Boosting endoplasmic reticulum folding capacity reduces unfolded protein response activation and intracellular accumulation of human kidney anion exchanger 1 in Saccharomyces cerevisiae

Human kidney anion exchanger 1 (kAE1) facilitates simultaneous efflux of bicarbonate and absorption of chloride at the basolateral membrane of α‐intercalated cells. In these cells, kAE1 contributes to systemic acid–base balance along with the proton pump v‐H+‐ATPase and the cytosolic carbonic anhydrase II. Recent electron microscopy analyses in yeast demonstrate that heterologous expression of several kAE1 variants causes a massive accumulation of the anion transporter in intracellular membrane structures. Here, we examined the origin of these kAE1 aggregations in more detail. Using various biochemical techniques and advanced light and electron microscopy, we showed that accumulation of kAE1 mainly occurs in endoplasmic reticulum (ER) membranes which eventually leads to strong unfolded protein response (UPR) activation and severe growth defect in kAE1 expressing yeast cells. Furthermore, our data indicate that UPR activation is dose dependent and uncoupled from the bicarbonate transport activity. By using truncated kAE1 variants, we identified the C‐terminal region of kAE1 as crucial factor for the increased ER stress level. Finally, a redistribution of ER‐localized kAE1 to the cell periphery was achieved by boosting the ER folding capacity. Our findings not only demonstrate a promising strategy for preventing intracellular kAE1 accumulation and improving kAE1 plasma membrane targeting but also highlight the versatility of yeast as model to investigate kAE1‐related research questions including the analysis of structural features, protein degradation and trafficking. Furthermore, our approach might be a promising strategy for future analyses to further optimize the cell surface targeting of other disease‐related PM proteins, not only in yeast but also in mammalian cells.

aggregations in more detail. Using various biochemical techniques and advanced light and electron microscopy, we showed that accumulation of kAE1 mainly occurs in endoplasmic reticulum (ER) membranes which eventually leads to strong unfolded protein response (UPR) activation and severe growth defect in kAE1 expressing yeast cells. Furthermore, our data indicate that UPR activation is dose dependent and uncoupled from the bicarbonate transport activity. By using truncated kAE1 variants, we identified the C-terminal region of kAE1 as crucial factor for the increased ER stress level. Finally, a redistribution of ER-localized kAE1 to the cell periphery was achieved by boosting the ER folding capacity. Our findings not only demonstrate a promising strategy for preventing intracellular kAE1 accumulation and improving kAE1 plasma membrane targeting but also highlight the versatility of yeast as model to investigate kAE1-related research questions including the analysis of structural features, protein degradation and trafficking.
Furthermore, our approach might be a promising strategy for future analyses to further optimize the cell surface targeting of other disease-related PM proteins, not only in yeast but also in mammalian cells.

Take Away
• We analysed the intracellular transport of human kAE1 to the yeast plasma membrane.
• We studied the effect of human kAE1 expression on yeast growth and UPR activation.
• We investigated the impact of different kAE1 truncation variants on UPR induction • We implemented intervention strategies to improve PM targeting of kAE1.

| INTRODUCTION
Saccharomyces cerevisiae is often used to study specific aspects of various biological relevant cellular processes such as intracellular trafficking, signalling and/or protein degradation (Berner et al., 2018;Lashhab et al., 2019;Winters & Chiang, 2016). Aside from the high conservation of major cellular processes between yeast and humans, yeast cells further possess critical advantages compared with human cell culture including affordability and rapid cultivation, availability of a broad range of genetic manipulation tools (e.g., plasmid and knock-out strain collections) and the potential for performing high-throughput screening assays (Kolb et al., 2011;Mackie & Brodsky, 2018). Nevertheless, heterologous expression of human proteins often failed and/or ended up with proteins lacking their native function, localization and/or folding state (Bonar & Casey, 2010;Groves et al., 1999;Kolb et al., 2011;Romanos et al., 1992). A current example represents human kidney anion exchanger 1 (kAE1) which was found to be trapped in intracellular membrane structures after its heterologous expression in S. cerevisiae (Bonar & Casey, 2010;Groves et al., 1999;Sarder et al., 2020).
In humans, kAE1 is located at the basolateral membrane of α-intercalated cells (α-ICs) in the connecting tubules and the collecting duct and is part of a cellular machinery that regulates the acid-base homeostasis in the kidney (Lashhab et al., 2018;Roy et al., 2015).
More precisely, membrane permeable CO 2 is normally enzymatically hydrated in α-ICs by carbonic anhydrase II (CA II) to carbonic acid (H 2 CO 3 ). After spontaneous dissociation into H + and bicarbonate (HCO 3 À ), kAE1 is responsible for the reabsorption of HCO 3 À into the blood and the coupled influx of chloride ions (Cl À ) into α-ICs. At the same time, apically located v-H + -ATPases mediate the excretion of H + into the urine. Consequently, proton excretion thus allows acid-base control in the kidney (Alper et al., 2002;Cordat & Casey, 2009;Lashhab et al., 2018). However, in mammals, mutations in kAE1 are associated with the development of a pathological phenotype, known as distal renal tubular acidosis (dRTA) (Alper, 2010;Lashhab et al., 2018;Trepiccione et al., 2017). Although it is speculated that some disease-causing kAE1 mutations lead to intracellular retention and/or mistrafficking of the anion transporter, the components of the intracellular kAE1 transport machinery are still not well characterized, and the actual role of the disturbed transport in dRTA is highly controversial (Cordat et al., 2006;Devonald et al., 2003;Lashhab et al., 2018;Toye et al., 2004).
Our recent study on human kAE1 indicated that yeast might represent a promising alternative model organism to address specific aspects of intracellular transport, function and degradation of complex membrane proteins (Sarder et al., 2020) (Weill, Arakel, et al., 2018;Weill, Yofe, et al., 2018). As previously suggested (Sarder et al., 2020) (Sarder et al., 2020). Therefore, it is likely that kAE1 accumulation either disturbs or at least negatively affects the structure of the yeast ER.
Next, we further assessed whether N-terminal addition of eGFP affects kAE1 targeting to the yeast cell surface. It has already been described that N-terminal modifications do not affect kAE1 targeting to the cell surface in mammalian cells (Beckmann et al., 2002). By using fluorescence-based SIM, colocalization of kAE1 with the yeast PM was visible in Δist2 cells expressing both eGFP-kAE1 and the PM marker Pma1p-mRFP ( Figure 1c) (Grossmann et al., 2007;Papouskova et al., 2017), suggesting that the N-terminal eGFP fusion does obviously not prevent the PM targeting of kAE1 in yeast. However, we cannot completely exclude the possibility that an N-terminal modification of kAE1 impairs its proper in vivo transport. The additional deletion of IST2, a major tethering factor for PM/cortical ER contact sites, should thereby lead to a better separation of kAE1 signals located in cortical ER or PM structures (Manford et al., 2012;Sarder et al., 2020;Wolf et al., 2012).
In summary, our data demonstrate that full-length kAE1 is predominantly located in ER/cortical ER membrane structures, confirming the previous EM data and indicating that the anion transporter is not efficiently released from the yeast ER. However, colocalization of kAE1 with COPII vesicles, early Golgi compartment and PM demonstrates the potential delivery of a minor kAE1 pool through the secretory pathway to the yeast cell surface.

| Yeast cells expressing various kAE1 variants show cell growth defects
In yeast cells, it is known that heterologous expression of foreign proteins can cause cell growth defects often accompanied by ER stress induction (Kintaka et al., 2016). In order to verify whether kAE1 expression influences the growth behaviour of yeast cells, we determined growth rates of wild-type BY4742 cells expressing either various unmodified or HA/FLAG-tagged kAE1 or an EV. As illustrated in Figure 1d,e, cells expressing the EV showed the highest growth rate whereas the expression of the full-length kAE1 variants kAE1 WT and kAE1 HA significantly slowed cell growth. Interestingly, expression of N-terminally truncated kAE1 (kAE1 B3Mem ) (Bonar & Casey, 2010;Groves et al., 1996) and an inactive kAE1 E681Q variant (Jennings & Smith, 1992) lacking its bicarbonate exchange function did not improve cell growth.

| Structural kAE1 yeast homologue Bor1p shows striking similarities in UPR induction
So far, it seems that the intensity of UPR activation correlates with the level of kAE1 expression, similarly to numerous other protein overexpression studies that have previously described this accumulation phenomenon (Sarder et al., 2020;Umebayashi et al., 1997). It is plausible that an overload of the cellular folding machinery in the ER inevitably results in an accumulation of misfolded proteins such as kAE1 in this study. However, the effect of kAE1 was out of scale compared with the overexpression of the yeast HDEL receptor Erd2p.
For this reason, we asked to which extent the expression of other yeast membrane proteins, including the structural yeast kAE1 homologue Bor1p (Coudray et al., 2017;Thurtle-Schmidt & Stroud, 2016), influence UPR induction by determining Kar2p protein levels ( Figure 3a). As expected, expression of all yeast membrane proteinswith exception of the yeast HDEL receptor Erd2p-led to a significant UPR activation indirectly seen by increased Kar2p expression levels.
Surprisingly, Bor1p showed the strongest UPR induction of all yeast proteins. Taking into account that kAE1 expression level was two times higher than Bor1p, both proteins show a similar, disproportional effect on ER stress induction. Therefore, the mere expression did not seem to be the sole determinant for the massive UPR activation.

| Structural features in the C-terminal region of kAE1 contribute to UPR induction in yeast
Given the structural homology between Bor1p and kAE1, we further asked if structural features might play a role in Ire1p-mediated UPR and TMD 5 to 14 (Reithmeier et al., 2016). Because N-glycosylation of full-length kAE1 was not detected in S. cerevisiae (Bonar & Casey, 2010;Sarder et al., 2020) and it is reported that deglycosylation probably affects the folding and aggregation state of kAE1 in oocytes, we asked whether Asn 577 plays a specific role in UPR induction (Groves & Tanner, 1994). Based on the AE1 crystal structure (Reithmeier et al., 2016), both variants should be unable to retain an intact core and/or gate domain; however, our data of kAE1 E681Q already suggest that kAE1 function is not responsible for the observed UPR induction. As shown in Figure 3c, cells expressing . Mean values ± SEM (n = 3) are indicated (*p < 0.05, **p < 0.01 and *** p < 0.001, one-way ANOVA). (b) Schematic outline of various wild-type and truncated kAE1 variants used in (c). The cytosolic N-terminus kAE1 is shown in blue and the cytosolic C-terminus in black. Additionally, the N-glycosylation site (Asn 577 ) and the HA-epitope tag (HA) are illustrated in the different constructs. (c) (left) Western analysis of BY4742 cells expressing EV pYES, kAE1 HA , kAE1 WT and C-terminal and N-terminal truncated kAE1 variants (kAE1 B3mem , kAE1 1-491aa and kAE1 1-666aa ). EV cells in the absence (EV) or presence of DTT (EV + DTT) serve as negative and positive control for UPR induction. Blot was probed with primary antibodies against Kar2p, kAE1 (anti-BRIC170) and the loading control Pgk1p. (middle) Relative Kar2p expression level compared with EV situation (normalized to Pgk1p). Mean values ± SEM (n = 3) are indicated (*p < 0.05, **p < 0.01 and *** p < 0.001, one-way ANOVA). (right) kAE1 expression level relative to kAE1 WT [Colour figure can be viewed at wileyonlinelibrary.com] expression levels and, therefore, might not be important for UPR induction (Figure 3c).
2.6 | Increased ER folding capacity reduces UPR activation in kAE1 expressing cells and significantly improves kAE1 transport to the cell periphery Finally, we searched for solutions to counteract the accumulation and UPR activation induced by full-length kAE1 expression. In yeast, different strategies have already been reported to enhance the folding and secretion capacity of the ER and, thereby, improving proper protein folding and/or targeting in the secretory pathway. Here, we tested whether the additional overexpression of different ER proteins involved in protein folding influences UPR activation. As shown in Figure 4a, under strong galactose-induced kAE1 expression conditions, we could not observe any effect on Kar2p expression after over expression. However, if kAE1 expression was decreased by using a β-estradiol-inducible expression system which allows a dosedependent kAE1 expression regulation, a significant reduction of Kar2p expression was observed in cells expressing Pdi1p or Emc1p. In contrast, the effect was not visible in EV control cells (Figure 4b).
Because relative kAE1 expression levels did not significantly differ between all strains, the missing upregulation of Kar2p is presumably mediated by the increased ER folding capacity after additional ER chaperone expression (Figure 4b). These findings nicely indicate that an improved kAE1 folding can prevent kAE1-mediated UPR induction in yeast. Consistently, if one would expect that more correctly folded kAE1 is present under these conditions, an improved kAE1 processing within the secretory pathway and trafficking to the yeast PM should take place. Indeed, a drastic shift in the localization of kAE1 was visible in SIM images. As expected and illustrated in Figure 4c, cells simultaneously expressing eGFP-kAE1 and the EV pBG1805 showed a distinct overlap of kAE1 signals (green) with the mCherry-labelled F I G U R E 4 Increased endoplasmic reticulum (ER) folding capacity reduces unfolded protein response (UPR) activation. (a) (Top) Western analysis of BY4742 cells expressing eGFP-kAE1 WT together with the empty vector (EV) pBG1805 or the indicated yeast ER chaperones. EV cells without an additional kAE1 plasmid served as negative control. Blot was probed with primary antibodies against Kar2p, kAE1 (anti-BRIC170) and the loading control Pgk1p. Anti-HA antibodies were used to detect expression of the different ER chaperones. (Down) Relative Kar2p expression level compared with EV situation (normalized to loading control pPgk1p). Mean values ± SEM (n = 3) are illustrated. (b) (left) Western analysis of BY4742 GEV cells expressing eGFP-kAE1 WT together with the EV pBG1805 or the indicated yeast ER chaperones. kAE1 expression was induced with 30-nM β-estradiol. EV cells without an additional kAE1 plasmid served as negative control. Blot was probed with primary antibodies against Kar2p, kAE1 (anti-BRIC170) and the loading control Pgk1p. Anti-HA antibodies were used to detect expression of the different ER chaperones. (middle) Relative Kar2p expression level compared with EV situation (normalized to loading control pPgk1p). (right) Relative kAE1 expression level compared with EV situation (normalized to loading control Pgk1p). Mean value ± SEM (n = 3) are illustrated (*p < 0.05, **p < 0.01 and *** p < 0.001, one-way ANOVA). (c) Structured illumination microscopy (SIM) images of yeast cells expressing mCherry-tagged ER marker Sec63p (mCherry) together with the eGFP-tagged kAE1 (GFP-KAE1) and the indicated ER chaperones. pBG1805 (EV) served as negative control without any ER chaperone expression. Scale bar: 5 μM. (d) SIM images of yeast cells expressing mRFP-tagged PM marker Pma1p (mRFP) together with the eGFP-tagged kAE1 (GFP-KAE1) and the indicated ER chaperones. EV pBG1805 served as negative control. Scale bar: 5 μM [Colour figure can be viewed at wileyonlinelibrary.com] ER marker Sec63p (red). In contrast, additional expression of ER chaperones led to a dramatic change in the intracellular kAE1 localization.
All cells showed a similar GFP signal distribution in the form of ringlike structures at the cell periphery. Moreover, SIM experiments with an mRFP-labelled PM marker (Pma1p) showed colocalization of the peripheral GFP signals with the yeast PM ( Figure 4d). These data strongly support our hypothesis that misfolded kAE1 accumulates in the ER, whereas an improved ER folding capacity is capable to at least partially restore the native folding state of the anion transporter under a regulated and dose-dependent kAE1 expression system.

| DISCUSSION
Various studies in the last decades have highlighted the model organism yeast as alternative expression platform for human proteins to investigate fundamental cellular processes including protein degradation, biosynthesis and transport (Sarder et al., 2020). However, heterologous expression of human proteins (e.g., membrane proteins) is not trivial and lack of expression and/or loss of function are common issues in this context (Bonar & Casey, 2010;Groves et al., 1999;Kolb et al., 2011). Furthermore, the use of strong promoters and multicopy vectors exceed in many cases the folding capacity of yeast cells, causing aggregation and/or accumulation of misfolded proteins (Kintaka et al., 2016;Umebayashi et al., 1997). Most recently, we have performed a pilot study in S. cerevisiae showing the successful expression of full-length variants of the human kAE1 with correct PM localization and biological activity. Nevertheless, a major drawback of the current model system was the tremendous kAE1 accumulation observed in intracellular compartments and low abundance of PM-localized kAE1 (Sarder et al., 2020).
By performing colocalization experiments with fluorescently labelled kAE1 and organelle markers for the secretory pathway, we now provide a more precise picture of the intracellular kAE1 trafficking in yeast. As expected from our previous EM analyses (Sarder et al., 2020), the vast majority of full-length kAE1 colocalizes with ER and cortical ER structures. However, some kAE1 signals overlap with the COPII vesicle marker Sec13p and the Golgi marker Anp1p, indicating an anterograde transport of kAE1 from the ER to the Golgi apparatus. Additionally, early and late endosomal compartment markers did not show any colocalization with the fluorescent anion transporter whereas the overlap of kAE1 with the PM marker Pma1p was clearly visible. Because the colocalization experiments of kAE1 and Pma1p were performed in the genetic background of a yeast Δist2 mutant, it is most likely that the overlap is not caused by kAE1 signals derived from the cortical ER. Surprisingly, our data showed costaining with Golgi structures and PM but not with endosomal structures. In our opinion, the most likely explanation for this observation is an endosome-independent transport of kAE1 directly from the trans-Golgi network (TGN) to the yeast PM. In yeast, several alternative PM transport routes (e.g., exomer) have been described which allow direct cargo transport from the Golgi to the cell surface (Spang, 2015;Wang et al., 2006). Furthermore, endosome-independent trafficking pathways are also found in mammalian cells (Chen et al., 1998;Parmar & Duncan, 2016;Wakana et al., 2012). Interestingly, it has been postulated that kAE1 can reach the PM by an endosome-independent transport mechanism (Junking et al., 2014).
Alternatively, the lack of colocalization with endosomal markers may simply reflect the transitory nature of the colocalization in endosomes that was not captured in our pictures. In the future, it would be interesting to study the exit of kAE1 from the yeast TGN in more detail to address how adequate the yeast model system can mimic the native PM targeting of kAE1 in mammalian cells.
Based on our biochemical data, it is obvious that heterologous kAE1 expression induces a massive UPR response in S. cerevisiae.
Both upregulation of the ER chaperone Kar2p and the increased activation of UPR-responsive elements demonstrate that yeast cells show an increased stress response in the presence of kAE1. Consistently, inactivation of the Ire1p pathway by using Δire1 and Δhac1 cells completely prevented Kar2p induction under induced kAE1 expression. Because UPR activation represents a cellular response to misfolded proteins (Kimata & Kohno, 2011), we conclude that the majority of newly synthesized kAE1 is not properly folded and, hence, predominantly aggregates and/or accumulates in the yeast ER. Consequently, the release of misfolded kAE1 from the ER is interrupted, resulting in its inefficient targeting to the cell surface.
Diminished growth rates under kAE1 expression further support our hypothesis. Interestingly, neither inactive kAE1 E681Q nor N-terminally truncated kAE1 B3Mem was capable of restoring cell growth and of preventing UPR response.
Previous yeast studies already speculated that structural features in the N-terminal region of kAE1 might be the reason for its intracellular ER retention (Groves et al., 1999). Because removal of the N-terminus did not show an increased PM localization in the previous EM study (Sarder et al., 2020) and did not show an impact on UPR induction, it is unlikely that this region plays an essential role in intracellular kAE1 retention. However, we now demonstrate that structural features in the C-terminus encompassing TMD 9-14 and the cytosolic carboxyl-terminus are partially involved in the ER stress induction of kAE1 expressing cells (Figure 3b,c). Surprisingly, the structural kAE1 homologue Bor1p showed a comparably high UPR response after its overexpression whereas the expression of other yeast membrane proteins only moderately induced the upregulation of the ER stress indicator Kar2p (Coudray et al., 2017;Thurtle-Schmidt & Stroud, 2016).
Remarkably, overexpression of the yeast HDEL receptor Erd2p that served as a yeast endogenous protein overexpression scenario did not significantly activate the UPR pathway in yeast cells. Due to HDEL-dependent retention function, the increased cellular Erd2p quantity presumably accelerates the retrieval of ER-luminal proteins back to the ER. Because most ER chaperones, including Kar2p and Pdi1p, contain C-terminal HDEL-retention motifs, a higher ER folding capacity could be expected which simultaneously prohibits the accumulation of misfolded and UPR-inducing proteins (Becker et al., 2016;Semenza et al., 1990).
Our hypothesis that structural features in kAE1 and Bor1p trigger Ire1p-mediated UPR induction was supported by additional truncation experiments. Indeed, the expression of shorter kAE1 variants lacking parts of the C-terminal region (kAE1 1-491aa and kAE1 1-666aa ) significantly reduced cellular Kar2p levels. Given that Kar2p upregulation is dose dependent and not completely abolished in the truncated variants, we therefore speculate that structural properties between TMD 10-14 and the cytosolic C-terminus are required for the kAE1-induced UPR activation in yeast. However, a more detailed analysis of other truncated kAE1 variants is required to precisely map the C-terminal region responsible for the reduced UPR activation. It is likely that the cytosolic C-terminal part of kAE1 induces the UPR phenotype. In mammalian cells, it is known that the cytosolic C-terminus represents an interaction hotspot that modulates the proper transport of kAE1 to the basolateral membrane (Almomani et al., 2012;Su et al., 2011;Su et al., 2017;Toye et al., 2004). Theoretically, the mere lack of such an interaction partner required for proper folding could induce the observed accumulation and misfolding of kAE1 in yeast. In this context, simultaneous coexpression of glycophorin A, a known kAE1 interaction partner, was shown to improve the PM targeting of kAE1 in yeast cells and frog oocytes (Groves et al., 1999;Groves & Tanner, 1992;Young et al., 2000). However, it should be noticed that several known dRTA-causing mutations are located in this C-terminal kAE1 region from TMD 10 to 14 (p.R901X, p.D905dup, p. D905Gfs15, p.E906K and p.M909T). A couple of these mutations (e.g., p.G701D or p.S773P) also lead to incorrect folding and/or distinct trafficking defects such as ER or Golgi retention in mammalian MDCK cells (Cordat et al., 2006). In future studies, it would be interesting to understand to which extend these mutations trigger UPR activation in mammalian cells and if chemical chaperones or coexpression of ER chaperones can prevent kAE1 mistrafficking and eventually restore kAE1 functionality in α-ICs.
To perform a genome-wide screening for potential kAE1 transport modulators, the inefficient kAE1 targeting to the secretory pathway represents a major drawback of the model system introduced by Sarder et al. (2020). However, our present data provide a first strategy to overcome this issue and optimize the yeast model system for future transport screening studies. Surprisingly and unexpected, overexpression of an individual ER chaperone not only prevents UPR activation under moderate kAE1 expression levels but also dramatically improves PM targeting of kAE1. These results further strengthen the idea that kAE1 misfolding in the ER is the major bottleneck for studying kAE1 trafficking in yeast. However, UPR activation was not prevented if kAE1 expression exceeds a certain limit, demonstrating that a fine-tuning of the expression parameters is crucial to achieve optimized conditions for kAE1 folding and trafficking. Notably, the selection of the yeast strain has also a strong impact on the kAE1 distribution at the yeast PM. Although BY4741 and BY4742 cells show a clear kAE1 localization at the PM after ER chaperone overexpression, the fluorescence pattern in BY4741 and BY4742 cells differs.
Although the microscopic and biochemical results strongly support our hypothesis that kAE1 is more efficiently folded and transported to the PM after expanding the ER folding capacity, future EM studies in analogy to the previous EM study would be essential to finally prove our assumption (Sarder et al., 2020).
It should be noticed that overexpression of cytosolic and ER-localized chaperones is a widely used strategy to prevent or counteract the misfolding and aggregation of various proteins in different species. For example, the commercially available GroEL-GroES system from Escherichia coli is often used in bacteria to improve the yield and biological activity of recombinantly expressed proteins (Lamppa et al., 2013;Nishihara et al., 1998).
Previous studies already used Kar2p (BiP) or Pdi1p overexpression in S. cerevisiae to boost the secretion of recombinant proteins including β-glucosidase, α-amylase and others (Hou et al., 2012). It is further reported that the expression of Hsp70, Hsp90 and BiP can prevent and/or rescue the misfolding of disease-related PM proteins. Therefore, Pdi1p represents a potential candidate not only to improve PM transport of kAE1 but also to foster PM targeting of other kidney disease-related proteins which have already been studied in yeast as model organism (e.g., aquaporins and CTFR) (Kolb et al., 2011). So far, Emc1p overexpression has not been described as a strategy to improve PM targeting of membrane pro- Collectively, our study highlighted the manifold options of yeast as model system to investigate specific aspects of kAE1 physiology in the future. Moreover, our data provide-for the first time-deeper insight in the intracellular transport of human kAE1 in S. cerevisiae and illustrate the reason and consequences of its tremendous accumulation seen in the recent EM study. Finally, we further provide a simple strategy to prevent the intracellular trapping of the human anion exchanger, enabling an improved kAE1 targeting to the yeast cell surface. With this novel knowledge, we believe that genome-wide screenings to identify candidate genes involved in kAE1 transport are now imaginable and will hopefully initiate further kAE1-related research in S. cerevisiae.

| Cultivation and transformation of yeast cells
Yeast cultivation was routinely performed at 30 C and 220 rpm in shaking flasks using standard YPD, synthetic complete or drop/out media containing 2% glucose or 3% galactose. S. cerevisiae strains used in this study are listed in Table S1. A previously described standard protocol was used for yeast transformation (Becker et al., 2016).

| Chromosomal integration of Pma1p-mRFP in Δist2 cells via homologous recombination
For chromosomal integration of the mRFP-tagged PM marker Pma1p, plasmid YIp128-Pma1-mRFP (Grossmann et al., 2007) was initially linearized with BglII, separated via agarose gel electrophoresis and subsequently purified via a gel extraction kit (Omega). Then, Δist2 BY4741 cells (Papouskova et al., 2017) were transformed with the linearized plasmid and selected on leucine d/o plates at 30 C for at least 3 days. Successful homologous recombination of selected clones was verified by Western blot analysis (data not shown) and SIM.

| Monitoring of UPR activation via Kar2p expression
In brief, cells carrying the indicated plasmids were inoculated in 5 ml of the appropriated d/o glucose medium for 18 h. For inducing GAL1 promoter-driven protein expression, cells (300-500 μl) were shifted to 10 ml of the appropriate d/o galactose medium and grown to OD 600 = 0.6-0.8. For ER stress induction, cells carrying an EV were additionally treated with 2.5 mM DTT or 2 μg/ml TM for 6 h. Next, cell pellets (OD 600 = 5) were harvested at 8000 rpm for 5 min, washed twice with distilled H 2 O and resuspended in 200-μl homogenization buffer (10 mM Tris, 1 mM EDTA, 1 mM PMSF, 0.1% β-mercaptoethanol, pH 7.4) containing protease inhibitors (EDTA-free, Roche). For cell lysis, samples were disrupted by a beat beater (Precellys Evolution, Peqlab) using glass beads and the following parameters: 6000 rpm, 3 Â 20-s shaking interval, 30-s break/interval.
After incubation of the cell lysates at 37 C for 15 min, cell debris were removed by a single centrifugation step at 4 C and 14,000 rpm (10 min), and supernatants were directly used for SDS-PAGE and immunoblotting.
For yeast growth rate calculation, measured OD 600 values of the different independent biological triplicates were plotted as a function of time, and data were fitted by exponential regression. Based on the obtained equation, mean growth rates per hour (curve slope) ± SD were determined.
For inducing kAE1 expression, the different strains were initially cultivated in 5-ml uracil d/o glucose medium and subsequently shifted to 5-ml uracil d/o galactose medium for 18 h at 30 C and 220 rpm.
To induce ER stress, cells carrying an EV were additionally treated with 2.5 mM DTT or 2-μg/ml TM for 6 h. Finally, 10-μl aliquots were spotted on poly-L-lysine-coated cover slips and incubated for 15 min to prevent cell movement during microscopy. To visualize UPR activation, GFP expression levels were monitored via Western blot or fluorescence microscopy using a Keyence BZ-8000 microscope (100x Oil immersion Plan Apo VC objective [1.4 NA]) with the preinstalled filter sets and standard settings for GFP (488 nm).

| EM analysis
Sample preparation, immunostaining and image acquisition were performed as previously described (Sarder et al., 2020). In brief, log phase yeast cultures (OD 600 = 0.6-0.8) of Δend3 cells expressing full-length kAE1 or EV were filtered into a paste which was pipetted into

| Structured illumination microscopy
In brief, cells were spotted on 25-mm round glass cover slips (Warner Instruments) following an 1-h incubation period to let them settle down to prevent cell movement during the experiment. Next, SIM images were acquired using an inverted Elyra PS1 microscopic system (Carl Zeiss Microscopy, GmbH, Jena) equipped with a 63Â Plan-Apochromat objective (NA = 1.4, Carl Zeiss Microscopy, GmbH, Jena). Samples were illuminated with the corresponding excitation wavelength for GFP, mRFP and mCherry. Acquisition and super resolution processing were performed with Zen 2012 software (Carl Zeiss, Microscopy GmbH, Jena, Germany). Further image analysis was carried out with the Fiji software (Schindelin et al., 2012). For observing dim structures, brightness and contrast of each fluorescence channel were separately adjusted.

SDS-PAGE was performed under nonreducing conditions in 10%
Tris-Tricine gels using a buffer system as previously described (Ploug et al., 1989). By performing semidry blotting (Jacobson & Kårsnäs, 1990), proteins were subsequently transferred onto PVDF membranes in the presence of transfer buffer (25 mM Tris, 190 mM glycine, 0.1% SDS, 20% methanol). Usually, kAE1 expression was validated using primary anti-kAE1 antibodies (BRIC170, recognizing an epitope in the amino acid region of 368-382) and visualized with secondary HRP-conjugated anti-rabbit antibodies. For Kar2p detection, blots were probed with anti-Kar2p and anti-rabbit-HRP antibodies. To monitor expression of yeGFP-kAE1, blots were probed with anti-GFP and anti-mouse-HRP antibodies. For visualization of the loading control Pgk1p (phosphoglycerate kinase 1), Western blots were incubated with anti-Pgk1 and HRP-coupled anti-rabbit antibodies. Finally, PVDF membranes were incubated with SuperSignalTM West Femto Maximum Sensitivity Substrate (ThermoScientific), and protein signals were visualized with Amersham Imager 600 (GE Healthcare). All antibody dilutions and sources are listed in Table S3.

| Data analysis and statistics
Statistical analysis was carried out in GraphPad Prism 8 (GraphPad Software, San Diego, California, USA). All pooled data were given as mean values ± SEM (unless otherwise stated). Statistical significance was assessed by one-way ANOVA based on biological replicates and at sample sizes of n ≥ 3 independent experiments (ns, not significant; *p < 0.05; **p < 0.01; *** p < 0.001).