ABSTRACT
The fission yeast small GTPase Rho2 regulates morphogenesis and is an upstream activator of the cell integrity pathway, whose key element, mitogen-activated protein kinase (MAPK) Pmk1, becomes activated by multiple environmental stimuli and controls several cellular functions. Here we demonstrate that farnesylated Rho2 becomes palmitoylated in vivo at cysteine-196 within its carboxyl end and that this modification allows its specific targeting to the plasma membrane. Unlike that of other palmitoylated and prenylated GTPases, the Rho2 control of morphogenesis and Pmk1 activity is strictly dependent upon plasma membrane localization and is not found in other cellular membranes. Indeed, artificial plasma membrane targeting bypassed the Rho2 need for palmitoylation in order to signal. Detailed functional analysis of Rho2 chimeras fused to the carboxyl end from the essential GTPase Rho1 showed that GTPase palmitoylation is partially dependent on the prenylation context and confirmed that Rho2 signaling is independent of Rho GTP dissociation inhibitor (GDI) function. We further demonstrate that Rho2 is an in vivo substrate for DHHC family acyltransferase Erf2 palmitoyltransferase. Remarkably, Rho3, another Erf2 target, negatively regulates Pmk1 activity in a Rho2-independent fashion, thus revealing the existence of cross talk whereby both GTPases antagonistically modulate the activity of this MAPK cascade.
INTRODUCTION
Protein S-acylation, also named protein palmitoylation, is a specific protein lipidation involving the thioesterification of selected cysteine residues within the target protein to the 16-carbon fatty acid palmitate (1). A number of proteins are palmitoylated in vivo in eukaryotes, from simple organisms like the yeast Saccharomyces cerevisiae to animal and human cell lines (2). Research on protein palmitoylation has drawn interest because of its regulatory and dynamic function and because some palmitoylated proteins are key players in the signaling mechanisms controlling cell proliferation, differentiation, and/or response to external stimuli (3). Prime examples of this control are members of the Ras and Rho family of small GTPases (4–7), which bind guanine nucleotides (GDP or GTP) and harbor intrinsic GTPase activity to hydrolyze the bound GTP (8). Guanine nucleotide exchange factors (GEFs) promote GTPase activation through dissociation and replacement of GDP by GTP (9). In addition, GTPase downregulation is exerted by GTPase-activating proteins (GAPs), which activate intrinsic GTPase activity by enhancing hydrolysis of GTP to GDP, and by GDP dissociation inhibitors (GDIs), which favor GTPase seizure to the cytosol in an inactive state (9).
Most Ras and Rho family GTPases undergo sequential lipid modifications for proper targeting to cellular membranes, which are essential for their biological activity (6, 10, 11). The first event of this sequence involves the covalent linkage of either farnesylpyrophosphate or geranylgeranylpyrophosphate to a cysteine residue located at a conserved C-terminal tetrapeptide motif known as the CAAX box (where A indicates an aliphatic residue and X is usually Ser, Met, Cys, Gln, Leu, or Ile) (2, 3). Then the AAX tripeptide is removed from the CAAX box by proteolytic cleavage, and the free carboxyl group of the isoprenylated cysteine is methylated by a specific isoprenylcysteine-O-carboxyl methyltransferase (12). However, in most cases additional motifs are needed to enhance and stabilize the membrane association of prenylated Ras and Rho GTPases. One is a cluster of polybasic amino acids located at a hypervariable region upstream of the CAAX box that favors electrostatic interaction with acidic membrane lipids (2). A second feature is the presence of one or two cysteine residues that become palmitoylated in vivo by a group of enzymes known as palmitoyltransferases (PTs), which locate to the endoplasmic reticulum and the Golgi region (3). Protein palmitoylation is a dynamic and reversible process, thus providing an attractive biological mechanism to compartmentalize both membrane targeting and signaling. Examples of the effects of palmitoylation dynamics on GTPase membrane sorting and function are the H- and N-Ras isoforms (6, 10) and Rho family small GTPases RhoB, TC10/RhoQ, and Rac1 (4, 5, 7).
The fission yeast Schizosaccharomyces pombe, a simple eukaryote whose signaling pathways show significant functional homology with those of higher cells (13), has a single Ras ortholog (Ras1), which is palmitoylated in vivo by Erf2 palmitoyltransferase (14, 15, 16). Notably, Ras1 signaling is spatially segregated so that cellular morphogenesis is regulated by an unpalmitoylated GTPase localized to the endomembranes, whereas mating pheromone signaling is dependent on palmitoylated Ras1 located on the plasma membrane (15). The hypothesis of compartmentalized Ras signaling has found strong support in recent studies performed with Ras orthologs from several organisms, from fungi (17–19) to higher eukaryotic cells (20). Fission yeast also contains six members of the small Rho GTPase family, namely, Rho1 to Rho5 and Cdc42. Among them, Rho2 GTPase is not essential but promotes the biosynthesis of the cell wall (1-3)α-d-glucan by activating α-glucan synthase Mok1 via the protein kinase C (PKC) ortholog Pck2 (21, 22). Moreover, Rho2 and Pck2 are the main positive regulators operating upstream of the cell integrity mitogen-activated protein kinase (MAPK) pathway (23, 24). This cascade, whose key element is the extracellular signal-regulated kinase (ERK)-type MAPK Pmk1, becomes activated in response to a variety of external stimuli (25) and regulates cell wall integrity maintenance, vacuole fusion, cytokinesis, and ionic homeostasis (25–28). In contrast to the situation in S. cerevisiae, where both Rho1 GTPase and Pkc1 are essential for the activation of MAPK Slt2/Mpk1 (29), Pmk1 can be activated in the absence of either Rho2 or Pck2, suggesting the existence of redundant and unknown upstream regulatory elements (24). Indeed, we have recently reported that Rho1 and Pck1 can also activate the Pmk1 pathway under very specific circumstances (30).
In this work we report that Rho2 undergoes in vivo palmitoylation at cysteine-196 within its C terminus to mediate its full targeting to the plasma membrane and have conducted a comprehensive study to investigate the biological significance of this lipid modification. Importantly, Rho2-dependent control of morphogenesis and signaling to the cell integrity pathway is exerted exclusively at the plasma membrane and not at other membrane compartments. While Erf2 PT is the major protein responsible for palmitate transfer to Rho2, our results reveal the existence of an interfering mechanism by which Rho3, another Erf2 substrate, negatively modulates Rho2-Pmk1 signaling independently of Rho2.
MATERIALS AND METHODS
Strains, media, growth conditions, and gene disruption.The S. pombe strains used in this work (listed in Table 1) derive from control strain MI200, which expresses a genomic Pmk1-HA6H fusion (24, 25). They were grown with shaking at 28°C in rich (yeast extract-sucrose [YES]) or Edinburgh minimal medium (EMM2) with 2% glucose and supplemented with adenine, leucine, histidine, or uracil (100 mg/liter; Sigma Chemical) (31). Mutant strains were obtained either by standard transformation procedures or by mating followed by random spore analysis (31). Cells transformed with pREP3X-based plasmids expressing different GTPase constructs were grown in liquid EMM2 with thiamine (5 mg/liter) and either plated in the same medium with or without the vitamin (viability assays) or transferred to EMM2 lacking thiamine for 16 to 24 h for Pmk1 activation assays and cell morphology scoring (see below). The ras1+, rho3+, rdi1+, akr1+, pfa3+, pfa5+, erf2+, pfa3+, and erf4+ null mutants were obtained by entire deletion of the corresponding coding sequences and their replacement with the G418 (kanR) or nourseothricin (natR) cassettes by a PCR-mediated strategy using plasmid pFA6a-kanMX6 or pFA6a-natMX6, respectively, as the template.
S. pombe strains used in this studya
Gene fusion, site-directed mutagenesis, and expression plasmids.To construct integrative plasmid pIL-rho2:HA:CCIIS, the rho2+ open reading frame (ORF) plus regulatory sequences were amplified by PCR using fission yeast genomic DNA as the template and employing the 5′ oligonucleotide PRho2-5 (CCTTATCTAGATCACGGGTCTGCGTTGGC), which hybridizes at positions 674 to 656 upstream of the rho2+ ATG start codon and contains an XbaI site, and the 3′ oligonucleotide Rho2HACAAX-3 (ACTTACCCGGGTTATGAAATGATGCAGCATTTTGTAGAACTCTTGCCCGCATAGTCAGGAACATCGTATGGGTAGCCGTCATTTTCCGAATCCCGAACTG), which hybridizes at the 3′ end of the rho2+ ORF and incorporates a 63-nucleotide sequence (underlined) encoding one hemagglutinin (HA) epitope (sequence GYPYDVPDYAG) followed by the 10 C-terminal amino acids of the Rho2 GTPase (sequence SSTKCCIIS) (see Fig. 1C) and a SmaI site (restriction endonuclease sites in these and the sequences below are in italics). The resulting PCR fragment was digested with XbaI and SmaI and cloned into the integrative plasmid pIL-GFP (25). Integrative plasmids expressing HA-fused Rho2 mutants were obtained by PCR using plasmid pIL-rho2:HA:CCIIS as the template by employing the 5′ oligonucleotide PRho2-5 and the following mutagenic 3′ oligonucleotides: Rho2B3 (ACTTACCCGGGTTATGAAATGATGCAGGATTTTGTAGAACTCTTGCCCGC) to obtain the C196S mutation, Rho2C3 (ACTTACCCGGGTTATGAAATGATGGAGCATTTTGTAGAACTCTTGCCCGC) to obtain the C197S mutation, Rho2D3 (ACTTACCCGGGTTATGAAATGATGGAGGATTTTGTAGAACTCTTGCCCGC) to obtain the C196S C197S mutation, and Rho2K3 (ACTTACCCGGGTTATAAAATGATGCAGCATTTTGTAGAACTCTTGCCCGCATAGTCAGG) to obtain the S200L mutation (base changes are in boldface). Substitution of the 10 C-terminal amino acids of Rho2 for the equivalent residues from Rho1 (sequence TKKKKRCILL) was achieved by using the 3′ oligonucleotide Rho2E3 (ACTTACCCGGGTTATAGAAGGATGCA GCGTTTTTTCTTCTTCGTGCCCGCATAGTCAGGAACATCGTATGGGTAGCCGTCATTTTCCGAATCC). Subsequent mutations within this motif were obtained sequentially employing PRho2-5 and the 3′ oligonucleotides Rho2G3 (CCTTACCCGGGTTATAGAAGGATGCAGCATTTTTTCTTCTTCGTGCCCGC), to obtain the R196C mutation, and Rho2J3 (CCTTACCCGGGTTATGAAAGGATGCAGCATTTTTTCTTCTTCGTGCCCGC), to obtain the R196C L200S mutation. To construct an integrative plasmid expressing a Rho2 version fused to HA and followed by the last 25 C-terminal amino acids (sequence -KKSKPKNSVWKRLKSPFRKKKDSVT) from the non-CAAX GTPase Rit, we used plasmid pIL-rho2:HA:CCIIS as the template, employing the 5′ oligonucleotide PRho2-5 and the 3′ oligonucleotide Rho2R3 (ACTTACCCGGGTTAAGTTACTGAATCTTTCTTCTTCCGGAATGGTGATTTTAGCCTCTTCCATACACTGTTTTTGGGCTAGATTTTTTACCACCACCGTCATTTTCCGAATCCCG). To construct integrative plasmids expressing wild-type or mutant N-terminal green fluorescent protein (GFP)-fused versions of Rho2, PCR fragments were amplified using as the template DNA from strain PPG4549 (32), which expresses a N-terminally tagged version of Rho2 under the control of its natural promoter, the 5′ oligonucleotide PRho2-5, and the 3′ oligonucleotides Rho2HACAAX-3, Rho2B3, Rho2C3, Rho2D3, Rho2E3, Rho2G3, Rho2J3, Rho2K3, and Rho2R3, as described above. In all cases the purified PCR fragments were digested with XbaI and SmaI and cloned into plasmid pIL-GFP (25). The resulting integrative plasmids were digested at the unique NruI site within leu1+, and the linear fragments were transformed into rho2Δ strain MI700 (24). leu1+ transformants were obtained, and the correct integration of the fusions was verified by both PCR and Western blot analysis.
Wild-type and chimeric Rho2 overexpression constructs were obtained by PCR amplification of the corresponding ORF using the plasmids described above as the templates with the 5′ oligonucleotide Rho2X-5 (ACTTACTCGAGAGTGTTTAATCCGCTCCC; contains an XhoI site) and the 3′ oligonucleotides Rho2HACAAX-3, Rho2B3, Rho2C3, Rho2D3, Rho2E3, Rho2G3, Rho2J3, and Rho2K3. In all cases the purified PCR products were digested with XhoI and SmaI and cloned into the expression plasmid pREP3X (33).
Stress treatments and detection of activated Pmk1.For stress-induced activation of Pmk1, log-phase cell cultures (optical density at 600 nm [OD600] = 0.5) growing at 28°C in YES were supplemented with either potassium chloride (Sigma Chemical) or caspofungin (a kind gift from Merck). Hypotonic treatment was achieved by growing cells in YES medium supplemented with 0.8 M sorbitol (Sigma Chemical) and subsequently transferring them to the same medium without polyol. The cells from 50 ml of culture were harvested at different times by centrifugation at 4°C and washed with cold phosphate-buffered saline (PBS), and the yeast pellets were immediately frozen in liquid nitrogen for subsequent analysis. Cell extracts were prepared under native conditions employing chilled acid-washed glass beads and lysis buffer (10% glycerol, 50 mM Tris-HCl [pH 7.5], 15 mM imidazole, 150 mM NaCl, and 0.1% Nonidet P-40 plus a specific protease and phosphatase inhibitor; Sigma Chemical). Cleared lysates were obtained, and HA6H (HA with 6 histidines)-tagged Pmk1 was purified with Ni2+-nitrilotriacetic acid (Ni2+-NTA)–agarose beads (Qiagen) (24). The purified proteins were resolved in 10% SDS-PAGE gels and transferred to Hybond-ECL membranes (GE Healthcare). Dual phosphorylation in Pmk1 was detected with rabbit polyclonal anti-phospho-p44/42 (Cell Signaling), whereas total Pmk1 was detected with mouse monoclonal anti-HA antibody (12CA5; Roche Molecular Biochemicals). In some experiments mouse monoclonal anti-PSTAIR (anti-Cdc2; Sigma Chemical) was used as a loading control. The immunoreactive bands were revealed with an anti-mouse horseradish peroxidase (HRP)-conjugated secondary antibody (Sigma) and the ECL system (Amersham-Pharmacia). Densitometric quantification of Western blot signals was performed using ImageJ (34).
In vivo palmitoylation assay.We essentially followed a scaled-down version of the original acyl-biotinyl switch assay developed by Wan et al. (1) and described in reference 17. S. pombe strains expressing control mutant Rho2-HA alleles were grown in YES (100 ml) to a final OD600 of 0.8. Cells from 50-ml cultures were resuspended in 1 ml lysis buffer (see above) containing 10 mM N-ethylmaleimide (NEM; Sigma Chemical) plus protease inhibitors (Sigma Chemical). Cell extracts were processed exactly as described previously (17), including NEM removal by repeated chloroform-methanol precipitation, treatment including or not including 0.7 M hydroxylamine in the presence of biotin-BMCC {1-biotinamido-4-[4′-(maleimidomethyl)-cyclohexyl carboxamido]butane; Thermo Scientific}, and recovery of acyl-biotinylated proteins with NeutrAvidin beads (Thermo Scientific). After washings, the NeutrAvidin-bound proteins were eluted in Laemmli sample buffer, subjected to SDS-PAGE, and analyzed by Western blotting with mouse monoclonal anti-HA antibody (12CA5; Roche Molecular Biochemicals) as described above.
Determination of GTPase activity in vivo.The amount of GTP-bound Rho2 was analyzed using a Rho-GTP pulldown assay as previously described (35). This assay was performed with growing cultures (25°C) of strains expressing genomic versions of distinct lipidated versions of the Rho2-HA fusion. Cell extracts were obtained using 350 μl of lysis buffer (50 mM Tris-HCl [pH 7.5], 20 mM NaCl, 0.5% NP-40, 10% glycerol, 0.1 mM dithiothreitol [DTT], 1 mM NaF, and 2 mM MgCl2 containing 100 μM p-aminophenyl methanesulfonyl fluoride, and a protease inhibitor cocktail; Sigma Chemical). Twelve micrograms of glutathione S-transferase (GST)–RBD (rhotekin Rho-binding domain), previously obtained from Escherichia coli DNA expression, purified, and coupled to GST beads was used to precipitate the GTP-bound Rho2 from 3 mg of the total cell lysates. The extracts were incubated with the GST-RBD beads for 2 h at 4°C, washed four times, and blotted with anti-HA high-affinity monoclonal antibody (Roche Molecular Biochemicals) to detect the corresponding GTP-bound Rho2-HA. Total amounts of Rho2-HA from extracts were determined by Western blotting using the anti-HA monoclonal antibody, and the immunoreactive spots were revealed with an anti-rat HRP-conjugated secondary antibody (Sigma) and the ECL system.
Subcellular fractionation by sucrose density gradient centrifugation.Cells from 200 ml of exponentially growing cultures in YES medium (OD600 = 0.8) were collected, resuspended in 1.5 ml of lysis buffer D (17% [wt/vol] sucrose, 50 mM Tris-HCl [pH 7.5], and 1 mM EDTA plus specific protease and phosphatase inhibitors; Sigma Chemical), and cell extracts were prepared by employing chilled acid-washed glass beads. The cleared lysates (1 ml) were applied on top of tubes containing 13.5 ml of 10 to 65% (wt/vol) linear sucrose gradients in 50 mM Tris-HCl, pH 7.5, and 1 mM EDTA. The tubes were placed in an SW41Ti rotor (Beckman) and centrifuged at 100,000 × g for 20 h (4°C). After centrifugation, 0.4-ml fractions were sequentially obtained from the bottom of the tubes, and 10 μl from each sample was subjected to SDS-PAGE and Western blot analysis with a mouse monoclonal anti-GFP antibody (Roche Molecular Biochemicals) to detect GFP-Rho2 fusions (see above). Localization of distinct subcellular membrane fractions along the gradients was achieved by using the following primary antibodies: rabbit polyclonal anti-Pma1 antibody (y-300; Santa Cruz Biotechnology) as a marker for plasma membrane localization (36), rabbit polyclonal anti-Pep12 antibody (Molecular Probes) for endosomal localization (37), anti-GFP antibody for localization of the SPAC1B2.03c-GFP fusion at the endoplasmic reticulum (38), and rabbit polyclonal anti-ATP6V1B2 (Vma2) antibody (Abnova) for vacuole membrane localization (39).
Plate assay of stress sensitivity for growth.S. pombe control and mutant strains were grown in YES liquid medium to an OD600 of 0.5. Appropriate decimal dilutions were spotted in duplicate on YES solid medium or in the same medium supplemented with different concentrations of MgCl2 (Sigma), FK506 (Alexis Biochemicals), or caspofungin (Merck). Plates were incubated at 28°C for 3 days and then photographed.
Fluorescence microscopy.Images of GFP-fused GTPases were obtained with an Olympus 1X71 microscope equipped with a personal Delta Vision system and a Photometrics CoolSnap HQ2 camera. Stacks of 5 z-planes, 0.2 μm apart, were acquired across the width of the cells. Images were then deconvolved using the Sofworx software from Applied Precision. All fluorescence images shown correspond to a single middle plane from these z-series after deconvolution. The plot profile feature from ImageJ software was used to obtain fluorescence density histograms across the longitudinal axis of cells in the late G2 phase of the cell cycle and to determine the percentage of plasma membrane targeting in different mutants. Calcofluor white was employed for cell wall/septum staining.
Reproducibility of results.All experiments were repeated at least three times. Depending on the experiment, mean relative units ± standard deviations (SD) and/or representative results are shown. P values were analyzed by the unpaired Student t test.
RESULTS
Fission yeast Rho2 is palmitoylated in vivo.To analyze the biological relevance of lipid modifications in Rho2-dependent signaling we first tested the ability of different genomic epitope-tagged versions of Rho2 to suppress phenotypes associated with the lack of this GTPase. These include the VIC phenotype (ability to grow in the presence of magnesium chloride and FK506, the specific inhibitor of calcineurin), the sensitivity to caspofungin (a β-glucan synthase inhibitor), and defective Pmk1 activation during a saline osmotic stress (23, 24). Expression of versions of Rho2 N- or C-terminally tagged with 3HA (3HA-Rho2) or HA followed by 10 C-terminal residues of the GTPase (sequence KSSTKCCIIS; Rho2-HA-CCIIS) (Fig. 1A) was able to fully suppress both the VIC and caspofungin-sensitive phenotypes of a rho2Δ strain (see Fig. S1A in the supplemental material). However, the degree of suppression in rho2Δ cells expressing versions of Rho2 either N- or C-terminally tagged with GFP was not complete (see Fig. S1A). Importantly, Rho2-HA-CCIIS and GFP-Rho2-HA-CCIIS fusions were able to fully and partially restore, respectively, Pmk1 activation in response to a saline stress in rho2Δ cells (see Fig. S1B). Therefore, mutant strains based on these constructs were selected for the experiments described below.
Rho2 is palmitoylated in vivo. (A) C-terminal sequences of Rho2 constructs employed include a single HA tag followed by the KSSTKCCIIS motif (control) and mutated versions in which cysteine-196 (SCIIS), cysteine-197 (CSIIS), or both residues (SSIIS) were replaced by serine. (B) Conserved C-terminal sequences in selected human RhoB family small GTPases and Rho GTPases from budding and fission yeasts. Positively charged residues are underlined. Prenylated cysteine residues (F, farnesylation; GG, geranylgeranylation) are in boldface, whereas palmitoylatable cysteines (P, palmitoylation) are in boldface and italics. H. sapiens, Homo sapiens. (C) Strain LSM400 (Rho2-HA-CCIIS) was grown in YES medium to mid-log phase in the absence or presence of 100 μM 2-bromopalmitate (2-BP) for 3 h and treated with 0.6 M KCl. Aliquots were harvested at timed intervals, and activated and total Pmk1s were detected with anti-phospho-p44/42 and anti-HA antibodies, respectively. R.U, relative units. (D) Rho2 palmitoylation assayed by the acyl-biotinyl switch assay (upper blot) in cell lysates from strains LSM400 (Rho2-HA-CCIIS, control), LSM401 (Rho2-HA-SCIIS), LSM402 (Rho2-HA-CSIIS), and LSM403 (Rho2-HA-SSIIS). Biotinylation is specific for proteins containing a free sulfhydryl generated after hydroxylamine cleavage of a thioester bond, indicating palmitoylation. Total extracts from the strains were included as controls (lower blot). Rho2 was detected by employing anti-HA antibody. (E) SDS-PAGE of cell lysates from strains described for panel D. Pmk1 and Rho2 were detected by immunoblotting with anti-HA antibody, and Rho2 levels were quantified using Pmk1 as an internal control. *, P < 0.05 in mutant strains compared to the control.
In vivo farnesylation of Rho2 cysteine-197 by farnesyltransferase Cpp1 is critical for membrane binding and function of this GTPase within the Pmk1 cell integrity pathway (23). However, as in other Rho GTPases, there is a second cysteine residue immediately upstream of the farnesylated cysteine, which might be palmitoylated in vivo (cysteine-196) (Fig. 1A and B). Pretreatment of cell cultures with the irreversible palmitoylation inhibitor 2-bromopalmitate significantly reduced Pmk1 activation in response to salt stress (Fig. 1C), suggesting that this is indeed the case. We thus employed site-directed mutagenesis to obtain integrative plasmids expressing genomic Rho2-HA fusions followed by the KSSTKCCIIS motif, in which cysteine-196, cysteine-197, or both residues were replaced by serine (Fig. 1A). These constructs were separately transformed into a rho2Δ strain, and a modified version of the acyl-biotinyl switch assay was employed to assess Rho2 palmitoylation in vivo (1, 17). As indicated in Fig. 1D, wild-type Rho2 was palmitoylated in vivo, while no palmitoylation was detected in strains expressing either Rho2-HA-SCIIS, Rho2-HA-CSIIS, or Rho2-HA-SSIIS. The absence of palmitoylation in cells expressing unfarnesylated Rho2 (CSIIS mutant) agrees with the dogma that palmitoylation in prenylated-palmitoylated proteins requires prior prenylation (3).
In the palmitoylation assays, total protein levels of the unfarnesylated GTPases Rho2-HA-CSIIS and Rho2-HA-SSIIS appeared increased compared to those in strains expressing either wild-type or unpalmitoylated Rho2 (Fig. 1D). Quantitative Western blot analysis of strains expressing each of the Rho2 mutant versions in a Pmk1-HA background (loading control) confirmed this observation (Fig. 1E). Also, the previously described mobility shift in Rho2 migration due to lack of prenylation (23) was evident in Rho2-HA-CSIIS and Rho2-HA-SSIIS mutants but not in control or Rho2-HA-SCIIS cells (Fig. 1E). Taken together, the above results strongly suggest that Rho2 GTPase is palmitoylated in vivo at cysteine-196 in fission yeast.
Palmitoylation enhances Rho2 targeting to the plasma membrane.To study the subcellular localization of each of the Rho2 variants, we constructed rho2Δ strains expressing wild-type, unpalmitoylated (SCIIS), or unprenylated (CSIIS or SSIIS) HA-tagged versions of Rho2 fused to GFP at their N termini. Fluorescence microscopy revealed that wild-type Rho2 was located mostly at the plasma membrane in growing cells (Fig. 2A). This pattern changed drastically in the unpalmitoylated mutant, with Rho2 mainly localizing to endomembranes (endoplasmic reticulum, Golgi region), with minor fluorescence at the plasma membrane (Fig. 2A). Unprenylated GFP-Rho2 mutants did not bind to membranes and displayed a strong nucleocytoplasmic distribution (Fig. 2A). Quantification of the relative fluorescence across the cells confirmed the observed patterns (Fig. 2A). To further analyze the role of Rho2 palmitoylation in specific membrane targeting, we performed subcellular fractionation of cell extracts from strains expressing wild-type and unpalmitoylated versions of Rho2 by using continuous 10 to 65% sucrose gradient density centrifugation. Rho2 was determined in each fraction with anti-GFP antibodies, and the following marker proteins of known subcellular localization were used as reference controls: Pma1 ATPase, for the plasma membrane (36); Pep12, as an endosomal SNARE protein (37); a C-terminal GFP-fused version of SPAC1B2.03C, which localizes to the membrane endoplasmic reticulum (38); and Vma2, a peripheral vacuole membrane protein (39). As Fig. 2B shows, wild-type Rho2 mostly colocalized with fractions corresponding to the plasma membrane (Pma1), although some protein was also detected in endosomal fractions. Importantly, unpalmitoylated Rho2 displayed a major distribution shift along the gradient, which was absent from plasma and endosomal membranes but was enriched in endoplasmic reticulum and also in vacuolar membrane fractions (Fig. 2B). Indeed, it has been found that unpalmitoylated Ras accumulates on internal membranous structures that resemble vacuoles both in budding and fission yeasts (16, 40). These results indicate that in vivo palmitoylation is a critical requirement for Rho2 targeting to the plasma membrane.
Palmitoylated Rho2 is targeted to the plasma membrane and is essential during morphogenesis but not for GTPase activity. (A) Deconvolved images of cells from strains LSM500 (GFP-Rho2-HA-CCIIS, control), LSM501 (GFP-Rho2-HA-SCIIS), LSM502 (GFP-Rho2-HA-CSIIS), and LSM503 (GFP-Rho2-HA-SSIIS) grown in YES medium and observed by fluorescence microscopy. GFP fluorescence density histograms across the longitudinal axis (dotted white lines) of representative cells in late G2 phase are shown. (B) Density gradient centrifugation in 10 to 65% sucrose of cell extracts from growing cultures of strains LSM521 (GFP-Rho2-HA-CCIIS, SPAC1B2.03c-GFP, control) and LSM522 (GFP-Rho2-HA-SCIIS, SPAC1B2.03c-GFP). Aliquots from the indicated fractions (fraction 1 is the bottom of the tube) were subjected to Western blot analysis with anti-GFP antibody to detect GFP-Rho2 and SPAC1B2.03c-GFP (endoplasmic reticulum marker) fusions, anti-Pma1 antibody (plasma membrane marker), anti-Pep12 antibody (endosomal marker), and anti-ATP6V1B2 (Vma2) antibody (vacuole marker). (C) Strain MI700 (rho2Δ) was transformed separately with plasmid pREP3X-Rho2-HA-CCIIS (control), pREP3X-Rho2-HA-SCIIS, pREP3X-Rho2-HA-CSIIS, or pREP3X-Rho2-HA-SSIIS and thereafter grown for 18 h in the absence of thiamine (−B1). Cell morphology was analyzed by fluorescence microscopy after staining of cells with calcofluor white. (D) Serially diluted cells from transformants described for panel C were spotted on EMM2 plates with (+B1) or without (−B1) 5 mM thiamine and incubated for 4 days at 28°C. (E) Cell extracts from strains expressing genomic integrated versions of rho2+ alleles described for panel C were precipitated with GST-RBD and blotted against anti-HA antibody (top). Total HA-Rho2 was estimated by Western blotting with anti-HA antibody (middle), and Pmk1-HA was detected in the same extracts as the loading control (bottom). Quantification of the ratio of GTP bound to total Rho2 is shown.
Plasma membrane localization of palmitoylated Rho2 is essential during morphogenesis but not for Rho2 activity.Rho2 is involved in the regulation of cell morphogenesis and polarity, and its overexpression induces cell swelling and cell wall defects leading to cell death (22). Farnesylated cysteine-197 is critical to promote these phenotypes, since overexpression of a Rho2-HA-CSIIS mutant does not induce morphological changes and/or lethality (23). As predicted, whereas overexpression of wild-type rho2+ in rho2Δ cells strongly altered cellular morphology and inhibited cell growth (Fig. 2C and D), expression of unprenylated Rho2 (CSIIS or SSIIS mutant) had no effect on both phenotypes (Fig. 2C and D). Importantly, cells overexpressing nonpalmitoylated Rho2 (SCIIS) grew normally without obvious morphological alterations (Fig. 2C and D). Since these phenotypes might be due to lack of GTPase activity in cells expressing unlipidated Rho2 versions, we performed pulldown experiments to determine the amount of endogenous GTP-bound Rho2 in rho2Δ cells expressing genomic versions of either the control Rho2 or the SCIIS, CSIIS, or SSIIS mutant fused to HA. Notably, compared to control cells, rho2Δ cells expressing unlipidated versions of Rho2 showed a clear increase in the amount of GTP-bound Rho2 (ranging from 1.8-fold greater than that of the control cells in the SCIIS mutant to 6- and 5-fold greater in the CSIIS and SSIIS mutants, respectively) (Fig. 2E). These results clearly demonstrate that, whereas in vivo palmitoylation at cysteine-196 and plasma membrane targeting are crucial for Rho2 control of cell growth and morphogenesis, Rho2 activation can take place independently of protein lipidation and plasma membrane binding.
Role of palmitoylation in Rho2-dependent signaling to the Pmk1 MAPK cascade.Rho2 controls the cell integrity pathway in fission yeast (23, 24). Therefore, we comparatively analyzed basal Pmk1 phosphorylation levels in growing cultures of a rho2Δ mutant and in cells expressing genomic versions of Rho2 (control), Rho2-HA-SCIIS, Rho2-HA-CSIIS, and Rho2-HA-SSIIS. Deletion of rho2+ decreased significantly the basal Pmk1 phosphorylation (Fig. 3A) (24), which also remained very low in strains expressing unpalmitoylated and/or unfarnesylated versions of the GTPase, compared to that in control cells (Fig. 3A). Determination of the VIC phenotype yielded similar results, with all strains remaining VIC positive except those expressing wild-type Rho2 (Fig. 3B). Interestingly, the VIC phenotype of cells expressing unpalmitoylated Rho2 (SCIIS) was not as strong as that of cells lacking Rho2 or expressing unprenylated versions of the GTPase, suggesting that it might retain some minor signaling activity (Fig. 3B). Thus, Rho2 palmitoylation is needed in fission yeast for full effectiveness of the cell integrity pathway during cell growth.
In vivo palmitoylation of Rho2 is essential for signaling to the cell integrity pathway during vegetative growth and in response to stress. (A) Strains MI700 (rho2Δ), LSM400 (Rho2-HA-CCIIS; control [C]), LSM401 (Rho2-HA-SCIIS), LSM402 (Rho2-HA-CSIIS), and LSM403 (Rho2-HA-SSIIS) were grown in YES medium, and activated and total Pmk1s were detected with anti-phosho-p42/44 and anti-HA antibodies, respectively. (B) Strain MI102 (pmk1Δ) and those described for panel A were grown in YES medium, and serially diluted cells were spotted on YES plates supplemented with either 0.1 M or 0.2 M MgCl2 plus 1 μg/ml FK506 (VIC). (C) Strains described for panel A were grown as above and treated with either 0.6 M KCl or 1 μg/ml caspofungin or subjected to a hypotonic treatment. (D) Serially diluted cells of strains described for panel B were spotted on YES plates supplemented with 0.8, 1, or 1.2 μg/ml caspofungin. (E) Strain MI700 (rho2Δ, Pmk1-HA) was transformed with plasmid pREP3X-Rho2-HA-CCIIS (control), pREP3X-Rho2-HA-SCIIS, pREP3X-Rho2-HA-CSIIS, or pREP3X-Rho2-HA-SSIIS, and the corresponding transformants were grown for 20 h in the presence or absence of thiamine (B1). Both activated and total Pmk1s were detected as previously described.
Pmk1 activation induced by osmotic saline or hypotonic stresses is totally dependent upon the signaling mediated by Rho2 (24). As shown in Fig. 3C, in contrast to results for control cells, Pmk1 remained unactivated in cells expressing either unpalmitoylated or unfarnesylated Rho2 versions when subjected to the above stresses. On the other hand, a moderate increase in Pmk1 phosphorylation was still detected in Rho2-HA-SCIIS cells treated with the β-glucan synthase inhibitor caspofungin (Fig. 3C). This result is consistent with previous findings showing that MAPK activation triggered by cell wall stress is not completely dependent on this GTPase (24). Interestingly, mutant strains expressing unlipidated versions of Rho2 were as sensitive to caspofungin as rho2Δ cells, whereas cells expressing wild-type Rho2 grew normally in the presence of the drug (Fig. 3D). Pmk1 hyperactivation is responsible for cellular lethality during rho2+ overexpression (23). Considering the above evidence, we anticipated that overexpression of unpalmitoylated Rho2 would not induce MAPK phosphorylation. Indeed, strong Pmk1 activation was observed only after overexpression of wild-type Rho2 (Fig. 3E). Taken together, our results support the hypothesis that in vivo palmitoylation of Rho2 is essential for signaling to the cell integrity pathway during vegetative growth and in response to stress.
Artificial plasma membrane targeting bypasses the Rho2 need for palmitoylation in order to signal.The results obtained so far show that in vivo palmitoylation of Rho2 enhances its plasma membrane localization and function. Then it could be predicted that a nonlipidated plasma membrane-tethered version of the GTPase might be able to circumvent the palmitoylation requirement for signaling. To test this hypothesis, we constructed strains expressing Rho2 or GFP-Rho2 versions lacking its natural CCIIS motif and followed by the last 25 amino acids from the hydrophobic C terminus of the mammalian plasma membrane nonlipidated GTPase Rit (41). This domain has been successfully employed for plasma membrane targeting in other signaling proteins, such as Ras1 (15) and Pmk1 (42). Similar to findings in cells expressing palmitoylated Rho2, the GFP-Rho2-RitC fusion displayed strong plasma membrane localization (∼75%) (Fig. 4A) and colocalized with fractions corresponding to the plasma membrane marker Pma1 (Fig. 4B). Remarkably, levels of basal Pmk1 phosphorylation were virtually identical in cells expressing either wild-type (palmitoylated) Rho2 or the Rho2-RitC fusion (Fig. 4C), and the Rho2-RitC fusion suppressed the VIC-positive and caspofungin-sensitive phenotype of rho2Δ cells to an extent similar to that by wild-type Rho2 (Fig. 4D). Finally, Pmk1 activation during osmotic saline or hypotonic stress in control cells was very similar to that in Rho2-RitC cells (Fig. 4E). These results suggest that Rho2 signaling is mostly dependent on plasma membrane localization.
Artificial plasma membrane targeting bypasses the Rho2 need for palmitoylation in order to signal. (A) Deconvolved images of cells from strains LSM500 (GFP-Rho2-HA-CCIIS, control) and LSM970 (GFP-Rho2-HA-RitC). The percentage of GFP-Rho2 at the plasma membrane (PM) with respect to the whole cell is shown at the right (n ≥ 15 cells). (B) Density gradient centrifugation of cell extracts from cultures of strains LSM521 (GFP-Rho2-HA-CCIIS, SPAC1B2.03c-GFP, control) and LSM970 (GFP-Rho2-HA-RitC). Aliquots from the indicated fractions were subjected to Western blot analysis with anti-GFP and anti-Pma1 antibodies (plasma membrane marker). (C) Strains LSM400 (Rho2-HA-CCIIS, control), MI700 (rho2Δ), LSM403 (Rho2-HA-SSIIS), and LSM971 (Rho2-HA-RitC) were grown in YES medium, and activated and total Pmk1s were detected with anti-phosho-p42/44 and anti-HA antibodies, respectively. (D) Serially diluted cells of strains described for panel C were spotted on YES plates supplemented with either 0.05 M or 0.1 M MgCl2 plus 1 μg/ml FK506 (VIC) or with 0.6 or 1 μg/ml caspofungin. (E) Strains LSM400 and LSM971 were treated with either 0.6 M KCl (left) or subjected to a hypotonic treatment (right). Activated and total Pmk1s were detected as previously shown.
A C-terminal polybasic sequence interferes with in vivo Rho2 palmitoylation depending on its prenylation context.We have recently demonstrated that the essential GTPase Rho1, in addition to Rho2, is an upstream regulator of the cell integrity pathway in fission yeast (30). Unlike Rho2, Rho1 lacks palmitoylatable cysteine residues and harbors a C-terminal polybasic sequence followed by an adjacent cysteine residue that is geranylgeranylated in vivo (Fig. 5A) (43). Rho1 displays weak plasma membrane localization and is enriched at the growing tips during interphase and at the septum prior to cytokinesis (44). Compared to wild-type Rho2, a GFP-tagged Rho2 chimera in which its natural CAAX box was replaced by the last 10 Rho1 C-terminal amino acids (Rho2-TKKKKRCILL) (Fig. 5A) showed decreased plasma membrane targeting (∼45%) and increased its localization at endomembranes and at the cell nucleus (Fig. 5B). Unlike wild-type Rho2, overexpression of Rho2-TKKKKRCILL was not lethal (Fig. 5C). Basal Pmk1 phosphorylation in cells expressing a genomic version of this chimera was lower than that in control cells (Fig. 5D), and this resulted in a detectable VIC phenotype (Fig. 5E). Moreover, maximal Pmk1 activation in response to osmotic saline stress and hypotonic shock was similarly reduced by approximately 45% in Rho2-TKKKKRCILL cells compared to the wild type (Fig. 5F). These results reinforce the notion that in fission yeast Rho2 palmitoylation at the CCIIS motif is essential for robust membrane sorting and downstream signaling.
Rho1 C-terminal polybasic sequence disturbs palmitoylation and signaling of geranylgeranylated but not farnesylated Rho2. (A) C-terminal sequences of Rho2 and Rho2-Rho1 chimeras. Rho1 C-terminal sequences are underlined, whereas amino acids derived from Rho2 terminal sequences are in bigger letters. (B) Deconvolved images of cells from strains LSM500 (GFP-Rho2-HA-CCIIS, control), LSM505 (GFP-Rho2-HA-RCILL), LSM506 (GFP-Rho2-HA-CCILL), and LSM507 (GFP-Rho2-HA-CCILS). The percentage of GFP-Rho2 at the plasma membrane with respect to the whole cell is shown at the right (n ≥ 15 cells; *, P < 0.05 in mutants compared to the control). (C) Strain MI700 (rho2Δ, Pmk1-HA) was transformed separately with plasmid pREP3X-Rho2-HA-CCIIS (control), pREP3X-Rho2-HA-RCILL, pREP3X-Rho2-HA-CCILL, or pREP3X-Rho2-HA-CCILS, and serially diluted cells from the respective cultures were spotted on EMM2 plates with (+B1) or without (−B1) 5 mM thiamine and incubated for 4 days at 28°C. (D) Strains LSM400 (Rho2-HA-CCIIS, control), LSM405 (Rho2-HA-RCILL), LSM406 (Rho2-HA-CCILL), LSM407 (Rho2-HA-CCILS), and MI700 (rho2Δ) were grown in YES medium, and both activated and total Pmk1s were detected as described for Fig. 1. (E) Strains described for panel D were grown in YES medium, and serially diluted cells were spotted on YES plates supplemented with 0.05 M or 0.1 M MgCl2 plus 1 μg/ml FK506 (VIC). (F) Strains described for panel D were either treated with 0.6 M KCl or subjected to a hypotonic treatment. (G) Rho2 palmitoylation assayed by the acyl-biotinyl switch assay in cell lysates from strains described for panel D.
Intriguingly, a Rho2 chimera in which arginine-196 within the Rho1 C-terminal motif was replaced by a palmitoylatable cysteine residue (sequence TKKKKCCILL; replaced residue underlined) (Fig. 5A) also showed altered plasma membrane localization (Fig. 5B) and remained unpalmitoylated in vivo (Fig. 5G). This mutant displayed a defective activation of the Pmk1 cascade, similar to cells expressing the Rho2-TKKKKRCILL construct (Fig. 5C to F). These findings suggested that the presence of the polybasic motif and/or the prenylation type (geranylgeranylation instead of farnesylation) hampers the in vivo palmitoylation of Rho2-TKKKKCCILL. We thus substituted serine for the terminal leucine in this construct to favor cysteine-197 farnesylation in vivo (Rho2-TKKKKCCILS; replacement residues underlined) (Fig. 5A). Notably, a GFP-fused version of this chimera was able to exit the cell nucleus, to increase its targeting to the plasma membrane, and to become palmitoylated in vivo, albeit to a lower degree than wild-type Rho2 (Fig. 5B and G). Most importantly, this mutant recovered the phenotypes associated with Rho2 function, including lethality in response to overexpression (Fig. 5C) and stress-induced Pmk1 activation (∼80% maximal Pmk1 activation compared to control cells) (Fig. 5F). Since substitution of farnesylation by geranylgeranylation does not affect Rho2 localization, palmitoylation, and signaling capabilities (see Fig. S2 in the supplemental material), these findings support the hypothesis that the presence of the Rho1 C-terminal polybasic sequence disturbs the in vivo palmitoylation of geranylgeranylated but not farnesylated Rho2.
Rho GDI Rdi1 negatively regulates the Pmk1 MAPK cascade in a Rho2-independent fashion.Rho GDIs negatively regulate Rho-mediated signaling by extracting Rho proteins from membranes to form cytoplasmic complexes in which the prenylated CAAX box fits into the hydrophobic pocket formed by the Ig-like β sandwich of the GDI (45). Also, the presence of adjacent palmitoylated cysteine residues likely renders GTPases insensitive to the GDI-mediated cytosolic sequestration (46). Similar to S. cerevisiae, fission yeast harbors a single ORF (SPAC6F12.06) encoding a GDI ortholog named Rdi1, which binds to Rho1, Rho4, and Cdc42 but not to Rho2 or Rho3 (47). rdi1Δ cells are viable but divide at a slightly smaller size than control cells (13.1 ± 0.54 μm versus 14.3 ± 0.73 μm, respectively) and show minimal morphological defects at normal growth conditions (not shown). Basal Pmk1 phosphorylation increased in growing rdi1Δ cells compared to control cells, suggesting that Rdi1 is a negative regulator of the cell integrity pathway (Fig. 6A). As expected, this control is not exerted through Rho2, since the low basal Pmk1 activity in rho2Δ cells was still enhanced by simultaneous deletion of the rdi1+ gene (Fig. 6A). Moreover, Pmk1 activation kinetics in salt-stressed rdi1Δ cells was similar to that shown by control cells except for a slight overall increase in MAPK activity (Fig. 6B). The most likely target to be negatively regulated by Rdi1 might be Rho1, considering that it has a functional relationship with the Pmk1 cascade and that it is not palmitoylated in vivo (see above). In this case, Rdi1 removal should then correct the defective Pmk1 activation in cells expressing the Rho2-TKKKKRCILL chimera (Rho1 tail) in response to stress (Fig. 5F). Indeed, rdi1+ deletion elicited a robust recovery in Pmk1 activation in Rho2-RCILL cells (Fig. 6C). Therefore, the presence of a Rho1 tail appears to be sufficient to allow Rdi1 to negatively regulate downstream signaling from a Rho2-Rho1 chimera. However, Pmk1 activation was not restored in cells expressing an unpalmitoylated version of Rho2 (SCIIS tail) in an rdi1Δ background (Fig. 6D), suggesting that palmitoylation is not the only factor preventing Rho2 targeting by Rdi1.
Rho GDI Rdi1 negatively regulates the cell integrity pathway through a Rho1-dependent mechanism. (A) Exponentially growing cultures of strains MI200 (control), MI700 (rho2Δ), LSM750 (rdi1Δ), and LSM760 (rho2Δ rdi1Δ), were grown in YES medium. Both activated and total Pmk1s were detected with anti-phospho-p44/42 and anti-HA antibodies, respectively. (B to D) Strains LSM400 (Rho2-HA-CCIIS, control) and LSM770 (Rho2-HA-CCIIS rdi1Δ) (B), LSM405 (Rho2-HA-RCILL) and LSM780 (Rho2-HA-RCILL rdi1Δ) (C), and LSM401 (Rho2-HA-SCIIS) and LSM790 (Rho2-HA-SCIIS rdi1Δ) (D) were treated with 0.6 M KCl. Aliquots were harvested at timed intervals, and activated and total Pmk1s were detected as described above.
DHHC acyltransferase Erf2 palmitoylates Rho2 and Rho3 to antagonistically regulate the Pmk1 MAPK cascade.Protein palmitoylation is primarily brought about by members of the DHHC family of palmitoyltransferases (2, 3). Fission yeast's genome contains five ORFs encoding putative DHHC members: SPAC2F7.10 (Akr1), SPBC691.01 (Pfa5), SPBC2F12.15c (Pfa3), SPBC13G1.07 (Swf1; essential), and SPBC3H7.09 (Erf2). Since palmitoylation is absolutely necessary for Rho2-dependent signaling to the cell integrity pathway, we reasoned that Pmk1 activation status would provide an excellent tool to identify the enzyme(s) responsible for in vivo palmitate transfer to this GTPase. Comparative analysis of Pmk1 activation in salt-stressed cultures of strains harboring single deletions in each of the nonessential PTs described above showed clearly decreased MAPK kinase activation in erf2Δ cells and also in a mutant lacking the Erf2 regulatory protein Erf4 (SPAC3F10.07c) (48) (Fig. 7A). A similar decrease in Pmk1 activation was observed in erf2Δ cells subjected to hypotonic stress (Fig. 7B). However, Rho2 might be a substrate for other PTs besides Erf2-Erf4, since some Pmk1 activation was still detected in stressed erf2Δ erf4Δ cells. Notably, defective Pmk1 activation in erf2Δ cells was not aggravated in an erf2Δ akr1Δ pfa3Δ pfa5Δ quadruple mutant (Fig. 7C). In comparison to control cells, erf2Δ cells displayed increased endomembrane localization of a GFP-Rho2 fusion, and this was accompanied by a notable reduction in plasma membrane targeting (∼40%) (Fig. 7D) and palmitoylation (∼60%) (Fig. 7E). Unexpectedly, the above defects were not accompanied by a reduction in basal Pmk1 phosphorylation or a noticeable VIC phenotype (Fig. 7F and G). A possible explanation for this discrepancy is that Erf2 has other targets which, when not palmitoylated, suppress the VIC phenotype regardless of Rho2 mislocalization. Indeed, we noticed that a mutant strain lacking Rho3, which is the only known target for Erf2-Erf4 besides Ras1 in fission yeast (16, 49), exhibited strong chloride sensitivity compared to control cells, unlike the rho2Δ mutant (Fig. 8A). Moreover, the basal Pmk1 phosphorylation in rho3Δ cells increased compared to that in control cells, indicating that Rho3 is a negative regulator of the cell integrity pathway (Fig. 8B). Importantly, this control is independent of Rho2 function, since both the low basal Pmk1 activity and the VIC phenotype in rho2Δ cells were suppressed in rho2Δ rho3Δ and rho2Δ erf2Δ double mutants (Fig. 8B and C). Taken together, these results suggest that the Erf2 PT is mainly responsible for Rho2 palmitoylation in vivo and that Rho2 and Rho3 regulate the basal activity of the cell integrity pathway in an antagonistic fashion (Fig. 8D).
Erf2 acyltransferase is the major protein responsible for Rho2 palmitoylation in vivo. (A) Exponentially growing cultures of strains MI200 (control), LSM810 (akr1Δ), LSM820 (pfa3Δ), LSM830 (pfa5Δ), LSM840 (erf2Δ), and LSM850 (erf4Δ) were treated with 0.6 M KCl. Aliquots were harvested at timed intervals, and activated and total Pmk1s were detected with anti-phospho-p44/42 and anti-HA antibodies, respectively. (B) Strains MI200 (control) and LSM840 (erf2Δ) were subjected to a hypotonic treatment, and both activated and total Pmk1s were detected. (C) Strains MI200 (control), LSM840 (erf2Δ), and LSM860 (erf2Δ akr1Δ pfa3Δ pfa5Δ) were subjected to hypotonic treatment, and both activated and total Pmk1s were detected. (D) Deconvolved images of cells from strains LSM500 (GFP-Rho2-HA-CCIIS, control) and LSM841 (GFP-Rho2-HA-CCIIS erf2Δ). The percentage of GFP-Rho2 at the plasma membrane with respect to the whole cell is shown at the right (n = 14 cells; *, P < 0.05 in erf2Δ cells compared to the control). (E) Rho2 palmitoylation assayed by the acyl-biotinyl switch assay (upper blot) in cell lysates from strains LSM400 (Rho2-HA-CCIIS, control), LSM842 (Rho2-HA-CCIIS erf2Δ), and LSM852 (Rho2-HA-CCIIS erf4Δ). (F) Exponentially growing cultures of strains MI700 (rho2Δ), LSM401 (Rho2-SCIIS), MI200 (control), LSM840 (erf2Δ), and LSM860 (erf2Δ akr1Δ pfa3Δ pfa5Δ) were grown in YES medium, and both activated and total Pmk1s were detected as above. (G) Serially diluted cells of strains described for panel F were spotted on YES plates supplemented with 0.05 M or 0.1 M MgCl2 plus 1 μg/ml FK506 (VIC) and incubated for 3 days at 28°C.
Erf2-palmitoylated Rho2 and Rho3 regulate the cell integrity MAPK pathway in an antagonistic fashion. (A) Serially diluted cells of strains MI200 (control), MI700 (rho2Δ), LSM901 (rho3Δ), and LSM902 (ras1Δ) were spotted on YES plates supplemented with 0.2 or 0.3 M MgCl2 and incubated for 3 days at 28°C. (B) Strains MI200 (control), MI700 (rho2Δ), LSM901 (rho3Δ), LSM903 (rho2Δ rho3Δ), LSM840 (erf2Δ), and LSM904 (rho2Δ erf2Δ) were grown in YES medium, and activated and total Pmk1s were detected with anti-phospho-p44/42 and anti-HA antibodies, respectively. (C) Serially diluted cells of strains described for panel B were spotted on YES plates supplemented with 0.05 M or 0.1 M MgCl2 plus 1 μg/ml FK506 (VIC) and incubated for 3 days at 28°C. (D) Erf2 palmitoylates Rho2 and Rho3 to antagonistically regulate the Pmk1 MAPK cascade (see the text for details).
DISCUSSION
Rho2 performs two fundamental tasks in fission yeast: modulator of cell wall integrity and morphogenesis (21, 22) and upstream activator of the cell integrity MAPK pathway (23, 24). Like most members of the Ras superfamily of small GTPases, Rho2 shows a CAAX box at its C terminus, and in vivo farnesylation of cysteine-197 within this motif is important for its plasma membrane binding and function (23). In this study we demonstrate that Rho2 is also palmitoylated in vivo at cysteine-196 and that this posttranslational modification is essential for Rho2 physiological roles. Rho2 target Pck2 is the main upstream kinase responsible for activation of the Pmk1 MAPK module in fission yeast (28). The fact that specific stimuli like hyperosmotic and hypotonic stresses signal to this cascade through Rho2 and Pck2 following a linear nonbranched pathway (24) provided an excellent biological readout to study the impact of palmitoylation on Rho2 signaling. Indeed, the lack of morphological defects and lethality upon overexpression of a nonpalmitoylated version of the GTPase (SCIIS mutant), plus complete abrogation of specific Rho2-dependent signaling to the Pmk1 cascade in Rho2-HA-SCIIS cells, clearly demonstrate the key contribution of palmitoylation to Rho2 functions.
Signal transduction of palmitoylated GTPase Ras1 is spatially compartmentalized in fission yeast (15). Whereas unpalmitoylated Ras1 localized to the endomembranes activates an Scd1-Cdc42 pathway that mediates cell polarity and protein trafficking, mating pheromone signaling is regulated exclusively by the palmitoylated GTPase localized to the plasma membrane by targeting Byr2 MAPK kinase kinase (MAPKKK) (15). In contrast, unpalmitoylated Rho2, which is mostly targeted to endomembranes, did not suppress the phenotypes of a rho2Δ strain, such as altered ionic homeostasis and defective Pmk1 activation in response to stress. A Rho2 version fused to a nonpalmitoylatable hydrophobic motif (RitC) elicited strong plasma membrane localization and bypassed the Rho2 need for palmitoylation in order to signal. However, substitution of a Rho2 tail for the Rho1 C-terminal motif, which includes a polybasic cluster followed by a geranylgeranylated cysteine residue (43), decreased its plasma membrane association and resulted in moderate signaling defects. These findings support the contention that Rho2 signaling is mostly restricted to the plasma membrane and is not associated with other cellular membrane systems. This scenario seems adequate from a mechanistic perspective because, in contrast to Ras1 effectors like Scd1, all the known direct downstream targets of Rho2, such as PKC orthologs Pck1 and Pck2, are predominantly plasma membrane proteins (38, 50). Thus, the effector localization pattern could be the main determinant to define spatial Rho2 signaling in fission yeast. Moreover, Rho2 membrane localization and biochemical activation appear uncoupled, since higher enzyme activity was detected in cells expressing versions of the GTPase that cannot be lipidated. This increase might be due to lack of negative regulation by Rho2 GAPs Rga2, Rga4, and/or Rga7, all of which are mainly localized to the plasma membrane (32, 51, 52). Previous studies demonstrated that Rac1, a Rho GTPase from higher eukaryotic cells which is palmitoylated in vivo and functions from the plasma membrane (7), shows GTPase activity in the cytosol independent of membrane recruitment (53). Nevertheless, the fact that unlipidated Rho2 is nonfunctional further confirms that its GTPase activity is biologically relevant only when executed at the plasma membrane.
The finding that a change in the prenylation type does not affect the signaling of palmitoylated Rho2 is congruent with previous studies suggesting that farnesylation is as important as geranylgeranylation for Rho GTPase function (5). Human GTPases from the RhoB family TC10/RhoQ and Rac1 harbor polybasic motifs besides palmitoylated and prenylated cysteine residues (Fig. 1B), and both geranylgeranylation and the polybasic motif are required for Rac1 palmitoylation in vivo (7). However, inclusion of a palmitoylatable cysteine residue downstream the Rho1 C-terminal polybasic sequence in a Rho2-Rho1 chimera neither resulted in detectable palmitoylation in vivo nor restored Rho2 downstream signaling. Notably, Rho2 structural and signaling features were regained when a change in the terminal amino acid within the CAAX box directed the chimeric GTPase to become farnesylated in vivo. Thus, the presence of a polybasic motif might play an interfering role during GTPase palmitoylation depending on the prenylation context. In yeasts, the C-terminal polybasic sequences are present only in Rho GTPases lacking palmitoylatable cysteine residues (Fig. 1B). The coexistence of a polybasic motif plus a palmitoylatable cysteine residue(s) in specific Rho GTPases from higher eukaryotes might be a late evolutionary acquisition to increase their functional flexibility in terms of subcellular localization and substrate accessibility. Remarkably, the polybasic motifs in Rac1 and TC10/RhoQ are positioned four amino acids apart from the palmitoylated cysteines (Fig. 1B). It is possible that the interfering mechanism described here is due, at least in part, to the absence of an amino acid spacer between the polybasic motif and the palmitoylatable cysteine, an issue that deserves further investigation.
GDI-Rho interaction is limited to nonpalmitoylated GTPases Cdc42 and RhoA in humans (46) and to Cdc42, Rho1, and Rho4 in both budding and fission yeast (47, 54). This appears to be due to the fact that palmitoylation near prenylated cysteines impairs Rho docking at the GDI hydrophobic pocket (45). We have found that there is increased basal Pmk1 activation in rdi1Δ cells and that Rho2 function is independent of Rdi1. However, replacement of its natural CAAX box by the nonpalmitoylatable Rho1 tail restored normal Pmk1 activation in response to stress. This strongly suggests that Rdi1 negatively regulates Pmk1 activity by limiting the signaling functions of Rho1, which is a true upstream activator of the cell integrity pathway (30). Interaction studies have demonstrated that Lys-185 and Arg-186 within the polybasic carboxyl terminus of human Cdc42 form hydrogen bonds with specific acidic residues located at the RhoGDIα hydrophobic pocket which likely contribute to membrane release of the GTPase (45). Two equivalent lysine residues (underlined) are conserved in the Rho1 tail (TKKKKRCILL) but are absent in Rho2 (KSSTKCCIIS) (Fig. 1B), suggesting that both basic amino acids might be essential to allow effective in vivo regulation of prenylated Rho1 by Rdi1.
An intriguing question in protein palmitoylation studies refers to the mechanism whereby different PTs combine specificity and functional redundancy against a given substrate (55). Here we show that Rho2 is a substrate for Erf2 palmitoyltransferase and its accessory protein Erf4. It has been proposed that Erf2 belongs to class III DHHC proteins due to its preference for palmitoylated lipidated Ras and Rho GTPases (56). Whereas in budding yeast Erf2-Shr5 (Erf4 ortholog) mediates palmitoylation of Ras2, Rho2, and Rho3 (57), in fission yeast Ras1 and Rho3 are substrates for the Erf2-Erf4 complex (16, 49). Accordingly, compared to control cells, erf2Δ cells showed reduced Rho2 palmitoylation, plasma membrane localization, and Pmk1 activation during osmotic saline and hypotonic stresses. Fission yeast Erf2 localizes to the trans-Golgi network, where it palmitoylates Ras1 to allow its efficient delivery to the plasma membrane (16). The localization pattern displayed by unpalmitoylated Rho2 suggests that Erf2-mediated palmitoylation also occurs at this specific subcellular location. However, simultaneous deletion of nonessential PTs Akr1, Pfa3, and Pfa5 did not exacerbate the low Pmk1 activation under stress, suggesting that either the essential PT Swf1 or a PT-independent mechanism is responsible for the remaining Rho2 palmitoylation and signaling in erf2Δ cells.
Our work also demonstrates that Rho3, a known Erf2 target (49), is a negative regulator of the cell integrity pathway in fission yeast. Epistasis and biochemical analysis revealed that Rho3 negatively regulates basal Pmk1 phosphorylation in a Rho2-independent fashion, with the low basal Pmk1 activity and the VIC phenotype being suppressed in rho2Δ cells by simultaneous deletion of Rho3. These results explain why the cell integrity pathway is still functional in vegetative erf2Δ cells despite the observed decrease in Rho2 palmitoylation and discloses the existence of a mechanism for antagonistic regulation of MAPK activity by two GTPases (Rho2 and Rho3) which are targets for the same PT (Erf2) (Fig. 8D). Rho3 is involved in polarized cell growth by regulating proper localization of formin For3 and the exocyst (58, 59) and in the control of Golgi-endosome trafficking (60). Interestingly, increased expression of Erf2/Erf4 enhances the palmitoylation state of Rho3 to promote meiotic entry (49). Future work will draw a comprehensive picture of the intricate mechanisms governing the functional connections among these important regulatory proteins.
ACKNOWLEDGMENTS
We thank J. Campoy for advice during density gradient centrifugation experiments and F. Garro for technical assistance. We thank D. Posner for language revision.
This work was supported by grants BFU2011-22517 (Ministerio de Economía y Competitividad, Spain) and 15280/PI/10 (Fundación Séneca, Región de Murcia, Spain) to J.C. and BFU2010-15641 (Ministerio de Economía y Competitividad; Spain) to P.P. The European Regional Development Fund provided cofunding from the European Union. L.S.-M. and M.M. are predoctoral (Formación de Personal Investigador) and postdoctoral (Juan de la Cierva Program) researchers, respectively, from Ministerio de Economía y Competitividad, Spain.
FOOTNOTES
- Received 19 November 2013.
- Returned for modification 6 December 2013.
- Accepted 7 May 2014.
- Accepted manuscript posted online 12 May 2014.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/MCB.01515-13.
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
