifugation. (A) A flow-chart of the fractionation protocol is shown. Clarified total cell lysate (T) was subjected to a high-speed spin to resolve a supernatant (S1) and pellet (P1) fraction. The (P1) fraction was resuspended in either lysis buffer or lysis buffer containing 6M urea. The treated P1 was then subjected to a second high-speed spin to generate a supernatant (S2) and pellet (P2) fraction. Aliquots of all fractions were then examined by SDS–PAGE and western blotting. Results are shown for the (B) SLC26A9 full-length protein, (C) the transmembrane-rich region (Δstas), and (D) the STAS domain. Controls included Pma1 (an integral membrane protein), Pdi1 (a peripheral membrane protein), and G6PD (a soluble protein), and their behavior in this analysis is shown for each experiment.

We then examined the behavior of the STAS domain. We found that this domain was membrane associated after the initial fractionation (Figure 7D, P1 fraction), but consistent with a lack of membrane integration the STAS domain was removed from the membranes with urea (Figure 7D, lane S2 ‘Urea’). Thus, STAS appears to be membrane-associated in yeast, perhaps consistent with its targeting by multiple ubiquitin ligases (see above), i.e. those that are both membrane-associated and that reside in the cytosol.

Prior work confirmed that chaperone requirements for the degradation of misfolded proteins in yeast can be translated into mammalian cells (see Introduction), even though in some cases the relative magnitude of post-ER quality control (i.e. lysosome-dependent degradation) versus ERAD is higher in human cells relative to yeast (see for example [66]). Therefore, we expressed the STAS domain in HEK293 cells. We first confirmed that STAS domain degradation was proteasome- (i.e. MG-132) dependent: As observed in yeast, the domain was degraded over time but was stabilized by treatment with MG-132 (Figure 8A,B). Next, to determine if Hsp70 facilitates STAS domain degradation in mammalian cells as it does in yeast, cycloheximide chases were again performed, but since thermosensitive Hsp70 mutant alleles are unavailable in higher cells, the cells were instead incubated with an Hsp70 inhibitor, VER155008, which blocks the ATP-binding site in Hsp70 but has no effect on the activity of other ATP-requiring chaperones, such as Hsp90 [108]. Although there was a lag-time for the effect of the compound to become apparent, STAS domain turnover was also slowed after 3 h when Hsp70 activity was inhibited with VER155008. To confirm these results, we modified the assay to examine the steady-state levels of STAS in the HEK293 cells after treatment with VER155008 (Figure 8C). Based on this analysis, STAS levels in the cell were ~2-fold higher in the presence of VER155008 compared with the DMSO control. Our combined results establish that Hsp70 is required for the degradation of the soluble STAS domain in both yeast and human cells. Our results additionally suggest that the recognition of full-length SLC26A9 for ERAD is initiated by the selective capture of the STAS domain by Hsp70.

Figure 8. STAS domain degradation in HEK293 cells is also proteasome- and Hsp70- dependent.

The STAS domain was transiently expressed in transfected HEK293 cells and stability was determined in a cycloheximide chase assay. (A) HEK293 cells were pre-treated for 2 h with 10 μM MG-132 to build-up the levels of the unstable STAS domain, and the cells were then incubated with DMSO, 10 μM MG-132, or 25 μM of the Hsp70 inhibitor, VER155008. A representative immunoblot corresponding to the quantified data is shown. (B) Quantified data based on four independent experiments is shown, ±SEM.*P < 0.05, **P < 0.005. (C) Steady state levels of the STAS protein in HEK293 cells are presented after a 2 h treatment with DMSO or 25 μM VER155008. Left, the graph represents the fold increase in STAS after VER155008 treatment compared with the DMSO control. Right, the primary data showing the relative levels of the STAS domain as well as the total protein loaded. Note that in this experiment the lag-time observed in parts A and B was absent, most likely due to uncharacterized antagonistic effects due to MG-132 pre-treatment. This is apparent by the overall decreased levels of the STAS domain in part A. Data represent the means ± SEM for three independent experiments. *P < 0.05.

Discussion

We report on the first analysis of the molecular requirements that lead to the selection and targeting of SLC26A9, a CFTR-interacting ion channel, for ERAD. To this end, we developed a new SLC26A9 inducible expression system in the yeast S. cerevisiae, an organism that allows for the rapid genetic dissection of the myriad pathways that lead to the disposal of immature ion channels and transporters by the ERAD pathway [109]. Interestingly, a non-trivial fraction of even the wild-type forms of many ion channels and transporters appear to fold inefficiently, an outcome that leads to ERAD targeting. For example, well over half of CFTR is selected for ERAD when stability is assessed in model cell types [7], although the level might be somewhat different when cells are propagated under more native-like conditions [110]. Our data indicate that SLC26A9 — and most notably the STAS domain — is no different: in both yeast and HEK293 cells, the STAS domain is also relatively unstable. These results are consistent with previous reports demonstrating that large quantities of insoluble SLC26A9 accumulate in BHK cells after treatment with MG-132 [38]. In the future, it will be important to measure whether SLC26A9 or CFTR mutants enhance the targeting of SLC26A9 for ERAD, a phenomenon that is commonly noted when CFTR or other ion channels and transporters harbor disease-associated alleles [77].

For many ERAD substrates, the domain or domains that target a protein for the ERAD pathway are poorly defined. In yeast, model ERAD substrates can be binned into ERAD-C, ERAD-M, and ERAD-L substrates, depending on whether the major folding lesion resides in the cytosol, membrane, or ER lumen, respectively [111]. In reality, however, defects in a single domain might impact the integrity of other regions within the protein, and indeed the categorization of ERAD substrates into these three distinct groups breaks down when select proteins are examined in yeast [112] or higher cells [113]. In contrast, the selection of SLC26A9 appears to represent a relatively ‘clean’ example of an ERAD-C substrate since the overall stability of the protein is dictated to a large extent by a single cytosolically disposed domain, i.e. the STAS domain. In addition, the Hsp70-dependence on the degradation of this domain almost fully accounts for the Hsp70-dependent degradation of the full-length protein. We also show that Hsp70 helps target SLC26A9 to the ubiquitin pathway, since disabling Hsp70 (via the use of the thermosensitive Ssa1 mutant) reduced substrate ubiquitination, although not completely. This might arise from the ssa1–45 allele retaining partial function under the conditions of the experiment, or due to the fact that Hsp70 also acts after substrate ubiquitination. Because inhibiting CFTR ubiquitination augments the correction of the F508del CFTR mutant by FDA-approved drugs [114,115], it is possible that similar strategies might one day be used to improve the biogenesis of trafficking-defective SLC26A9 mutants.

One surprising aspect of this study is that — given the complexity of the transmembrane domain of SLC26A9, which possesses 14 individual membrane-spanning helices — the folding efficiency of this domain is high, i.e. the Δstas construct was quite stable. This contrasts with the noted instability of transmembrane domain-containing proteins that are at least partially selected for ERAD-M in yeast [112,116,117]. In fact, a recent cryo-EM structure of SLC26A9 indicates that the transmembrane region is built to allow for uncoupled chloride transport [26]. However, unlike most oligomeric ion channels that possess multiple transmembrane segments, the formation of the SLC26A9 dimer is not mediated by the transmembrane domains. Instead, the transmembrane domains may fold independently of SLC26A9 dimerization, which could explain the relative stability of the Δstas construct.

In contrast to the transmembrane region, we suggest that the STAS domain folds inefficiently, i.e. it is unstable in both yeast and human cells. The STAS domain — and especially the N-terminal portion of the domain — is responsible for the interaction between the two SLC26A9 molecules in the dimer [26]. Thus, it is possible that the post-translational association between two STAS domains as SLC26A9 dimerizes in the cell represents a rate-limiting and problematic step in the folding pathway. Moreover, the STAS domain possesses an intrinsically disordered region that might lead to further destabilization. The PDZ tail also contains a putative disordered region. Intriguingly, the removal of these regions improves SLC26A9 cell surface residence and solubility in detergent [26]. Future work will be undertaken to test these and other hypotheses on the rate-limiting steps in the SLC26A9 folding pathway.

In conclusion, our work shows that complex, polytopic membrane proteins can possess both relatively stable as well as unstable domains whose contributions on overall stability can be differentiated. Moreover, the unstable domain can, in effect, function as a ‘degron’ [118]. In other words, a motif (such as STAS) when attached to a stable polypeptide (i.e. Δstas) is sufficient to direct the fusion protein to the ubiquitin-proteasome system. By dissecting the contributions of each domain, we also discovered that — consistent with prior studies — the Hsp70 chaperone facilitates the delivery of SLC26A9 to the ubiquitination machinery. Based on the importance of SLC26A9 in health and disease, and the generation of drugs that modify protein folding and chaperone function [58,119], our study serves as a first step toward the goal of regulating SLC26A9 stability alone or in combination with vital partners, such as CFTR.

Abbreviations

CFTR

Cystic Fibrosis Transmembrane Conductance Regulator

ER

endoplasmic reticulum

ERAD

endoplasmic reticulum associated degradation

G6PD

anti–glucose-6-phosphate dehydrogenase

STAS

sulfate transporter and anti-sigma factor antagonist

Data Availability

All original data related to work described in this paper are available from the authors upon request.