Analysis of Functional Domains in Rho5, the Yeast Homolog of Human Rac1 GTPase, in Oxidative Stress Response

The small GTPase Rho5 of Saccharomyces cerevisiae is required for proper regulation of different signaling pathways, which includes the response to cell wall, osmotic, nutrient, and oxidative stress. We here show that proper in vivo function and intracellular distribution of Rho5 depends on its hypervariable region at the carboxyterminal end, which includes the CAAX box for lipid modification, a preceding polybasic region (PBR) carrying a serine residue, and a 98 amino acid–specific insertion only present in Rho5 of S. cerevisiae but not in its human homolog Rac1. Results from trapping GFP-Rho5 variants to the mitochondrial surface suggest that the GTPase needs to be activated at the plasma membrane prior to its translocation to mitochondria in order to fulfil its role in oxidative stress response. These findings are supported by heterologous expression of a codon-optimized human RAC1 gene, which can only complement a yeast rho5 deletion in a chimeric fusion with RHO5 sequences that restore the correct spatiotemporal distribution of the encoded protein.


Introduction
The necessity to react to changing environmental conditions has led to the establishment of complex but largely conserved signal transduction pathways in all eukaryotic cells from yeast to humans. Many of those pathways use monomeric G-proteins of the Ras superfamily as molecular switches, which are active in their GTP-bound and inactive in their GDP-bound state and serve as central and highly conserved components [1]. To promote activation, guanine nucleotide exchange factors (GEFs) trigger the dissociation of GDP and its replacement by the intracellularly prevailing GTP. GTPase activating proteins (GAPs) stimulate hydrolysis of GTP and thus inactivation of the G-protein [2]. Frequently, guanine nucleotide dissociation inhibitors (GDIs) stabilize the GDP-bound state by preventing the association with GEFs and simultaneously sequester the GTPase from its membrane compartment [3]. The importance of the Ras-type GTPases is reflected in a variety of diseases associated with their malfunctions, ranging from developmental disorders and modulation of the immune system to neurodegenerative diseases (for further reading, see references in studies [4,5]).
The ras homology (Rho) subfamily of these GTPases in the model yeast Saccharomyces cerevisiae is represented by six members, namely Rho1 to Rho5 and Cdc42, which regulate signaling pathways involved in polarized growth, exocytosis and cytokinesis [6]. While the roles of the two essential members, Rho1 in cell wall integrity (CWI) and Cdc42 in establishment of cell polarity under different physiological conditions, have been thoroughly investigated, the diverse biological functions of Rho5 in yeast, and its Rac1 homologs in other fungi and humans, are only beginning to emerge [7,8].
Deletions of the gene in haploid S. cerevisiae strains were first reported to be hyper-resistant to cell wall stress agents and suggested a negative regulatory role for Rho5 in cell wall integrity (CWI) signaling [9]. Moreover, hyper-activation of Rho5 resulted in an increased sensitivity to salt stress, linking its function to the high osmolarity glycerol (HOG) signaling pathway [10]. In addition, the observation that rho5 mutants were hyper-resistant to oxidative stress provided a link to apoptosis [11] and autophagy/mitophagy, which also involves CWI and HOG signaling [12,13].
We recently showed that Rho5 interacts with a dimeric GEF constituted of yeast Dck1 and Lmo1, with deletion mutants in either of the encoding genes resembling rho5 deletions in their hyper-resistance toward oxidative stress-induced cell death [14]. All three components of the putative trimeric Dck1-Lmo1-Rho5 (DLR) complex rapidly translocate from the cell periphery to mitochondria under such stress conditions, followed by mitophagy and cell death [14]. The translocation of Rho5 occurs only in the presence of both Dck1 and Lmo1 and is also triggered by glucose starvation, thus linking the DLR complex to nutrient signaling [15].
An essential characteristic of GTPases to fulfil their biological functions in different subcellular compartments is their association with membranes, which is primarily achieved by a modification of the C-terminally located CAAX-motif (i.e., a cysteine residue followed by two aliphatic amino acids and a variable C-terminal residue). Processing of this motif results in removal of the three terminal amino acids, a carboxymethylation, and geranyl-geranylation of the cysteine, altogether providing a hydrophobic anchor for association with target membranes [16]. A polybasic region (PBR) preceding the CAAX motif is also present in many Rho-type GTPases, where the positive charges electrostatically interact with the negatively charged membrane phospholipids to enhance membrane association and specificity [17,18]. Together with some preceding amino acid residues, the two motifs have been designated as a hyper-variable region, which is involved in many interactions with effector proteins and the spatiotemporal distribution [17].
Yeast Rho5 and its human homolog Rac1 share the basic features of Rho-GTPases, reflected in the conservation of their primary sequences ( Figure 1A). Thus, the P-loop and the two switch regions involved in nucleotide binding are highly conserved from yeast to humans. The P-loop contains a glycine residue substituted by a valine in an oncogenic, constitutively active human Ras variant (G12V; [19]), and hyperactivation of yeast Rho5 is also observed in the respective mutant as well as the Q91H substitution within the switch II region [9]. The CAAX box in both Rho5 and Rac1 is preceded by a PBR region, which only in the yeast protein carries a serine residue. Rho5 also differs from Rac1 by a 98 amino acid insertion prior to the PBR region, as well as a shorter insertion of 30 amino acids between the switch I and switch II domains ( Figure 1A). The hypervariable region at the C-terminal end of human Rac1 has been suggested to have a major influence on its intracellular distribution and physiological targets [17], and we reasoned that this is where important regulatory properties of the yeast homolog should be encoded, and may be altered by genetic manipulations to study the phenotypic outcome. We therefore set out to investigate the structural requirements that govern the translocation of Rho5 to mitochondria under oxidative stress, other than its obligatory association with the dimeric GEF. In this context, we focused on the role of conserved primary sequence features, especially those located in the C-terminal part of the GTPase, and their role in response to oxidative stress. Knowledge gathered from these experiments was used to construct a functional Rac1-Rho5 hybrid protein that complements a yeast rho5 deletion and may be a valuable tool to study human Rac1 functions. Alignment of the primary sequences of Rho5 and Rac1. Identical amino acid residues are shaded in blue, conserved ones in grey. Domains relevant for the activation of the GTPases (P-loop, switch I and II regions), as well as the conserved sequences at the C-terminal end (PBR = polybasic region; CAAX = site of lipid modification, with a cysteine, two aliphatic and one variable terminal residues) are highlighted. The orange box designates a 98 amino acid extension specific for Rho5 of Saccharomyces cerevisiae. Accession numbers for the coding sequences are KZV08440 for S. cerevisiae Rho5, and CAB53579 for human Rac1. (B) Schematic representation of mutants used in this study. Designations of Rho5 and Rac1 variants used in the text are depicted at the left. Exchanges of amino acid residues are shown in red letters, including the valine near the N-terminus in the G12V mutant. Dark and light grey lines represent the catalytic domains of Rho5 and Rac1, respectively. Rectangles in orange designate the yeast-specific sequences of the 98 aa extension. The polybasic regions are highlighted in dark blue for Rho5 and light blue for Rac1, the CAAX motif in green for Rho5 and yellow for Rac1. For localization studies, all constructs were N-terminally tagged with GFP and expressed from CEN/ARS vectors. * C110 and C17 were exclusively used as N-terminal GFP fusions and not as untagged versions at the genomic locus. (A) Alignment of the primary sequences of Rho5 and Rac1. Identical amino acid residues are shaded in blue, conserved ones in grey. Domains relevant for the activation of the GTPases (P-loop, switch I and II regions), as well as the conserved sequences at the C-terminal end (PBR = polybasic region; CAAX = site of lipid modification, with a cysteine, two aliphatic and one variable terminal residues) are highlighted. The orange box designates a 98 amino acid extension specific for Rho5 of Saccharomyces cerevisiae. Accession numbers for the coding sequences are KZV08440 for S. cerevisiae Rho5, and CAB53579 for human Rac1. (B) Schematic representation of mutants used in this study. Designations of Rho5 and Rac1 variants used in the text are depicted at the left. Exchanges of amino acid residues are shown in red letters, including the valine near the N-terminus in the G12V mutant. Dark and light grey lines represent the catalytic domains of Rho5 and Rac1, respectively. Rectangles in orange designate the yeast-specific sequences of the 98 aa extension. The polybasic regions are highlighted in dark blue for Rho5 and light blue for Rac1, the CAAX motif in green for Rho5 and yellow for Rac1. For localization studies, all constructs were N-terminally tagged with GFP and expressed from CEN/ARS vectors. * C110 and C17 were exclusively used as N-terminal GFP fusions and not as untagged versions at the genomic locus.

The Carboxyterminal CAAX Motif of Rho5 is Required for Plasma Membrane Association and Physiological Function
Sequence alignments revealed a strong conservation between yeast Rho5 and human Rac1, with some specific features unique to the yeast GTPase ( Figure 1A). In order to study the physiological significance of the Rho5 domains, we started by investigating the importance of the conserved C-terminus for its physiological function and intracellular distribution related to oxidative stress. A schematic summary of the mutant alleles studied in this work is presented in Figure 1B. First, we exchanged the cysteine at position 328 for a leucine to eliminate the presumed prenylation and carboxymethylation site and compared this mutant to wild type, a complete rho5 deletion, and a strain expressing the constitutively active RHO5 G12V allele. To avoid phenotypic variations in growth assays, which may be caused by fluctuating plasmid copy numbers, the wild-type gene and the mutant alleles were introduced at the native RHO5 locus using SkHIS3 as a selection marker inserted 3 to the coding sequence. Growth was monitored in rich medium under standard growth conditions and under oxidative stress. Confirming previous observations, the rho5 deletion was hyper-resistant toward hydrogen peroxide, whereas the constitutively active variant RHO5 G12V was hyper-sensitive as compared to wild type ( Figure 2A). As expected, neither allele impaired growth in the absence of oxidative stress. Next, we used the previously observed synthetic lethality of rho5 and sch9 deletions to test the in vivo function of our mutant alleles [15]. Whereas the rho5 deletion did not produce any viable progeny in a sch9 background, the constitutively active RHO5 G12V variant produced small colonies typical of RHO5 sch9 segregants ( Figure 2B). On the other hand, the rho5 C328L mutant, which lacks the cysteine residue required for lipid modification, resembled the characteristics of the rho5 deletion both in growth on rich media with and without oxidative stress and in its synthetic lethality in combination with a sch9 deletion (Figure 2A,B). We concluded that the GTPase is not functional in the absence of lipid modification.
To follow the intracellular distribution, CEN/ARS vectors encoding GFP-Rho5 fusions were introduced into a rho5 deletion strain, which also carried an mCherry mitochondrial marker construct. A strain with the wild-type GFP-Rho5 fusion showed the expected plasma membrane (PM) localization under standard growth conditions and the rapid translocation to mitochondria under oxidative stress, as reported previously ( Figure 2C; [14]). Likewise, the activated GFP-Rho5 G12V resided predominantly at the PM under standard growth conditions but displayed an incoherent distribution within different cells of the same culture upon addition of hydrogen peroxide, varying from mainly mitochondrial (41%), through cytosolic (48%), to some PM-associated (11%; Figure 2C). On the other hand, a GFP-tagged Rho5 T17N mutant, which cannot be activated, did not interfere with mitochondrial localization, since 73% of the cells showed colocalization of the GFP signal with the mitochondrial marker under oxidative stress (Supplementary Figure S1).
As expected from the lack of a lipid modification, GFP-Rho5 C328L entirely failed to localize to the PM in life-cell fluorescence microscopy but appeared rather uniformly distributed within the cytosol under standard growth conditions ( Figure 2C). A substantial portion of cells still showed translocation of the GFP signal to mitochondria upon exposure to oxidative stress (36% as compared to 70% of cells with the wild-type GFP-Rho5 fusion). Thus, although causing a significant reduction, lipid modification is not absolutely required for mitochondrial translocation. Growth and response to oxidative stress of Rho5 G12V and Rho5 C328L mutants. (A) Comparative growth behavior of strains with the indicated RHO5 alleles. Growth curves were recorded in rich medium in the absence or presence of hydrogen peroxide as indicated. (B) Tetrad analyses of crosses from strains carrying the indicated RHO5 allele to a sch9 deletion strain performed on rich medium (YEPD). Note that sch9 deletions grow more slowly and produce much smaller colonies than wildtype or rho5 deletions, and that rho5 sch9 double deletions are not viable. Yellow circles indicate tetrads carrying a wild-type RHO5 allele and a sch9 deletion, red circles mark the double mutants of sch9 with the RHO5 alleles indicated above each image. Parental strains crossed to HOD348-4D (sch9) from left to right were: HD56-RHwt (RHO5), HOD295-7D (rho5), HD56-R8 (RHO5 G12V ), and HD56-RH C328L Figure 2. Growth and response to oxidative stress of Rho5 G12V and Rho5 C328L mutants. (A) Comparative growth behavior of strains with the indicated RHO5 alleles. Growth curves were recorded in rich medium in the absence or presence of hydrogen peroxide as indicated. (B) Tetrad analyses of crosses from strains carrying the indicated RHO5 allele to a sch9 deletion strain performed on rich medium (YEPD). Note that sch9 deletions grow more slowly and produce much smaller colonies than wild-type or rho5 deletions, and that rho5 sch9 double deletions are not viable. Yellow circles indicate tetrads carrying a wild-type RHO5 allele and a sch9 deletion, red circles mark the double mutants of sch9 with the RHO5 alleles indicated above each image. Parental strains crossed to HOD348-4D (sch9) from left to right were: HD56-RHwt (RHO5), HOD295-7D (rho5), HD56-R8 (RHO5 G12V ), and HD56-RH C328L (RHO5 C328L ). See Table 1 for genotypes. (C) Life-cell fluorescence microscopy of a rho5 deletion strain with a genomic IDP1-mCherry fusion as a mitochondrial marker (HCSO76-1A) expressing N-terminal GFP fusions of the Rho5 variants as indicated, each from a CEN/ARS vector. Cells were grown on synthetic media selecting for maintenance of the plasmid and either examined without further treatment (upper panels), or after addition of 4.4 mM hydrogen peroxide for less than 15 min (lower panels). The percentage of cells displaying association of the GFP signal with the plasma membrane (pm), being cytosolic or associated with internal structures (int), and those showing colocalization with the mitochondrial marker (mit) were determined for the total number of cells counted in each sample (n). Plasmids employed were pJJH1639 = GFP-Rho5; pLAO2 = GFP-Rho5 G12V , and pCSO18 = Rho5 C328L ( Table 2). Images in the mCherry channel were obtained by 0.5 s exposures, those in the GFP channel by 2 s exposures. The size bar in the upper left differential interference contrast (DIC) image represents 5 µm, applicable to all images in the panels of the same strain.

Rho5 Requires the Basic PBR Region for Proper Plasma Membrane Association and Physiological Function
To study the function of the PBR sequence preceding the CAAX-box, we first constructed a mutant in which the six lysine residues were replaced by alanines (Rho5 pbrK6A ; Figure 1B). Again, the respective allele was introduced at the native RHO5 locus using SkHIS3 as a selection marker inserted 3 to the coding sequence. The RHO5 pbrK6A strain grew like wild type under standard conditions but showed an increased resistance to hydrogen peroxide, although not reaching that of a rho5 deletion strain ( Figure 3A). This suggested a partial functionality of this mutant with regard to oxidative stress response. Likewise, a synthetic growth defect was observed in crosses with the sch9 deletion ( Figure 3B). Thus, although segregants containing the mutant rho5 pbrK6A allele were viable in the sch9 background, they produced much smaller colonies than those carrying RHO5 sch9. Similar to the CAAX-box mutant, GFP-Rho5 pbrK6A failed to associate with the PM under standard growth conditions, but upon exposure to hydrogen peroxide still translocated to mitochondria in 32% of the cells observed ( Figure 3C). We also noticed that GFP-Rho5 pbrK6A was not dispersed evenly within the cytosol, but frequently associated with ring-like structures, which we could not yet identify, but they were clearly different from mitochondria.
A distinguishing feature of S. cerevisiae Rho5 as compared to human Rac1 is the presence of a serine residue within the PBR, which has been shown to modulate membrane association of another human Ras-type GTPase depending on its phosphorylation state [20]. Site-specific mutants carrying either an alanine residue that cannot be phosphorylated (Rho5 S326A ) or a phosphomimetic glutamate substitution (Rho5 S326E ) at this position both produced a functional Rho5, as judged from their growth curves in the presence of hydrogen peroxide ( Figure 3A) and their viability in a sch9 background ( Figure 3B). While strains producing Rho5 S326A behaved basically like wild-type in the growth experiments, while strains with Rho5 S326E showed a slight increase in H 2 O 2 -resistance ( Figure 3A). GFP-fusions of the GTPase with either the serine or the glutamate substitution localized to the PM under normal growth conditions very similar to wild type. They also translocated predominantly to mitochondria under oxidative stress, but with a larger fraction remaining at the PM as compared to the wild type ( Figure 3C). Mitochondrial translocation under stress appeared to be somewhat reduced for GFP-Rho5 S326E (58%) as compared to the 70% of both GFP-Rho5 ( Figure 2C) and GFP-Rho5 S326A ( Figure 3C).
Taken together, these results indicate that the basic character of the PBR region is required for proper PM association of Rho5 and influences the association of Rho5 with mitochondria under oxidative stress, whereas the serine residue appears to be of minor importance.

The C-Terminal Extension is Required for Rho5 Function in Yeast
A distinctive feature of S. cerevisiae Rho5 versus human Rac1 is the presence of a 98 amino acid extension between residues P221 and D320 prior to the PBR-CAAX regions ( Figure 1A), which does not resemble any protein domains available in the databases, and whose functional significance has not yet been studied. Thus, we removed this sequence (with the resultant protein designated Rho5 ∆222-319 ) and tested the response of the respective mutant to oxidative stress. Compared to the wild-type, rho5 ∆222-319 cells displayed the hyper-resistance typical of a complete rho5 deletion and were synthetically lethal with a sch9 deletion, indicating that the GTPase is not functional in vivo ( Figure 4A,B). Interestingly, a GFP-Rho5 ∆222-319 fusion did not associate with the PM under normal growth conditions and translocation to mitochondria was drastically reduced under oxidative stress ( Figure 4C). Since this GFP-Rho5 ∆222-319 fusion resembled the mutants in the CAAX motif and the pbrK6A variant in their failure to associate with the PM, the minimal requirement of the Rho5 C-terminus for such an association became an issue. In order to address this question, we introduced a number of different GFP fusion constructs (compare Figure 1B) and determined their intracellular distribution.   Figure 2. (B) Tetrad analyses of crosses from strains carrying the indicated RHO5 allele to a sch9 deletion strain performed on YEPD. Parental strains employed in crosses to HOD348-4D (sch9) were HOD337-7D (RHO5 pbrK6A ), HD56-RH S326A (RHO5 S326A ), and HD56-R7 (RHO5 S326E ). Yellow and red circles indicate segregants with a RHO5 sch9 and a mutant RHO5 sch9 constitution, respectively. (C) Life-cell fluorescence microscopy of a rho5 deletion strain carrying IDP1-mCherry as a mitochondrial marker (HCSO76-1A) and the RHO5 alleles indicated as N-terminal GFP fusions introduced on a CEN/ARS vector. Plasmids employed were pCSO35 = GFP-Rho5 pbrK6A , pCSO19 = GFP-Rho5 S326A , pCSO31 = GFP-Rho5 S326E ( Table 2). Growth conditions, scales, and determination of distribution numbers were as described in the legend of Figure 2.   Table 2 for more information on plasmid features. Growth conditions, scales, and determination of distribution numbers were as described in the legend of Figure 2.
In order to pinpoint the requirements for PM association, we first demonstrated that a GFP fusion with the C-terminal 110 amino acids of Rho5 (GFP-Rho5 C110 ), which comprises the entire extension terminated by the PBR/CAAX motifs, correctly localizes to the PM under standard growth conditions ( Figure 4D). It largely remains at that position after addition of hydrogen peroxide, with only a small fraction translocating to mitochondria. As expected from the importance of the basic PBR sequence demonstrated above, substitution of the lysines by alanine residues (K6A) impaired PM localization of this construct. The Rho5 sequence was then further truncated to include only the last 17 amino acids in the GFP fusion (GFP-Rho5 C17 ), leaving three more basic residues prior to the PBR sequence. This fusion localized at the PM in the absence of oxidative stress, whereas translocation to mitochondria was virtually absent after application of hydrogen peroxide, with the majority of the signal observed in the cytosol and non-mitochondrial internal structures ( Figure 4D).
These observations prompted us to revisit the GFP-Rho5 ∆222-319 fusion, which completely lacked PM localization. We introduced the five amino acids preceding the PBR sequence present in the GFP-Rho5 C17 construct, to obtain GFP-Rho5 ∆222-314 . Indeed, with this variant GFP signals were predominantly found at the PM under normal growth conditions and the fusion protein behaved very similar to GFP-Rho5 C17 under oxidative stress, with the majority of signals detected in the cytosol and internal structures. However, a higher percentage of cells displayed colocalization with mitochondria for GFP-Rho5 ∆222-314 under these conditions ( Figure 4C). Nevertheless, substitution of the wild-type RHO5 allele by the one encoding Rho5 ∆222-314 neither reduced the hyper-resistance of the cells toward hydrogen peroxide in growth experiments ( Figure 4A) nor compensated the synthetic lethality with a sch9 deletion ( Figure 4B), indicating that the additional five amino acids cannot restore the in vivo function of the internally truncated GTPase.
In summary, while the C-terminal 17 amino acids of Rho5 are sufficient to direct the protein to the PM, neither these nor the entire C-terminal 110 amino acids confer an efficient translocation to mitochondria under oxidative stress.

In Vivo Trapping of GFP-Rho5 to the Mitochondrial Surface Affects Its Physiological Functions
To further address the question of the physiological significance of Rho5 localization, a GFP trap tethering the GTPase to the mitochondrial surface was employed. To construct this mitochondrial GFP-trap, we fused the GFP-binder sequence to the N-terminus of the transmembrane domain of the outer mitochondrial protein Fis1 on a yeast integrative plasmid and inserted it at the leu2-3,112 locus. As expected from the presence of a wild-type FIS1 allele in these strains, segregants carrying this construct did not display any growth defects in tetrad analyses. Such a segregant was crossed with a strain carrying a GFP gene fusion at the RHO5 locus to obtain haploid strains with combinations of GFP-RHO5, FIS1 TMD -GB and the respective wild-type loci. Fluorescence microscopy demonstrated that the traps worked as expected, i.e., they either recruited GFP-Rho5 exclusively to mitochondria in the GFP-RHO5, FIS1 TMD -GB segregants, both in the absence and presence of hydrogen peroxide ( Figure 5A). Growth curves showed that segregants with separate GFP-RHO5 and FIS1 TMD -GB alleles reacted similar to wild type to hydrogen peroxide ( Figure 5B). On the other hand, the strain expressing both GFP-RHO5 and FIS1 TMD -GB resembled the hyper-resistance observed for the rho5 deletion, demonstrating that the GTPase is not functional when trapped exclusively to mitochondria. We reasoned that this phenotype could be attributed to a necessity for Rho5 to be activated at the PM prior to its translocation in order to fulfil its function in oxidative stress response. Therefore, we introduced an activated GFP-RHO5 G12V allele at the native RHO5 locus and combined it with a FIS1 TMD -GB background by crossing and tetrad analysis. Segregants with GFP-Rho5 G12V trapped to the mitochondria displayed wild-type sensitivity toward hydrogen peroxide demonstrating in vivo functionality of the GTPase, while strains only producing GFP-Rho5 G12V without the trapping device proved to be hyper-sensitive, as observed above for the non-tagged version ( Figure 5C).  Table 1. Growth curves were recorded in synthetic medium in the absence or presence of hydrogen peroxide as indicated. (C) Comparative growth behavior of strains with an activated GFP-RHO5 G12V allele. Growth curves were recorded in synthetic medium as described in (B).
GFP fusion constructs showed that while the unmodified GFP-Rac1 did not localize at the PM or mitochondria both in the absence and in the presence of hydrogen peroxide, GFP-Rac1-Rho5 C110 appeared almost exclusively at the PM under normal growth conditions and a substantial portion translocated to mitochondria under oxidative stress ( Figure 6C). Finally, substitution of just the Rac1 PBR and CAAX motifs for yeast Rho5 C17 conferred PM localization to the respective GFP fusion under standard growth conditions but failed to translocate to mitochondria under oxidative stress ( Figure  6C).  Table 1. Growth curves were recorded in synthetic medium in the absence or presence of hydrogen peroxide as indicated. (C) Comparative growth behavior of strains with an activated GFP-RHO5 G12V allele. Growth curves were recorded in synthetic medium as described in (B).

A C-Terminally Modified Human Rac1 Partially Complements a rho5 Deletion
Since impairment of human RAC1 functions is associated with a variety of diseases, we decided to explore the possibility of using S. cerevisiae as a heterologous host for its expression and molecular analysis. For this purpose, a codon-optimized RAC1 sequence was cloned into a CEN/ARS vector and expressed either under the control of the native RHO5 promoter or that of the strong glycolytic PFK2 promoter. Neither construct complemented the rho5 deletion, as judged from the retained hyper-resistance toward oxidative stress ( Figure 6A). In order to rule out that the lack of complementation was due to a failure of Rac1 to be activated by the yeast GEF Dck1/Lmo1, a constitutively active RAC1 G12V allele was also expressed, but did not complement the rho5 defects, either.  Table  2). (B) A heterozygous diploid obtained by crossing HAJ216-A (rho5) with HAJ195-A (sch9) was used as a recipient for the plasmids described in (A), with omission of pLAO4, and transformants were subjected to tetrad analyses. Yellow circles highlight segregants with a sch9 deletion only, red circles non-viable segregants with the genotype rho5 sch9, and red squares designate double deletion segregants rescued by the plasmid-borne gene copy. (C) Life-cell fluorescence microscopy of a rho5 Figure 6. Growth and response to oxidative stress of yeast strains producing human Rac1 and its derivatives. (A) Comparative growth behavior of transformants of a rho5 deletion (HAJ216-A) expressing the indicated GTPase variants from a centromeric vector. Plasmids introduced were YCplac111 (vector), pJJH1637 (pRHO5 wt ), pLAO4 (pRAC1), and pCSO85 (pRAC1-RHO5 C110 ; see Table 2). (B) A heterozygous diploid obtained by crossing HAJ216-A (rho5) with HAJ195-A (sch9) was used as a recipient for the plasmids described in (A), with omission of pLAO4, and transformants were subjected to tetrad analyses. Yellow circles highlight segregants with a sch9 deletion only, red circles non-viable segregants with the genotype rho5 sch9, and red squares designate double deletion segregants rescued by the plasmid-borne gene copy. (C) Life-cell fluorescence microscopy of a rho5 deletion strain producing Idp1-mCherry as a mitochondrial marker (HCSO76-1A) and N-terminal GFP fusions of the Rac1 variants indicated, expressed from a CEN/ARS vector. Plasmids employed were pLAO6 (Rac1), pCSO89 (Rac1-Rho5 C110 ), and pCSO100 (Rac1-Rho5 C17 ). See Table 2  Based on the functional significance of the C-terminal insertion of 98 amino acids for the yeast GTPase described above, we created a chimeric construct, in which coding regions for the PBR and CAAX motifs of RAC1 were replaced by the C-terminal sequences of RHO5 (encoding Rac1-Rho5 C110 ). Upon overexpression in a rho5 deletion, this construct restored the sensitivity toward hydrogen peroxide even more than that of the wild type, demonstrating its functionality in yeast ( Figure 6A). Complementation capacity was also confirmed by viability of strains carrying this plasmid in a rho5 sch9 double deletion background ( Figure 6B). It should be noted that, although growth could be restored in these transformants, colonies of the viable segregants were smaller than those carrying only the sch9 deletion with a RHO5 wild-type allele.
GFP fusion constructs showed that while the unmodified GFP-Rac1 did not localize at the PM or mitochondria both in the absence and in the presence of hydrogen peroxide, GFP-Rac1-Rho5 C110 appeared almost exclusively at the PM under normal growth conditions and a substantial portion translocated to mitochondria under oxidative stress ( Figure 6C). Finally, substitution of just the Rac1 PBR and CAAX motifs for yeast Rho5 C17 conferred PM localization to the respective GFP fusion under standard growth conditions but failed to translocate to mitochondria under oxidative stress ( Figure 6C).

Discussion
Although Rho5 has been identified as the last of the six yeast Rho-family members more than 15 years ago, then as a negative regulator of CWI signaling, its exact mode of action in various physiological processes is not yet fully understood [8]. Neither is the role of its protein domains, which comprise evolutionary conserved sequences as well as features specific for S. cerevisiae Rho5. The mutational analyses presented herein provide first experimental evidence for the functional significance of these domains in vivo.
For Rac1, a human homolog of yeast Rho5, a puzzling variety of physiological functions has been reported, which to a large extent may be mediated by what was defined as a hyper-variable region (HVR) at the C-terminal end of Rho-type GTPases (reviewed in reference [17]). The HVR comprises both the CAAX box with the cysteine residue as a target for lipid modification and the polybasic region (PBR), which are both conserved in yeast Rho5 (Figure 1), as well as some preceding amino acid residues. Substitution of the CAAX-box cysteine at position 189 in Rac1 for a serine abolished both prenylation and in vivo function [21]. Likewise, our Rho5 C328L variant proved to be non-functional in two different physiological read-outs, i.e., the hyper-resistance against hydrogen peroxide typical for complete rho5 deletions as seen in growth curves (oxidative stress response; [14]), and the synthetic lethality in combination with a sch9 deletion (carbon stress response; [15]). In vivo fluorescence of the GFP-tagged Rho5 C328L demonstrated that it does no longer associate with the PM under standard growth conditions, in contrast to the wild-type version. Nevertheless, the rapid translocation of GFP-Rho5 to mitochondria upon application of oxidative stress is not completely abrogated in the prenylation mutant. While this indicates that some interaction with the dimeric GEF Dck1/Lmo1 necessary for mitochondrial association is still possible, the moiety reaching the mitochondria is clearly not sufficiently active to mediate the oxidative stress response. Moreover, this provided a first hint that the GTPase has to be activated at the PM prior to its translocation in order to fulfil its physiological function. We presume that the less efficient mitochondrial translocation of the activated Rho5 G12V variant is due to a reduced affinity toward the dimeric GEF, since the latter should interact primarily with the non-activated conformation of the GTPase. Accordingly, the non-activatable Rho5 T17N translocated like wild type to mitochondria under oxidative stress, indicating that its interaction with Dck1/Lmo1 is not impaired.
PM association may also be aided by electrostatic interactions of the basic residues within the PBR with membrane phospholipids, depending on the combination of their arginine and lysine residues [22]. We here showed that the PBR of Rho5 is necessary but not sufficient for proper membrane targeting. More importantly, quantitative trapping of GFP-Rho5, but not that of its activated GFP-Rho5 G12V variant, resulted in a loss of function in the oxidative stress response. This further supports the notion that the GTPase needs to be activated at the PM prior to its translocation to mitochondria, in line with the general assumption that activation of Rho-type GTPases occurs while they are associated with membranes [23]. For Rho5, this activation would be mediated at the PM by its dimeric GEF Dck1/Lmo1, which is also required for its translocation to mitochondria [14]. This is reminiscent of the situation in mammalian cells, where insulin-induced activation of Rac1 is mediated by interaction with its GEF at the PM [24]. It is also consistent with other data on Rac1, where prenylation of the cysteine residue was shown to be required for its activation and its subsequent interaction with target proteins, which included the response to oxidative stress [25].
For Rac1, the HVR not only mediates membrane association through the prenylated cysteine residue and the PBR, but also determines its specific interaction with effector proteins and its subcellular distribution, which frequently involves some preceding amino acid residues [17]. We here found that a yeast-specific insertion of 98 amino acids prior to the PBR is required, but not sufficient, to ensure the rapid translocation of Rho5 to mitochondria under oxidative stress and trigger the proper cellular response. Thus, the phenotypes of Rho5 ∆222-319 and Rho5 ∆222-314 mutants resemble those of the complete rho5 deletion, demonstrating its functional importance. However, GFP-Rho5 C110 , which carries the entire insertion and the native C-terminal end, translocated poorly to mitochondria. This suggests that N-terminal sequences, presumably the switch regions interacting with the dimeric GEF Dck1/Lmo1, are also required for this process, as previously observed for interactions of Rac1 with some effectors [17]. We therefore propose that the yeast-specific insertion mediates the interaction of Rho5 with its effectors. Supporting this notion is the hydrophilic, probably surface-exposed nature of this sequence predicted by secondary structure analyses. In a recent paper, Singh and co-workers reported on a large number of putative Rho5 interactors, comprising, amongst others, the two eisosome components Lsp1 and Sur7 [26]. Lsp1 directs the membrane invaginations typical for eisosomes through its BAR domain [27]. Mammalian Rac1 was also found to interact with the SH3 domains of BAR domain proteins through its proline residues in the HVR, generating intracellular tubular structures [28]. While it is thus tempting to speculate that the tubular structures observed in some of our GFP-fusions with different Rho5 variants are caused by a dysregulation of its interaction with Lsp1, this remains to be determined.
Another interesting interaction was reported between mammalian Rac1 and a protein kinase C-related kinase (PRK1; [29]). PRKs are characterized by the presence of homology repeats, so-called HR1 domains, at their N-terminal end, which are also present in the unique, prototypic protein kinase C (Pkc1) of S. cerevisiae and interact with the Rho1-GTPase [30]. Pkc1 is a key component of the CWI pathway, to which Rho5 function was found to be related [9]. An evolutionary conserved interaction between Rho5 and Pkc1 would thus be an option, although it was neither detected in the work of Singh and co-workers [26] nor in direct two-hybrid interaction assays with the HR1 domains of Pkc1 [31]. Yet, since both assays were performed under standard growth conditions, it cannot be ruled out that an interaction might still occur in specific stress situations.
Mammalian PKC was found to phosphorylate K-Ras at a serine residue within its PBR, causing the dissociation of the GTPase from the PM and its association with mitochondria and other endomembranes [20]. Other GTPases of the Ras-superfamily were also found to be phosphorylated at serine residues within the PBR in order to modulate their interaction with downstream effectors [32][33][34][35][36]. In contrast, our data show that substitutions of serine 326 in the PBR of yeast Rho5 did neither drastically effect its distribution to different membranes, nor its physiological functions determined by the two different outreads. The minor increase in resistance of Rho5 S326E toward growth in the presence of hydrogen peroxide as compared to wild type could indeed be due to a weakened PM association caused by an electrostatic repulsion of the negative charge introduced. Although we thus cannot exclude the possibility that phosphorylation of serine 326 by Pkc1 or other kinases occurs in Rho5, we believe that it would be of minor regulatory importance, at least for the oxidative stress response and the reaction to carbon stress. Large-scale phosphoproteome studies performed under standard growth conditions further support this notion, since several serine residues appeared to be phosphorylated in Rho5 but not serine 326 [37].
A common feature of the HVR in mammalian GTPases is the presence of a nuclear localization sequence (NLS) within the PBR [18], which is also present in Rho5. In Rac1, the NLS and the preceding proline residues determine its cell-cycle dependent import into the nucleus [38]. The authors found that lipid modification of the CAAX-box cysteine impairs this function, since cysteine-substitution mutants showed a much stronger nuclear localization. In contrast, yeast GFP-tagged Rho5 C328L showed no sign of nuclear localization in the fluorescence images, and neither did the GFP fusion to the C-terminal 17 amino acids. We thus believe that a nuclear function of yeast Rho5 is highly unlikely.
In summary, the data discussed above show that a correct spatiotemporal distribution of Rho5, its activation either exclusively or predominantly at the PM, and its specific interactions with downstream effectors work hand in hand to ensure its physiological functions. Last but not least, these finding also enabled us to construct a hybrid human Rac1-Rho5 protein which at least partially complemented the rho5 deletion for both physiological read-outs. Besides providing the opportunity to study clinically important Rac1 variants in yeast, these results further underlined the importance of the yeast-specific insertion for proper spatiotemporal distribution and physiological function. It will be interesting to see if and how the Rho5 domains and their functions are conserved in other fungal Rho5 homologs, given their impact on hyphal morphologies with biotechnological applications [39].

Strains and Growth Conditions
All yeast strains used are derived from the haploid strain HD56-5A. An isogenic diploid was previously obtained from this strain by introducing the HO gene on a plasmid and subsequent selection for plasmid loss [40,41]. HD56-5A is also closely related to the commonly employed CEN.PK series [42], as it was one of the parental strains used in the construction of these strains. Yeast cells were cultured in liquid media or on plates at 30 • C. Crossing, tetrad analyses and other genetic techniques followed standard procedures as described in reference [43]. Rich medium (YEPD) contained 1% yeast extract, 2% Bacto peptone (Difco Laboratories Inc., Detroit, MI, USA) and 2% glucose (all w/v). Synthetic media were prepared with Difco yeast nitrogen base (Difco Laboratories Inc., Detroit, MI, USA) with ammonium sulphate as described in reference [43], with the addition of amino acids and bases using a mixture provided by MP Biomedicals (Eschwege, Germany; CSM-His-Leu-Trp-Ura) supplemented histidine, leucine, tryptophan, and uracil, with omission of either histidine, leucine, or uracil, if required for selection of plasmid maintenance or when the respective wild-type alleles of the encoding genes were employed as deletion markers. Two percent glucose (w/v) was added as carbon source (SCD). Then, 200 mg/L of G418 were used for selection of the kanMX marker. Hydrogen peroxide was added at the concentrations indicated to induce oxidative stress.
To record growth curves, cells were pre-grown overnight in either rich or synthetic medium at 30 • C on a shaker set at 180 rpm. Cells were inoculated in fresh medium at an OD 600 of 0.3 and grown again to mid-logarithmic phase. They were diluted to an OD 600 of 0.1 in fresh medium and distributed in 100 µL aliquots into 96 well plates (Thermo Scientific, Bremen, Germany), with or without hydrogen peroxide as indicated. Growth was recorded in a Varioscan Lux plate reader (Thermo Scientific, Bremen, Germany) following the increase in OD 600 by measuring each well once every 30 min for at least 18 h. Temperature was set at 30 • C and plates were moved with 1024 rpm with 5 s intervals, using the SkanIt Software 4.1 (Thermo Scientific, Bremen, Germany) for microplate readers, version 4.1.0.43. A factor of 3.546 was used for normalization with standard 1 cm cuvettes, as the path length in the cell culture was 0.282 cm. A minimum of two biological replicates (usually different segregants with the same mutant alleles obtained from tetrad analyses) and two technical replicates were recorded to generate each growth curve from mean values giving standard error bars.
Tetrad analyses were performed on YEPD plates using zymolyase 20T (MP Biomedicals) for digestion of the ascus walls and a Singer Instruments micromanipulator (Singer Instruments, Somerset, UK) as described in reference [15]. Plates were incubated for 3 days at 30 • C to allow for germination and growth of colonies. Images of the plates were then recorded by scanning with an Epson Perfection V500 Photo Scanner (Suwa, Nagano Prefecture, Japan). Brightness and contrast were adjusted for entire plates using Corel Photo Paint (Corel, Ottawa, ON, Canada). prior to selection of exemplary four tetrads each. A minimum of 18 tetrads were examined for each cross.

Plasmid and Strain Constructions
Modifications at the original genetic loci were achieved by using one-step gene replacement techniques [45]. To avoid problems with copy numbers and plasmid stability, RHO5 variants were tagged and used to substitute the gene at its original locus. In brief, cloned variants were equipped with the SkHIS3 cassette from the Longtine collection [46] by in vivo recombination in their 3 -non-coding regions, flanked by sequences homologous to the native RHO5 locus. From there, genomic sequences were replaced by homologous recombination selecting for histidine prototrophy. Mitochondria were visualized by a C-terminal Idp1-mCherry fusion encoded at the native IDP1 locus, obtained by in vivo recombination with a PCR-generated cassette using SkHIS3 as a selection marker. Sequences of all fusion constructs at their genetic loci are available upon request.
Trapping of GFP-Rho5 to mitochondria was achieved by fusion of the sequence encoding a GFP-binding monomeric antibody (GB; shortly referred to as "GFP-binder"), obtained by PCR from a template kindly provided by Roland Wedlich-Söldner [47], to the one encoding the transmembrane domain of the outer mitochondrial membrane protein Fis1 (GB-Fis1) in the yeast integrative plasmid YIplac128 [48] to obtain plasmid pLAO12. The GB-FIS1 gene was expressed under the control of a constitutive yeast PFK2 promoter tailored based on published promoter analyses [49], with 576 bp prior to the ATG translation start codon, which was previously shown to be suitable for heterologous expression of plant genes in yeast [50]. Intracellular distribution of GFP-Rho5 encoded on CEN/ARS vectors derived from YCplac33 or YCplac111 [48] was followed by fluorescence microscopy. Mutations at the 3 -end of the RHO5 open reading frame were introduced by inverse PCR using appropriate oligonucleotides. For human RAC1, string-DNA synthesis was ordered from GeneArt (Thermofisher, Bremen, Germany) with the open reading frame optimized for S. cerevisiae codon usage and first cloned into pUK1921 for sequence verification [51]. Overexpression of a RAC1-RHO5 hybrid construct obtained by in vivo recombination was achieved by placing the coding sequence under the control of the strong TEF2 promoter from S. cerevisiae carried on YCplac111 [48]. A list of all plasmids employed is given in Table 2, and their sequences and details on their constructions are available upon request.

Fluorescence Microscopy
For standard microscopic examination cells were grown to early logarithmic phase in SCD medium. To apply oxidative stress, 4.4 mM hydrogen peroxide was added for 5-15 min prior to image acquisition throughout all localization studies, since these conditions were optimized previously [14]. Fluorescence microscopy was performed with a Zeiss Axioplan 2 (Carl Zeiss, Jena, Germany) microscope equipped with a 100× alpha-Plan Fluor objective (numerical aperture 1.45) and differential-interference contrast (DIC), as described previously [14]. Images were acquired using a Photometrics CoolSNAP HQ Camera (Roper Scientific, Tucson, AZ, USA). Fluorescence was excited with a SPECTRA X light engine (Lumencor, Beaverton, OR, USA). For colocalization studies, the microscope was additionally equipped with an image-splitter (DualView, Photometrics, Tucson, AZ, USA). The setup was controlled by the Metamorph v6.2 program (Universal Imaging Corporation, Downingtown, PA, USA). Images in the mCherry channel were obtained by 0.5 s exposures, and those in the GFP channel by 2 s exposures. Brightfield images were acquired as single planes using DIC. Scale bars were generated using Metamorphs scale image command. Huygens essential software (Scientific Volume Imaging, Hilversum, Netherlands) was used for deconvolution of the images. In order to visualize colocalization of signals from the mCherry-and the GFP-channels from the same cell, the processed images were overlaid using Metamorphs' overlay images command.
A minimum of 100 cells (usually much more, with the total number "n" given in the figures) was examined for each strain and condition. Distributions of Rho5-GFP signals were judged to belong to one of three patterns, and the number of cells displaying association of the signal with the plasma membrane (pm), being either cytosolic or associated with internal structures (int), and those showing colocalization with the mitochondrial marker (mit) were determined. Acknowledgments: Technical assistance was provided by Andrea Murra and some preliminary experiments were carried out by Aileen Faist under the supervision of C.S. We thank Rosaura Rodicio for critical reading of the manuscript and fruitful discussions.

Conflicts of Interest:
The authors declare no conflict of interest.