Cells continuously adapt cellular processes by integrating external and internal signals. In yeast, multiple stress signals regulate pheromone signaling to prevent mating under unfavorable conditions. However, the underlying crosstalk mechanisms remain poorly understood. Here, we show that mechanical stress activates Pkc1, which prevents lysis of pheromone-treated cells by inhibiting polarized growth. In vitro Pkc1 phosphorylates conserved residues within the RING-H2 domains of the scaffold proteins Far1 and Ste5, which are also phosphorylated in vivo. Interestingly, Pkc1 triggers dispersal of Ste5 from mating projections upon mechanically induced stress and during cell–cell fusion, leading to inhibition of the MAPK Fus3. Indeed, RING phosphorylation interferes with Ste5 membrane association by preventing binding to the receptor-linked Gβγ protein. Cells expressing nonphosphorylatable Ste5 undergo increased lysis upon mechanical stress and exhibit defects in cell–cell fusion during mating, which is exacerbated by simultaneous expression of nonphosphorylatable Far1. These results uncover a mechanical stress–triggered crosstalk mechanism modulating pheromone signaling, polarized growth, and cell–cell fusion during mating.
Interplay between signaling networks determines proper regulation of cell growth, survival, and fate. In the budding yeast Saccharomyces cerevisiae, many signaling components have been identified and studied at the molecular and mechanistic levels (Brent, 2009; Alvaro and Thorner, 2016), but the dynamic interactions required to temporally and spatially orchestrate appropriate cellular responses remain poorly understood. One such case is the yeast pheromone response pathway, which is activated in a cell cycle–dependent manner in haploid cells upon contact with a partner of opposite mating type. Pheromone signaling is inhibited by CDK activity and thus restricted to the G1 phase of the cell cycle when both partners have an equal amount of DNA, assuring genomic integrity (Strickfaden et al., 2007). In addition, other intra- and extracellular stress signals impinge on the mating reaction to avoid cell lysis under conditions unfavorable for mating and cell–cell fusion. For example, signaling through the high-osmolarity glycerol pathway down-regulates the pheromone response pathway (Yamamoto et al., 2010; Vaga et al., 2014). Yeast mating thus offers an attractive system to study signal integration in cell fate determination, as multiple and at times opposing signaling inputs can be combined.
Pheromone signaling induces G1 cell cycle arrest, cell polarization, and initiation of a cell–cell fusion transcriptional program (Dohlman and Slessareva, 2006). All three responses are orchestrated by the scaffolds Ste5 and Far1, which function as assembly platforms and comprise important regulatory nodes that allosterically modulate signaling output (Ferrell and Cimprich, 2003). Mating pathway activation is triggered by pheromones binding to dedicated receptors, which leads to dissociation of the Gβγ heterodimer (Ste4–Ste18) from the α-subunit (Gpa1) of the heterotrimeric G protein. Ste5 and Far1 bind the released Gβγ complex through their RING-H2 domains at the plasma membrane. Ste5 membrane association further requires cooperative effects of the pleckstrin homology (PH) domain and the plasma membrane binding motif (PM) domain to increase cell membrane affinity (Winters et al., 2005). Pheromone signaling is down-regulated by negative feedback through the MAPK Fus3 (Choudhury et al., 2018; Repetto et al., 2018) and CDK1-mediated phosphorylation of multiple sites flanking the PM domain (Oehlen and Cross, 1994; Strickfaden et al., 2007) by electrostatic exclusion with the negatively charged head groups of the phospholipids (Strickfaden et al., 2007).
Membrane-bound Ste5 associates with and activates a MAPK module composed of Ste11, Ste7, and the MAPKs Fus3 and Kss1, while Far1 binds and activates the Cdc42 exchange factor Cdc24 to direct cell polarity (Dohlman and Slessareva, 2006). The PAK-like kinase Ste20 phosphorylates Ste11 in a Cdc42-dependent manner. Ste11 then activates Ste7, which in turn doubly phosphorylates Fus3 and Kss1. Activated Fus3 and Kss1 phosphorylate cytoplasmic and cell membrane–tethered targets, including Ste18 and Far1, to promote cell cycle arrest and translocate into the nucleus, activating a specific transcriptional program.
Polarized growth toward the mating partner and the subsequent cell–cell fusion events require spatial cell wall remodeling that renders cells vulnerable to lysis (Zarzov et al., 1996). Upon cell wall stress, Pkc1 is activated, which in turn triggers a MAPK cascade resulting in activation of the MAPK Mpk1 (Slt2; Levin, 2011) to reinforce cell wall synthesis (Smits et al., 1999). Pkc1 is also activated in mechanically stressed cells and prevents cell lysis at least in part by inhibiting actin-driven polarized growth (Delarue et al., 2017; Mishra et al., 2017). Indeed, Mpk1 pathway components are actively recruited to shmoo tips by interacting with the scaffold protein Spa2 (van Drogen and Peter, 2002), and mpk1Δ cells show increased lysis upon exposure to pheromones (Merlini et al., 2013; Engelberg et al., 2014).
Here, we show direct pheromone pathway modulation by the cell wall integrity (CWI) pathway upon mechanical stress. We found that physical pressure activates Pkc1, which in turn prevents Ste5 accumulation at shmoo tips. Molecular analysis revealed that phosphorylation of specific sites in the RING-H2 domains of Ste5 and Far1 interfere with their binding to Gβγ heterodimers, thereby inhibiting Fus3 activity. In the absence of this mechanism, cell viability is reduced due to increased lysis during pheromone-induced polarized growth and cell–cell fusion. Thus, timely inactivation of pheromone signaling by regulating Gβγ-mediated membrane association of the scaffold proteins Ste5 and Far1 is part of a Pkc1-dependent crosstalk mechanism to prevent cell lysis in response to mechanical stress and cell wall remodeling during cell–cell fusion.
Mechanical stress inhibits the pheromone response pathway in a Pkc1-dependent manner
Previously, we have shown that mechanical stress activates Pkc1 to protect cells from lysis partially by inhibiting polarized growth (Mishra et al., 2017). Indeed, inhibition of Pkc1 with cercosporamide, or overexpression of a dominant-negative Pkc1 allele (Pkc1K853R; Watanabe et al., 1994), increased lysis of pheromone-treated cells exposed to mechanical stress (Fig. 1, A–C). Because the MAPK Fus3 promotes polarized growth and shmoo formation, we used the Fus3 SKAR (synthetic kinase activity relocation) reporter (Durandau et al., 2015) to assess whether mechanical stress inhibits Fus3 activity. As expected, the SKAR reporter was mainly cytoplasmic in pheromone-treated cells but showed nuclear accumulation with mechanical pressure, implying that Fus3 activity is reduced in response to mechanostress (Fig. 1 D). Cercosporamide prevented Fus3 inhibition, and phenotypic analysis revealed that those cells displaying high Fus3 activity lyse (Fig. 1 E). Additionally, we quantified mechanostress induced lysis of cells expressing a NaPP1-inhibitable Fus3 mutant protein. Indeed, Pkc1-inhibited cell lysis was suppressed by simultaneous addition of NaPP1 (Fig. 1 F), indicating that Pkc1-dependent Fus3 inhibition protects cells from mechanostress induced lysis during mating. This Pkc1-dependent crosstalk to the pheromone pathway was unlikely to be caused by off-target effects of cercosporamide (Fig. S1). Surprisingly, cells lacking the MAPK Mpk1 were less prone to lyse than cercosporamide-treated cells (Fig. 1 G), implying that unknown Pkc1 targets must exist to protect cells from mechanostress-induced lysis in response to pheromones. To understand this crosstalk, we examined Ste5 localization upon mechanical stress. While triple Venus (tV)–tagged Ste5 (Ste5-tV) accumulated at shmoo tips in the absence of stress, Ste5-tV was dispersed upon mechanical pressure (Fig. 1 H), which activates both Pkc1 and Mpk1 (Fig. S2, A and B). Ste5 dispersal was blocked by addition of cercosporamide primarily by an Mpk1-independent mechanism (Figs. 1 H and S2 C), implying that Pkc1 may directly regulate membrane association of Ste5. Indeed, Ste5 dispersal was mimicked by expression of a dominant-active Pkc1 (Pkc1R398A; Fig. S2 D). Furthermore, expression of Pkc1R398A before pheromone treatment resulted in failure to recruit Ste5 and form shmoos (Fig. S2, E and F). In contrast, cells expressing a weakly constitutive allele of Bck1 (Bck1-20) responded normally to pheromone (Fig. S2 G). Together, these results imply that Pkc1 interferes with pheromone response, likely by preventing membrane association of Ste5 by an Mpk1-independent mechanism (Fig. 1 I).
Ste5 and Far1 are phosphorylated in their RING-H2 domain by Pkc1
We mapped Ste5 phosphorylation sites by tandem mass spectrometry (MS) analysis in cells exposed to α-factor. We purified Ste5-HTB under denaturing conditions, and enriched phospho-peptides using TiO2. This analysis identified over 40 Ste5 phosphorylation sites (Figs. 2 A and S3 and Table S1), among them several previously described or predicted sites, confirming that Ste5 is a major hub for regulation (Choudhury et al., 2018; Repetto et al., 2018). Interestingly, Ste5 was also phosphorylated on S185 close to the first pair of finger cysteine residues within its RING-H2 domain (Figs. 2 A and S3 A and Table S1). Sequence analysis indicates that the corresponding serine and the surrounding residues are largely conserved in Ste5-containing yeast species, and similar phosphorylatable residues are also found in the Far1 RING-H2 domain (Fig. 2 A). MS/MS analysis of purified Far1 confirmed that S210 and possibly also S208 and S211 in its RING-H2 domain were phosphorylated in cells exposed to α-factor (Figs. 2 A and S3 B and Table S2). The charged residues surrounding Ste5 S185 with downstream lysine residues do not conform to a MAPK consensus site but rather resemble motifs phosphorylated by Pkc (Kreegipuu et al., 1998). Due to the unfortunate physical properties of this particular phospho-peptide, we failed to establish selected reaction monitoring assays to reliably quantify S185 phosphorylation. While we could detect the phosphorylated S185-containing peptide in pheromone-treated WT and ste7Δ cells, the signal was too low to allow robust quantification across different conditions. Therefore, we were unable to determine whether S185 phosphorylation depends on Pkc1 or mechanical stress in vivo. We thus tested whether these serine residues in Ste5 and Far1 could be phosphorylated by Pkc1 in vitro. Indeed, Escherichia coli–expressed 6His-RING-H2 fragments of Ste5 (amino acids 149–238) and Far1 (amino acids 173–261) were phosphorylated by GST-Pkc1 affinity purified from yeast extracts and immobilized on GST beads (Fig. 2 B). To confirm that Pkc1 phosphorylates S185, we performed in vitro phosphorylation assays using either WT (Ste5149–238) or nonphosphorylatable (Ste5149–238 S185A) RING-H2 fragments of Ste5 and analyzed the phosphorylation status by nuclear magnetic resonance (NMR). Indeed, a new peak was identified in the 1H-15N correlation spectrum upon incubation with yeast extract (Fig. 2, C and D) or purified Xenopus laevis Pkc1δ (Fig. S4). This peak was absent when the Ste5149–238 S185A mutant protein was used as a substrate (Fig. 2 E) or when the kinase assays were performed in the presence of cercosporamide (Figs. 2 F and S4). No additional peaks were detected when the Ste5149–238 fragment was incubated with Pkc1, indicating that S185 is the dominant phosphorylation target in the RING-H2 domain.
To test whether Pkc1 and Ste5 directly interact in vivo, we used the M-track protein–protein proximity assay (Zuzuarregui et al., 2012; Brezovich et al., 2015) designed to capture short-lived interactions. Ste5 was tagged with the active enzymatic domain of the murine histone lysine methyltransferase Suv39 (HKMT-myc). Pkc1 was fused to the prey sequence protA-H3, harboring a histone H3 peptide sequence that becomes permanently methylated by Ste5-HKMT-myc upon close proximity. Indeed, as shown in Fig. 2 G, Pkc1 and Ste5 showed M-track proximity signals that increased with extended times of α-factor treatment, indicating a physical interaction between the proteins. We conclude that Pkc1 physically interacts with Ste5 at shmoo tips and directly phosphorylates S185 in its RING-H2 domain in in vitro experiments.
Pkc1-dependent phosphorylation of Ste5 inhibits signaling
We examined Ste5 S185 function in vivo using the nonphosphorylatable (Ste5S185A) and phosphomimetic (Ste5S185D) versions of GFP-tagged Ste5. Although proteins are expressed at comparable levels (Fig. 3 A), ste5Δ cells expressing Ste5S185D were unable to arrest their cell cycle in response to pheromone, while the halo in ste5Δ cells expressing Ste5S185A was similar in size when compared with WT controls (Fig. 3 B). After α-factor exposure, ste5Δ cells harboring untagged WT Ste5 or Ste5S185A strongly induced expression of the quadruple Venus (qV) fluorescence reporter expressed from the FIG1 promoter (pFIG1-qV), a bona fide Ste12 target, while cells expressing Ste5S185D did not induce even after prolonged exposure (Fig. 3 C). These results suggest that phosphorylation of Ste5 at serine 185 interferes with pheromone signaling in vivo.
We also analyzed WT Far1 and the corresponding nonphosphorylatable Far13A (S208A, S210A, and S211A) and the phosphomimetic Far13E (S208E, S210E, and S211E) versions for their ability to polarize cells toward artificial pheromone gradients. While WT and both Far1 mutants were able to arrest the cell cycle as expected, Far13E exhibited impaired orientation function (Fig. 3 D), which is known to require an intact RING-H2 domain (Lee et al., 2012; Hegemann et al., 2015). Taken together, we conclude that phosphorylation of the RING-H2 domain inhibits Ste5 and Far1 function and renders the scaffolds signaling incompetent in vivo.
Phosphorylation of Ste5 S185 abolishes its interaction with Gβγ
We next investigated Pkc1-mediated Ste5 inhibition. Ste5 is recruited to the plasma membrane by binding to Gβγ through its RING-H2 domain (Whiteway et al., 1995), as well as by interactions between the PM domain and membrane-bound phospholipids (Strickfaden et al., 2007). While GFP-tagged WT Ste5 and the Ste5S185A mutant accumulated at tips of mating projections in WT cells (Fig. 4 A), the phospho-mimicking GFP-Ste5S185D mutant protein was absent from shmoo tips and instead was cytoplasmic. To corroborate these data, we compared the membrane recruitment kinetics of WT and mutant GFP-Ste5 expressed in cells exposed to pheromones (Fig. 4 B; Colman-Lerner et al., 2005). WT and the nonphosphorylatable Ste5S185A mutant showed indistinguishable membrane recruitment kinetics with a fast initial phase followed by a moderate decline due to negative feedback regulation (Yu et al., 2008). In contrast, no membrane recruitment was observed for GFP-Ste5S185D, even during early stages, indicating defective initial recruitment. If the membrane recruitment defect is solely responsible for the signaling defect of the Ste5S185D mutant, then its artificial recruitment to the cell membrane should be sufficient to restore signaling (Pryciak and Huntress, 1998). Indeed, pFIG1-qV reporter expression was restored in ste5Δ cells expressing a fusion construct from the estradiol-inducible GAL promoter of Ste5S185D with a constitutive transmembrane domain (TMD; Fig. 4 C). Taken together, we conclude that phosphorylation of serine 185 in the RING-H2 domain regulates membrane recruitment of Ste5.
We next used yeast two-hybrid and in vitro binding assays to examine whether phosphorylation of the RING-H2 domains of Far1 regulates its interaction with free Gβγ. While WT Far1 and the Far13A mutant readily interacted with Ste4 by two-hybrid assay, binding to phospho-mimicking Far13D was diminished (Fig. S5, A and B). Likewise, much less 6-His-Ste5S185D was retained on immobilized GST-Ste4 than WT (Fig. 4 D), suggesting that phosphorylation of S185 interferes with binding of Ste5 to Gβγ in vitro. We then performed NMR titration experiments using isotopically labeled (15N) Ste5 RING-H2 domain with unlabeled Gβγ (Ste4–Ste18) purified from yeast (Fig. 4 E), which allows fast and robust assessment of binding partners. Overlaying the NMR spectra of the unphosphorylated RING-H2 domain and the WT RING-H2 domain (Fig. 4 E, black peaks) overtitrated with Gβγ (red peaks) reveals distinctive patterns that indicate binding. Increase of total peak number from ∼90 (black) to ∼170 (red) is consistent with the observation that Ste5 undergoes asymmetric oligomerization while binding to Ste4 (Inouye et al., 1997), although it is unknown whether the asymmetric RING-H2 oligomer is bound to a single Ste4–Ste18 complex or to two complexes that are linked by the homodimerizing GST moiety. The low protein concentrations used for this in vitro assay (10 µM Ste5 and 30 µM Gβγ) and the substantial level of spectral rearrangements imply that the interaction occurs with nanomolar affinity. Analogous titrations with nonphosphorylatable S185A RING-H2 mutant domain similarly showed a specific interaction with Gβγ, although the perturbations were less pronounced (Fig. 4 E, middle panel), indicative of reduced affinity compared with the WT control. Importantly however, no changes in peak positions were detected when titrating Gβγ and the S185D RING-H2 mutant (right panel), demonstrating that the phospho-mimicking mutant protein fails to interact with Gβγ. We were able to assign ∼80% of the peaks in the [1H,15N]-HSQC spectra. This allowed deducing a structural model of the Ste5 RING-H2 domain, comprising three β-strands and two α-helices positioned by the two Zn2+ ions that are coordinated by the conserved cysteine and histidine residues (Fig. S5 C). The S185 residue is located between the two N-terminal β-strands, and its phosphorylation could either directly or indirectly regulate Gβγ binding as part of the binding motif or by inducing a conformational change. The NMR-based analysis provides strong evidence that phosphorylation of Ste5 at S185 regulates its interaction with Gβγ and thereby inhibits signaling by preventing membrane recruitment of the scaffold protein.
Pkc1-dependent phosphorylation of Ste5 is required to turn off pheromone signaling to prevent lysis during mechanostress and cell–cell fusion
We quantified by FACS analysis the transcriptional induction of the pFIG1-qV reporter in the presence of either WT Pkc1 or a dominant-active mutant (Pkc1R398A). After α-factor exposure in the presence of WT Pkc1, ste5Δ cells harboring untagged WT Ste5 or Ste5S185A strongly induced expression from the FIG1 promoter, while cells expressing Ste5S185D did not display induction, even after prolonged exposure (Fig. 5 A). In contrast, upon expression of Pkc1R398A, induction of pFIG1-qV was reduced (Fig. 5 B). Importantly, this reduction was at least partially alleviated in cells expressing the nonphosphorylatable (Ste5S185A) mutant as the only Ste5 copy. We next examined whether Pkc1-mediated crosstalk is physiologically important in vivo. Indeed, α-factor treated ste5S185A cells exposed to mechanical stress showed increased cell lysis compared with WT controls (Fig. 5 C), and this defect was accompanied by reduced ability to remove Ste5-tV from shmoo tips (Fig. 5 D). We conclude that phosphorylation of Ste5 in a Pkc1-dependent manner inhibits pheromone signaling, thereby reducing polarized growth at shmoo tips to prevent cell lysis.
Mechanical cell wall stress occurs during the mating response, in particular during cell–cell fusion. To test whether Pkc1-dependent crosstalk prevents cell lysis during mating, we performed quantitative mating assays. Cells expressing the nonphosphorylatable Ste5S185A mutant mated with reduced efficiency compared with WT controls (Fig. 6 A), which could be partially rescued by sorbitol (Fig. S5 D), indicative of a CWI defect. To examine whether this defect results from problems during cell–cell fusion, we designed a microfluidic device to trap mating-competent a- and α-cells to visualize the entire mating process and specific marker proteins by fluorescence microscopy in single cells (Fig. 6 B). Lysis during pheromone-induced cell–cell fusion was increased in cells expressing either Ste5S185A or Far13A (Fig. 6 C). When both mating partners express Ste5S185A and/or Far13A the lysis phenotype was exacerbated, implying that reduced pheromone signaling at the fusion site is necessary in both mating partners. Concomitantly, the time and position of the first bud emergence in viable zygotes was altered in Ste5S185A-expressing cells (Fig. 6 D), and this cell cycle reentry delay worsened in ste5S185A far13A double mutants. To corroborate these results, we compared Ste5 disappearance from the fusion site in ste5Δ cells expressing either GFP-tagged WT or nonphosphorylatable Ste5S185A. As expected, WT Ste5-GFP accumulated at sites of cell–cell contact and rapidly dispersed to the cytoplasm concomitant with cell–cell fusion (Fig. 6 E). However, GFP-Ste5S185A persisted at sites of cell–cell fusion (Fig. 6, F and G), suggesting that phosphorylation of S185 contributes to Ste5 turnover at the fusion site. Taken together, we propose that Pkc1 activated by mechanical stress restricts pheromone signaling during cell–cell fusion at least in part by triggering dissociation of Ste5 and Far1 from the fusion site to facilitate local cell wall remodeling and prevent cell lysis.
The scaffolds Ste5 and Far1 comprise important regulatory nodes that spatially and temporally orchestrate yeast mating. While their functions in establishing mating signaling and polarization are comparatively well understood, much less is known about their role in modulating signaling in stressed mating cells. Here, we show that direct phosphorylation of the RING-H2 domains of Ste5 and Far1 regulates their interaction with Gβγ in cells exposed to mechanical stress. Interestingly, this crosstalk mechanism is functionally important both during cell–cell fusion and upon cell wall compression by external mechanical stress. Although the CWI pathway is activated, its MAPK Mpk1 is mostly dispensable for this crosstalk. Instead, we found that Pkc1 triggers removal of Ste5 and probably Far1 from sites of polarized growth, thereby inhibiting pheromone signaling and polarized growth to prevent cell lysis.
Multiple mechanisms and kinases regulate Ste5 membrane association and modulate signaling
Previous work identified several phosphorylation-dependent mechanisms that regulate Ste5 and Far1 activity with altered environmental conditions, and pheromone activation tunes Ste5 output with positive and negative feedback loops. For example, our MS uncovered phosphorylation of four sites near the MAPK-docking site in Ste5, including T287, which was previously proposed to be targeted by Fus3 as part of a negative feedback circuit (Bhattacharyya et al., 2006; Malleshaiah et al., 2010), although this conclusion was recently challenged (Winters and Pryciak, 2019). All four sites are followed by prolines, including S276 located on the same tryptic peptide as T287. It is thus possible that analogous to T287, phosphorylation of these MAPK consensus sites may allosterically contribute to negative feedback regulation in the pheromone-signaling pathway.
Multiple kinases regulate Ste5 membrane association to perturb signaling. It is well established that CDK1 phosphorylates several residues within an amphipathic helix in the N-terminal domain of Ste5 (Strickfaden et al., 2007) and thereby explains why pheromone signaling is restricted to the G1 phase of the cell cycle (Oehlen and Cross, 1994). These and other sites in the PH domain are also phosphorylated by Fus3, constituting a negative feedback mechanism that regulates membrane turnover of Ste5 and tunes down its signaling activity during mating (Repetto et al., 2018). Our analysis identified additional phosphorylation sites on Ste5 and Far1 that conform to the minimal MAPK consensus sites, including T456 located within the Ste5 PH domain. Indeed, mutating T456 to a phospho-mimicking aspartic acid residue (Ste5T456D) reduced signaling output (unpublished data), while Ste5 signaling capacity was unchanged when the same site was mutated to a nonphosphorylatable valine residue. Thus, phosphorylation of multiple sites within distinct membrane-binding domains comprises negative feedback that cooperatively modulates the residence time of Ste5 at the plasma membrane. This multisite-phosphorylation mechanism is expected to set a threshold constraining signaling activity during mating and preventing Ste5 activation under conditions of high CDK1 activity in the cell cycle.
Pkc1 directly phosphorylates Ste5 S185 in the RING-H2 domain and prevents membrane localization by interfering with Gβγ binding
We identified a novel mechanism for how single-site phosphorylation regulates Ste5 membrane turnover in response to mechanical stress. Indeed, mutating serine 185, located in its RING-H2 domain, to the phosphomimetic aspartic acid directly interferes with Gβγ binding, suggesting that phosphorylation is sufficient to disrupt binding and that a similar mechanism may regulate the polarity scaffold Far1. Consistent with this notion, cells expressing Ste5S185D are unable to signal, and the mutant protein is not recruited to the plasma membrane upon pheromone exposure. This signaling defect can be rescued by artificially tethering the Ste5S185D mutant protein to the plasma membrane, demonstrating that the defect is mainly caused by abolished membrane recruitment. Mechanistically, phosphorylation of S185 within the RING-H2 domain may directly prevent Gβγ binding through electrostatic repulsion. Alternatively, NMR analysis revealed an induced-fit mechanism such that upon Gβγ binding, the Ste5 RING-H2 domain adopts a fold that may include an asymmetric dimerization typical for many known RING-H2 E3 ligases (Yudina et al., 2015; Sanchez et al., 2016). S185 phosphorylation may thus interfere with folding and prevent stabilization of the active conformation. Irrespective of the underlying mechanism, phosphorylation of a single site in the RING-H2 domain of Ste5 blocks its signaling function by interfering with Gβγ binding and thus membrane association. While we do not know the stoichiometry of S185 phosphorylation, this regulatory mechanism could function as an on/off switch at shmoo tips rapidly activated during stress responses rather than a global tunable rheostat typically observed for multisite phosphorylation.
Although we were unable to confirm that S185 phosphorylation depends on Pkc1 or mechanical stress in vivo, several lines of evidence suggest that the RING-H2 domain of Ste5 is directly phosphorylated by Pkc1. First, Pkc1 is activated by mechanical stress, and its activity is required to prevent cell lysis by inhibiting Fus3. Second, Pkc1 and Ste5 colocalize at shmoo tips and the site of cell–cell fusion. Third, Ste5 physically interacts with Pkc1, and this interaction increases during prolonged α-factor treatment. Fourth, S185 is phosphorylated by purified Pkc1 in vitro and S185 is phosphorylated in vivo in α-factor–treated cells. Finally, expression of a constitutively activate Pkc1 mutant interferes with FIG1-qV reporter expression in WT cells but much less in ste5S185A mutant strains, and ste5S185A cells are prone to lyse upon mechanical stress and during cell–cell fusion. Taken together, these data suggest that mechanical stress activates Pkc1, which in turn phosphorylates Ste5 on S185 located within the RING-H2 domain, thereby blocking its Gβγ-mediated membrane recruitment. While Pkc1 likely down-regulates Ste5 activity in response to mechanical stress, it is possible that other kinases similarly use S185 phosphorylation to inhibit pheromone signaling in response to other intrinsic or extrinsic stress conditions.
Pkc1-mediated inactivation of the pheromone-response pathway is required to prevent cell lysis of cells experiencing external or intrinsic mechanical stress
Cell wall stress activates the CWI signaling pathway, which through Pkc1 and its downstream MAPK Mpk1 rapidly prevents actin polarization and polarized growth (Levin, 2011). Mpk1 also regulates the production and assembly of cell wall components such as chitin β-1,3-glucan and several mannoproteins and thus repairs and strengthens the protective cell wall (Roemer et al., 1994). Recent evidence demonstrates that compressive mechanical stress is sensed by Mid2, which in turn rapidly triggers Pkc1 activation to prevent cell lysis during bud emergence and shmoo formation (Mishra et al., 2017). As expected, pheromone-treated mpk1Δ cells are prone to lyse upon mechanical stress. However, our results identified an additional, Mpk1-independent function of Pkc1 in regulating the scaffolds Ste5 and Far1 to tune down pheromone signaling. Indeed, Pkc1 triggers removal of Ste5 from shmoo tips, resulting in reduced Fus3 activity. Mechanostress-induced cell lysis of shmooing cells in the absence of Pkc1 activity is prevented by simultaneous inactivation of Fus3. Thus, analogous to Hog1 in response to high osmolarity (Hall et al., 1996), this crosstalk mechanism is functionally important to protect cells from lysis when exposed to mechanical stress.
Pkc1-mediated Ste5 phosphorylation also comprises an intrinsic regulatory mechanism temporally and spatially orchestrating mating. Indeed, in cell–cell fusion, cells have to locally break down their cell wall to allow membrane fusion (Merlini et al., 2013). These two steps are carefully coordinated, as early engagement could trigger osmotic shock and lysis. How proper timing is established remains unknown, but it has been proposed that cell fusion depends on Fus3 activity and requires particularly high levels of pheromones (Brizzio et al., 1996). In addition, cells engage an unknown protection pathway that results in Pkc1 activation to antagonize cell wall reorganization if cells are osmotically unstable or morphology is disrupted (Philips and Herskowitz, 1997). Using a microfluidic chip, we found that this Pkc1-dependent protection mechanism involves inhibition of pheromone signaling and polarized growth by inactivation of Ste5 and Far1. Pheromone signaling down-regulation may be necessary to allow efficient cell wall repair and reestablish osmobalance before the fusion process can continue. In absence of this mechanism, sustained Fus3 and Cdc24 activity with cell wall stress at the site of cell fusion may uncouple polarized growth from cell wall remodeling, ultimately leading to cell lysis. Alternatively, removal of Ste5 and Far1 from the fusion site may directly impact the fusion machinery. Indeed, Ste5 physically interacts with the membrane protein Fus1 (Nelson et al., 2004), which is required for polarization and efficient cell wall remodeling during the cell–cell fusion process. Taken together, our data suggest that crosstalk between Pkc1 and the pheromone signaling pathway coordinates cell–cell fusion and prevents untimely cell wall remodeling during the process.
Materials and methods
Yeast strains, plasmids, and growth conditions
All yeast strains and plasmids are listed in Tables 1 and 2, respectively. Yeast strains are derivatives of BY4741 (Brachmann et al., 1998) or w303. Gene fusions were generated by homologous recombination–based replacement of the endogenous gene, and expressed from their endogenous promoter unless otherwise indicated. Strains for all experiments were grown in synthetic media (0.17% yeast nitrogen base, 2% glucose, 0.5% NH4-sulfate, and amino acids). α-Factor (Genscript) was used at 2.7 µM concentration, unless indicated otherwise. To induce expression of proteins from the GAL1-promoter, 2% galactose was added for 2 h unless indicated otherwise to cells growing logarithmically in media containing 2% raffinose were added for 2 h.
Protein extracts and Western blotting
Protein extracts were prepared from TCA-fixed cells. TCA pellets were resuspended in 2× urea buffer (62.5 mM Tris, pH 6.8, 10% glycerol, 4% SDS, 5% β-mercaptoethanol, 8 M urea, and bromophenol blue) and vortexed 3 min at 4°C with 0.5-mm glass beads. After boiling, samples were analyzed using standard SDS-PAGE and Western blotting procedures. Antibodies used included α-GFP (11 814 460 001; Roche), α-PGK1 (A6457; Invitrogen), α-GST (G1160; Sigma), α-His (H1029; Sigma), α-me3K9H3 (NBP1-30141; Novus), GαR-HRP (170-6515; Biorad), and GαM-HRP HRP (170-6516; Biorad). The signal was visualized using SuperSignal West Pico PLUS (1863096; Thermo Fisher Scientific) and film (SuperRX, 47410 19236; Fuji). Protein extracts for in vitro kinase assays were prepared as follows: exponentially growing yeast cells were lysed by freezer milling (Kraft et al., 2012) and the powder was resuspended in PBS buffer with 3 mM EDTA, Roche Inhibitor Tablet, Sigma Inhibitors of yeast proteases, 0.5% Triton X-100, and 2 mM DTT. The extract was cleared by centrifugation at 235,000 ×g, and equal amounts of the supernatant were used for the in vitro kinase assays.
Microscopy, microfluidics, and image analysis
Images were acquired on fully automated inverted epifluorescence microscopes (Ti-Eclipse; Nikon) in an incubation chamber set to 30°C with Nikon CFI Plan Apochromat 60×, NA 1.4 objective, a Hamamatsu ORCA Flash 4.0 camera and Micromanager 1.4, and appropriate excitation and emission filters. A motorized XY-stage and piezo drive was used to acquire Z-stacks and multiple fields of view per time point.
Crosstalk between mechanostress and pheromone signaling was studied using a microfluidic device that can apply compressive mechanical stress (Mishra et al., 2017). MATa cells growing in mid-log phase were treated with 2.7 µM α factor for 100 min and loaded in the microfluidic chip, which was precoated with 1 mg/ml Con A. Where appropriate, cells were pretreated for 15 min with the solvent DMSO or, as indicated, with cercosporamide (7.5 µM; Merck) or NaPP1 (5 µM; Tocris Bioscience). Subsequently, mechanostress (7 psi pressure) was applied, and the cellular response was assessed using bright-field as well as fluorescence microscopy. Z-stacks of images were acquired for fluorescent reporters, and the images were analyzed on maximum intensity projections.
Cell viability was assessed using 0.025% trypan blue staining or loss of fluorescence after 30 min of mechanostress. For quantification, redistribution of Ste5-tV or Pkc1-GFP away from shmoo tips within 30 min of mechanopressure was scored as Ste5 or Pkc1 dispersal, respectively. Specifically, the intensities of 4 × 4 pixels (16 pixels square) in the shmoo and brightest area elsewhere were compared and scored as “dispersed” when the reduction at the shmoo area was reduced more than the control area. The number of cells that dispersed Ste5 or Pkc1 from shmoo tips within 30 min of applying mechanical pressure was compared with the number of cells that failed to do so. The Ste5 dispersal was analyzed for cells after induction of Pkc1K398A expression by 2% galactose in 96-well plates. In these experiments, cells with pronounced enrichment of Ste5-tV at shmoo tips were monitored every 10 min, and cells that lost Ste5-tV from shmoo tips within 2 h after addition of 2% galactose were scored manually as cells with Ste5 dispersal. Loss of Ste5-tV from shmoo tips was further confirmed by loss of polarized growth. In experiments involving NaPP1-mediated rescue of cell lysis, cells that had Ste5-tV enriched at shmoo tips but lost total fluorescence and changed bright-field contrast within 30 min of mechanostress were scored as dead cells. Cells with mating protrusions after 100 min of pheromone treatment were counted as shmooing cells.
Fus3 activity was monitored using the SKARS as described previously (Durandau et al., 2015). Briefly, cells displaying cytosolic localization of the reporter after pretreatment with α factor were monitored for 30 min under mechanostress conditions. Cells that showed enrichment of the reporter in the nucleus when subjected to pressure were scored as Fus3-inhibited cells, while cells that lysed and/or failed to enrich the reporter in the nucleus were scored as Fus3-active cells. In the lysed or protected cells, determined by the integrity of the nucleus (Hta2-CFP signal), Fus3 activity was scored based on SKARS localization. Cells that have SKARS in the cytosol before lysis or during 30 min of mechanostress in protected cells were scored as cells with high Fus3 activity. At least three independent experiments were quantified, with 50 cells analyzed for each condition. Error bars indicate SEM.
Microscopy was performed on the 96-well plates coated with 1 mg/ml Con A (Sigma) when a microfluidic platform was not used. To show that cercosporamide treatment fails to inhibit Fus3, WT cells expressing the Fus3 SKARS were allowed to adhere onto Con A–coated 96-well plates, and DMSO or drugs were applied with or without α factor for the indicated time. For assessing PKA and TORC1 activity, cells expressing GFP-tagged Gat1 or Sfp1 were treated with DMSO, 7.5 µM cercosporamide, or 200 nM rapamycin (LC Laboratories). Moreover, WT cells harboring the PKA reporter Msn2-NLS-GFP (Görner et al., 2002) were treated with DMSO or 7.5 µM cercosporamide or subjected to glucose starvation for 60 min. The localization of the GFP reporter was monitored microscopically every 10 min for 60 min. Note that the plates were not coated with Con A while monitoring Msn2-NLS-GFP dynamics.
Cell orientation assays in pheromone gradients were performed in homemade microfluidic gradient chips as described previously (Hegemann et al., 2015). Experimental cells were mixed with unlabeled WT control cells to internally control for gradient stability. For mating assay, MATa and MATα cells grown in exponential phase were mixed before the experiment and immediately loaded into the chip. All microfluidics devices (gradient and mating) were produced by soft lithography and the detailed process of wafer fabrication, polydimethylsiloxane baking, and bonding process were done as described previously (Lee et al., 2012). The height of the cell culture chamber in the mating chips (5 µm) is similar to that of yeast cells ensuring an identical focal plane to image cell mating and fusion for microscopy during the entire mating process. We continuously provide fresh media with very mild flow (0.2 µl/min), and the cells are trapped within close proximity so that they can efficiently mate with each other.
Automated image analysis was performed using YeastQuant software on raw images (Pelet et al., 2012) running in MATLAB. For Fig. 4 B, the Z-stack of images (13 steps with 0.225 µm distance) are projected using summation intensity and analyzed using Fiji (Schindelin et al., 2012). Yeast cell mating events were monitored by time-lapse microscopy, successively capturing images. The moment of cell death was identified by the sudden shrinkage of the cell body. Cell viability was defined as the ratio of the total number of mating pairs and cell death events.
Saturated overnight cultures in synthetic media were diluted and grown to early log phase (OD600 0.4). Where appropriate, genes under the GAL1 promoter were induced by addition of galactose to a final concentration of 2%. For the estradiol induction, strains containing the ADGEV constructs were treated with the indicated concentration of estradiol for 2 h. The pheromone response pathway was activated by the addition of 2.7 µM α-factor. Protein translation was stopped at the indicated times by the addition of cycloheximide (0.1 mg/ml). Fluorescence intensity of 10,000 cells was measured by FACS (FACSCalibur; 488 nm excitation 530 nm emission) 3 h after pathway induction. Customized MATLAB scripts were used to analyze the data.
Yeast two-hybrid and quantitative β-galactosidase assays
The two-hybrid assays were performed essentially as described previously (Möckli and Auerbach, 2004). Briefly, for each interaction pair and controls, several colonies were grown to an OD600 of 0.5–0.8. One absorbance unit of yeast cultures was pelleted, and cells were lysed, resuspended in 20 µl water, and then transferred to a transparent flat-bottom 96-well plate. 100 µl PXG buffer containing Bluo-gal (Invitrogen) was added, and the absorbance at 420 nm was measured with a flatbed scanner. Images were analyzed using ImageJ (National Institutes of Health).
Histidine-biotin tandem affinity purifications are based on methods described elsewhere (Tagwerker et al., 2006; Reiter et al., 2012), with the following modifications. Cells expressing either Ste5 C-terminally tagged with a modified HTB tag (HTBn: 12xHis-tag and two additional TEV cleavage sites) or HTBeaq-tagged Far1 (Reiter et al., 2012) were grown to mid-logarithmic phase (OD600 = 0.6–0.9), treated with 100 nM α-factor for the times indicated, harvested by filtration, and rapidly deep frozen in liquid N2. All subsequent steps were performed as previously described. In the case of in-gel digestion, HTB purified proteins bound to Streptavidin-Agarose resin (Thermo Fisher Scientific) were eluted by a 5-min incubation at 95°C in 1× Laemmli buffer. Proteins were separated on Novex Bolt 4–12% Bis-Tris Plus gels (Invitrogen) according to the manufacturer’s protocol. Protein bands were visualized using SimplyBlue SafeStain (Invitrogen) and further processed as described previously (Reiter et al., 2013). For Far1-Venus purification, cells were grown to mid-log phase and treated for 2 h with 2.87 μM α-factor. Cells were washed in PBS, rapidly frozen in liquid nitrogen, and later lysed using freezer milling (Kraft et al., 2012). The lysed powder was solubilized in xt3 buffer (10 mM Tris-HCl, pH 7.5, 150 mM KCl, 2 mM MgCl2, 0.1% NP-40, 1 mM DTT, and complete protease inhibitors, with EDTA [Roche], 100 µM PMSF, 1 mM Na3VO4, 10 mM Na4P2O7, 10 mM NaF, and 10 mM β-glycerophosphate) and centrifuged at 4,500 rpm for 5 min, and the remaining supernatant was centrifuged at 20,000 rpm for 10 min (both at 4°C). The cleared lysate was incubated with GFP-trap beads (Chromototek), washed extensively in xt3, and washed extensively in xt3 without detergent, and finally protein was eluted from beads using 200 mM glycine, pH 2.0. The eluate was neutralized in 1.5 M Tris, pH 9.2, and stored at −80°C before processing for MS analysis. In-solution digestion with trypsin was performed as described previously (Reiter et al., 2012). All tryptic peptides samples were subjected to a second reduction step, applying 30-min incubation with DTT (160 µg) at 56°C. Enrichment using TiO2 and MS analysis is based on methods described previously (Reiter et al., 2012). We used a LTQ FT Ultra MS (Thermo Fisher Scientific) and a LTQ Orbitrap Velos (Thermo Fisher Scientific) instrument for phosphorylation site mapping with settings as described previously (Reiter et al., 2012). Gel-purified protein samples were analyzed using a LTQ XL MS instrument (Thermo Fisher Scientific). MS raw files were processed using MaxQuant (Cox and Mann, 2008; Cox et al., 2011) software version 220.127.116.11 using standard settings, except the following modifications. Spectra were searched against the Saccharomyces Genome Database database (http://www.yeastgenome.org/) containing 6,717 entries (February 3, 2011) and including a list of 248 common laboratory contaminants as well as reversed versions of all sequences. The enzyme specificity was set to trypsin. A maximum of two missed cleavages was allowed. Phosphorylation of serine, threonine, and tyrosine residues, oxidation of methionine, and deamidation of asparagine was searched as variable modification. For stable isotope labeling using amino acids in cell culture (SILAC)–labeled samples, Lys6 and Arg6 were additionally selected. Carbamidomethylation of cysteine was searched as a fixed modification. A maximum of five modifications per peptide was allowed. The false discovery rate for peptide, protein, and site identification was set to 1%. All files except Exp1 were searched together. Minimum delta score for modified peptides was set to 6. The six raw files of Exp1 were searched separately with the same parameters as described previously (Reiter et al., 2012). The MS proteomics data have been deposited at the ProteomeXchange Consortium (Vizcaíno et al., 2013; http://www.proteomexchange.org) via the Proteomics Identifications (PRIDE) partner repository with the dataset identifier PXD004657.
M-track protein–protein proximity assay
Cells expressing HKMT-myc–tagged Ste5 (bait) and protA-H3–tagged Pkc1 (prey) were grown until mid-log phase; treated with α-factor (0.5 µg/ml final concentration) for 0, 30, and 60 min; and harvested by centrifugation. Protein extraction was performed under denaturing conditions (50 mM Tris/HCl, pH 8, 50 mM sodium phosphate buffer, pH 8, 8 M urea, 0.3 M NaCl, and 0.5% Nonidet P-40) by glass bead lysis. Whole-cell extracts were mixed with 2× urea buffer (62.5 mM Tris, pH 6.8, 10% glycerol, 4% SDS, 5% β-mercaptoethanol, 8 M urea, and bromophenol blue). After boiling, samples were analyzed using standard SDS-PAGE. Histone-H3 Lys 9 trimethylation (me3K9H3) of Ste5-protA-H3 was visualized by Western blot using an antibody recognizing me3K9H3 (NBP 1-30141; Novus Biochemicals). Loading was controlled using an antibody (12CA5) recognizing HA.
Protein purifications, kinase assays, and NMR analysis
Expression and purifications of the proteins used for NMR investigation followed protocols previously described (Walczak et al., 2014). Briefly, His-Z–tagged Ste5149–238 was expressed in E. coli BL21(DE3) cells that were lysed in binding buffer (500 mM NaCl, 50 mM Tris-HCl, 10% glycerol, 2 mM DTT, pH 8.0, EDTA-free Roche Inhibitor Cocktail tablet, 150 µM PMSF, 100 µM ZnCl2, and 4.5 mM imidazole) and bound to an Ni-NTA column. After washing with increasing amounts of imidazole (20, 60, and 100 mM) in binding buffer, Ste5149–238 was eluted with 300 mM imidazole in binding buffer before dialysis into 50 mM Tris-HCl, pH 8.0, and 2 mM DTT. Cleavage of the His-Z tag took place during dialysis using 3C PreCission Protease. Subsequently, Ste5149–238 was purified over an S-Sepharose column and dialyzed into NMR buffer (100 mM NaCl, 50 mM Tris-HCl, pH 7.2, and 100 µM ZnCl2).The GST-Ste4–Ste18 heterodimer was expressed in yeast cells that were lysed by freezer milling (Kraft et al., 2012). The lysed powder was dissolved in GST-binding buffer (PBS buffer with 3 mM EDTA, Roche Inhibitor Tablet, Sigma Inhibitors of yeast proteases, 0.5% Triton X-100, and 2 mM DTT), cleared by spinning at 235,000 ×g and bound to glutathione-S-transferase resin for 5–6 h. 6His-tagged Ste5 RING-H2 fragments were incubated at 4°C for 1 h on a rotating wheel, the resins washed with NMR buffer, and bound proteins were eluted with urea buffer. For NMR experiments, GST-Ste4–Ste18 was eluted with 40 mM reduced Glutathione and applied to a Superdex 75 size exclusion column using NMR buffer as described above.
The in vitro kinase assays for NMR analysis were performed with 150 µM Ste5149–238 (in 250 µl) with either assay buffer (3 mM MgCl2 and 1 mM ATP), 5 µl of PKCα (P61-10G; SignalChem), and 0.5-0.8× of Lipid Activator (L51-39, SignalChem) or 250 µl yeast extract complemented with 5 mM ATP and phosphatase inhibitors. Reaction proceeded immediately at room temperature. Heteronuclear correlation NMR experiments were acquired as previously described (Vuister et al., 1991; Schanda et al., 2005). The spectra were measured on Bruker AvanceIII NMR spectrometers equipped with cryogenically cooled probes. The NMR data were processed with the software TOPSPIN 3.2 (Bruker). For in vitro kinase assays using autoradiography detection, His-Z–tagged Ste5- and Far1-RING-H2 domains were expressed and purified as described above, but not cleaved. GST-Pkc1 was affinity purified from yeast extracts using glutathione-S-transferase resin, and the kinase assays were performed in the presence of γ-P32-ATP as described previously (Drogen et al., 2000).
Quantitative mating assays
The quantitative mating assays were essentially performed as described previously (Chenevert et al., 1994). Briefly, 106 of either MATa WT or Ste5S185A cells were mixed with 10 × 106 MATα cells, filtered onto 0.45-µm filters, and incubated on permissive YPD (where indicated containing 0.2 M sorbitol) plates for 60 min. Cells were resuspended and plated on SD plates allowing either only growth of MATa cells or diploids. The total number of colony-forming units was determined. All experiments were performed in triplicate.
The results of at least three independent experiments are presented as mean values, and the error bars represent SD or SEM, as indicated in the figure legends. Statistical significance was tested using the two-tailed Student’s t test in Microsoft Excel, and the results are indicated (***, P 0.001; **, P 0.01; *, P 0.05). In addition, we performed one-way ANOVA followed by post-test comparison (Dunn’s method) for the data shown in Fig. 6, D and G. The results are comparable to the ones from the t test, and the differences in the median values among the groups are greater than would be expected by chance.
Online supplemental material
Fig. S1 shows the specificity of cercosporamide toward Pkc1. Fig. S2 shows that Pkc1 activation inhibits the pheromone response. Fig. S3 shows mapping of phospho-sites on Ste5 and Far1. Fig. S4 shows NMR-based in vitro phosphorylation assays. Fig. S5 shows two-hybrid analysis of the Far1-RING-H2 domain with Ste4, ste5S185A mating efficiency and schematic representation of the Ste5 RING-H2 domain structure bound to Gβγ. Table S1 lists confidently assigned phosphorylation sites of Ste5. Table S2 lists confidently assigned phosphorylation sites of Far1.
We thank J. Tilma, I. Stoffel, J. Augustine, and the D-BIOL NMR platform for technical assistance, S. Honnappa (Novartis, Basel, Switzerland) and P.M. Pryciak (University of Massachusetts Medical School, Worcester, MA) for plasmids, and D. Hollenstein and M. Hartl for help with MS data analysis. We thank T. Mayor and A. Smith for critical reading of the manuscript and the Ammerer, Wider and Peter laboratories for helpful discussions.
This work was supported by the Christian Doppler Forschungsgesellschaft (W. Reiter and I. Dohnal), the European Commission Seventh Framework Project UNICELLSYS (W. Reiter and I. Dohnal), the ETH (grant ETH-23 10-2; M.J. Walczak), and the Swiss National Science Foundation Marie-Heim Vögtlein fellowship (Z. Yudina). Work in the M. Peter laboratory is supported by a European Research Council award, a Swiss National Science Foundation project grant, ETH, and the National Research Foundation of Korea Global Research Laboratory (NRF-2015K1A1A2033054).
The authors declare no competing financial interests.
Author contributions: F. van Drogen, R. Mishra, S.S. Lee, B. Hegemann, S. Pelet, Z. Yudina, and F. Rudolf performed the yeast, biochemistry, and quantitative microscopy experiments. S.S. Lee and R. Mishra designed and constructed the microfluidic device. F. Rudolf, W. Reiter, I. Dohnal, and G. Ammerer carried out the MS analysis, and M.J. Walczak, G. Wider, A. Binolfi, and P. Selenko performed the NMR experiments. F. van Drogen, R. Mishra, S.S. Lee, B. Hegemann, F. Rudolf, Serge Pelet, G. Wider, G. Ammerer, and M. Peter participated in the experimental design. F. van Drogen, R. Mishra, and M. Peter wrote the manuscript, with critical input from all authors.
F. van Drogen and R. Mishra contributed equally to this paper.
F. Rudolf’s present address is Dept. of Biosystems Science and Engineering, ETH Zürich, Basel, Switzerland.
M.J. Walczak’s present address is Captor Therapeutics Ltd., Wroclaw, Poland.
I. Dohnal’s present address is BIOMIN Research Center, Tulln, Austria.