2.1. Overview of the Ste20 Split-Ubiquitin Screen
In order to improve our understanding of Ste20 functions and mechanisms, we performed a split-ubiquitin screen to identify proteins that bind to Ste20. The split-ubiquitin technique is based on the ability of the N-terminal and C-terminal domains of ubiquitin to form a quasi-native ubiquitin (
Figure 1) [
23,
24]. If two proteins, which are attached to the N-terminal and C-terminal portion, respectively, bind to each other, the ubiquitin halves may be forced into proximity and a ubiquitin-like molecule assembles. Ubiquitin-specific proteases, present in the cytoplasm and nucleus, recognize the reconstituted ubiquitin but not its individual halves, and cleave off a reporter that is attached to the C-terminal ubiquitin portion. The technique described here employs a modified version of Ura3, an enzyme essential for uracil biosynthesis, as reporter [
24]. This Ura3 variant carries an additional arginine at the extreme N-terminus (RUra3). RUra3 attached to the C-terminal half of ubiquitin is stable and functional. In contrast, RUra3 cleaved off by a ubiquitin-specific protease is rapidly degraded because arginine is a destabilizing residue in the N-end rule pathway. Therefore, the interaction between two proteins fused to the N-terminal and C-terminal halves of ubiquitin results in uracil auxotrophy. Conversely, growth on 5-fluoroorotic acid (5-FOA) indicates a protein–protein interaction because Ura3 converts 5-FOA into the toxic compound 5-fluorouracil.
The split-ubiquitin technique has a number of advantages that make it the ideal tool for the identification of Ste20 interactors. This method is suitable for monitoring interactions with membrane proteins [
23,
24,
25] which is critical because Ste20 associates with the plasma membrane and vacuolar membranes [
14,
15,
16]. The split-ubiquitin assay can also detect weak and transient interactions in vivo [
23,
25], which allows for the isolation of substrates of the kinase Ste20.
In the screen, we identified 56 proteins as putative interactors of Ste20 (
Table 1). Five proteins have been shown by other groups to bind to Ste20 (
Table 2). Out of these, the interaction between Bem1 and Ste20 is well established (
Table 2). Like Ste20, Bem1 has a role in the three MAPK pathways’ regulating pheromone response [
26,
27], filamentous growth [
28] and hyperosmotic stress response [
28]. As a central scaffold protein, Bem1 not only binds to Ste20 but also to the related PAK Cla4 and their upstream activator Cdc42 [
26,
29,
30,
31]. Large-scale screens have also revealed that Mlc1 and Ssb2 physically interact with Ste20, and that Spb1 and Ubx7 are phosphorylated by Ste20 (
Table 2). Notably, the Bem1-Ste20 and Mlc1-Ste20 interactions observed by others were also detected using the split-ubiquitin technique (
Table 2). The fact that other groups independently isolated some of the proteins described here indicates that this approach robustly identifies Ste20 interactors.
We have previously characterized the interactions between Ste20 and five proteins identified in this screen (Sut1, Ncp1, Cbr1, Erg4, Vma13) (
Table 2). Importantly, it has also been shown that Ste20 modulates the processes these proteins mediate, namely sterol biosynthesis (Erg4, Cbr1 and Ncp1) [
32,
33], sterol uptake (Sut1) [
17] and V-ATPase activity (Vma13) [
14]. Taken together, we present here 46 putative Ste20 interactors not described previously.
To find out whether these interactions are physiologically relevant, we further analyzed them by including published observations. First, the interactions identified in the screen presented here seem to be highly specific. We performed another split-ubiquitin screen using the same library and Rdi1 as bait (
Table S1). We have chosen Rdi1 because, like Ste20, Rdi1 binds to Cdc42 [
34,
35]. However, Rdi1 is not a downstream effector of Cdc42. Instead, Rdi1 regulates the localization of Cdc42 by extracting it from membranes [
34,
36]. Rdi1, therefore, like Ste20, localizes to the cytoplasm, the plasma membrane and vacuolar membranes [
34,
36,
37]. Fifty different proteins have been identified in the Rdi1 split-ubiquitin screen (
Table S1). Only one protein (Mlc1) came up in both screens, even though Ste20 and Rdi1 both function around Cdc42 and both localize to the same compartments. This demonstrates that the interactions described here are highly specific. Using the split-ubiquitin assay, Ste20 and Rdi1 physically interact with a very different set of proteins.
The localization of the proteins identified here also suggests that the interactions are real. Importantly, all of the Ste20 interactors can be found in compartments where they could physically interact with Ste20. The huge majority of proteins localize to the cytoplasm and/or the nucleus (
Table 1). Some proteins that associate with the plasma membrane, and the membranes of the endoplasmic reticulum (ER) and vacuoles, are peripheral membrane proteins on the cytoplasmic side, such as the aforementioned Bem1 and the vacuolar V-ATPase subunit Vma13 [
26,
38]. For other proteins, at least a portion of the protein faces the cytoplasm. For example, a topology analysis of Ale1, an enzyme involved in glycerophospholipid biosynthesis, has shown that the protein spans the ER membrane multiple times [
39]. However, Ale1 also contains several cytoplasmic loops and a cytoplasmic C-terminus that is over 100 residues long. These proteins would, therefore, be accessible for Ste20.
For three proteins (Ima5, Kin82 and Pps1), subcellular localizations are not known. However, considering their functions and physical interactions with other proteins, it seems very likely that these proteins localize to the cytoplasm and/or nucleus. For example, Pps1 is involved in DNA replication [
40] and physically interacts with seven proteins (Ade13, Ccr4, Dhh1, Isw1, Mpt5, Nab2 and Ser3) all of which localize to the nucleus and/or cytoplasm [
41,
42,
43,
44,
45,
46]. It is, therefore, reasonable to assume that Pps1 can also be found in the nucleus and/or cytoplasm.
Ste20 plays key roles in filamentation, the pheromone response and the hyperosmotic stress response [
5,
6,
7,
8,
9,
10]. Initially, Ste20 has been shown to activate the respective MAPK cascades through Ste11 [
11,
12]. More recently, it has become clear that other Ste20 functions also contribute to these processes. For example, under hyperosmotic conditions, Ste20 phosphorylates histone H4 to attenuate transcription [
47] and Mrc1 phosphorylation by Ste20 prevents genomic instability caused by the collision of replication and transcription machineries [
48]. Since hyperosmotic stress response, pheromone response and filamentous growth are the best characterized processes of Ste20, we searched whether any of the proteins identified in the Ste20 split-ubiquitin screen are also involved in these processes. This is the case for 30 out of the 56 proteins identified in the screen (
Table 3). Based on mutant phenotypes, twenty proteins play a role in filamentation, eleven in hyperosmotic stress response, and seven in pheromone response. This suggests considerable functional overlap between Ste20 and the newly identified interactors.
We also checked published genetic interactions. With the exception of
BEM1, none of the genes identified in the split-ubiquitin screen have been reported to interact with
STE20.
BEM1 overexpression rescues mutant phenotypes of cells lacking
STE20 [
49,
50]. This is not surprising since the scaffold protein Bem1 binds activated Cdc42 and several of its effectors including Ste20 and Cla4. Increased Bem1 levels could lead to the activation of a parallel pathway that shares a function with Ste20.
Negative genetic interactions between
CLA4 and genes identified in the
STE20 split-ubiquitin screen are also interesting. The deletion of either
STE20 or
CLA4 does not affect the growth rate but cells lacking both genes are inviable [
51]. A negative genetic interaction between
CLA4 and another gene could mean that this gene shares a function with
CLA4 independently of
STE20 (
Figure 2). Alternatively, this gene could encode a factor that acts in the same pathway as Ste20, either as upstream activator or downstream effector (
Figure 2). In total, 12 out of the 56 genes identified in the
STE20 split-ubiquitin screen have previously been reported to display negative genetic interactions including synthetic lethality with
CLA4 (
Table 4). Importantly, none of the genes identified in the
STE20 split-ubiquitin screen have been reported to exhibit negative genetic interactions with
STE20, demonstrating that the genetic interactions with
CLA4 are highly specific. The combination of Ste20 protein–protein interactions and
CLA4 genetic interactions described here supports the notion that the corresponding proteins bind to Ste20 under physiological conditions and act in the same pathway as Ste20 (
Figure 2).
The proteins identified in the Ste20 split-ubiquitin screen can be grouped into a relatively small number of functional categories such as translation and protein folding, glycolysis and vesicle transport (
Table 1). This suggests that the proteins identified in the screen are not a random collection. Instead, the functional categories hint at potential cellular functions of Ste20, most of them having not been described previously. Several proteins could be in more than one functional group. For example, Sgt1 acts as a linker between Hsp90 and ubiquitin ligase complexes [
52,
53,
54]. Another example is Sut1, a nuclear protein [
55] that is listed here under lipid metabolism and homeostasis [
56].
Taken together, we identified here 56 proteins that bind to Ste20. Only a very small number of these interactions (Bem1, Cbr1, Erg4, Ncp1, Sut1 and Vma13) have previously been characterized, leaving 50 uncharacterized interactions. Our analysis and published data, including subcellular localizations, genetic interactions, protein–protein interactions and mutant phenotypes, suggest that these interactions are probably physiologically relevant.
2.2. Ste20 and Glucose Metabolism
Four of the proteins identified as Ste20 interactors are glycolytic enzymes (
Table 1), a quite high number considering that glycolysis is a series of only ten reactions (
Figure 3). We wanted to know whether other glycolytic enzymes also bind to Ste20. Since glycolysis, gluconeogenesis and the pentose phosphate pathway are intimately linked (
Figure 3) [
57,
58], all enzymes, major and minor isoforms [
59], of these pathways were tested for interaction with Ste20. Unexpectedly, all enzymes, with the exception of the two pyruvate carboxylase isoforms Pyc1 and Pyc2, bound to Ste20 (
Figure 4A). Notably, these two proteins are not specific for gluconeogenesis but are also involved in other pathways such as amino acid biosynthesis and the citric acid cycle. Despite the huge number of enzymes binding to Ste20, these interactions seem to be highly specific. Ste20 does not interact with other metabolic enzymes such as β-isopropylmalate dehydrogenase (Leu2) (
Figure 4A). This enzyme catalyzes a reaction in leucine biosynthesis, a pathway that is derived from pyruvate, the end product of glycolysis. Like all glucose metabolism enzymes tested here, Leu2 also localizes to the cytoplasm [
60]. Thus, Ste20 interactions seem to be limited to glucose metabolism enzymes. Furthermore, none of the glucose metabolism enzymes bound to Rdi1 (
Figure 4A), demonstrating again the high specificity of these interactions. Twf1, a cytoplasmic protein involved in actin filament organization [
61], which was identified in the Rdi1 split-ubiquitin screen (
Table S1), was included as another control. Twf1 does not interact with Ste20 but, as expected, binds to Rdi1 (
Figure 4A). Bem1 and Cdc42 also served as controls. As expected, Cdc42 binds to both Ste20 and Rdi1, whereas Bem1 only forms a complex with Ste20 but not with Rdi1. These findings are all in line with published observations (
Table 2) [
15,
34,
35,
62,
63] and highlight the specificity of Ste20 interactions using this assay.
It is noteworthy that the split-ubiquitin technique not only detects direct protein–protein interactions but also picks up indirect interactions [
64,
65]. It is, therefore, not clear whether the observed interactions with Ste20 are direct or indirect. Nevertheless, because of the high number of enzymes binding to Ste20, we tried to confirm at least one of the interactions using an independent experimental approach. To this end, immunoprecipitation of Ste20 and Pgk1, a protein that was identified in the screen, was attempted. Ste20 co-precipitated with Pgk1 (
Figure 4B), suggesting that the results of the split-ubiquitin assay reflect physiological interactions.
The pentose phosphate pathway plays a crucial role in the generation of reduced nicotinamide adenine dinucleotide phosphate (NADPH) in the cytoplasm (
Figure 3) [
57,
58]. NADPH is required for the regeneration of reduced glutathione, one of the most important antioxidants in the cell. The synthetic compound diamide oxidizes glutathione, resulting in a decrease in the cytoplasmic pool of reduced glutathione, which causes oxidative stress [
66,
67]. It is, therefore, not surprising that changing the activity of glucose metabolism enzymes that can alter NADPH levels has an effect on diamide sensitivity. For example, cells with reduced catalytic activity of the glycolytic enzyme triosephosphate isomerase (Tpi1) accumulate NADPH which confers diamide resistance [
68,
69]. Since Ste20 binds to glucose metabolism enzymes, it is tempting to speculate that Ste20 modulates the activity of these enzymes. It was, therefore, tested whether altering
STE20 levels affects diamide sensitivity.
STE20 overexpression resulted in an increase in diamide sensitivity (
Figure 5A), whereas cells lacking
STE20 exhibited diamide resistance (
Figure 5B). This suggests that Ste20 negatively modulates NADPH levels possibly through the regulation of glucose metabolism enzymes.
Figure 3.
Overview of glycolysis, gluconeogenesis and the pentose phosphate pathway. In glycolysis (left, red arrows), glucose is oxidized to pyruvate. In gluconeogenesis (left, red arrows), pyruvate is converted to glucose 6-phosphate. Most of the gluconeogenic enzymes are also used in glycolysis. Gluconeogenesis-specific enzymes are Pyc1, Pyc2, Pck1 and Fbp1. Unlike some higher eukaryotes, budding yeast has no glucose 6-phosphatase. In contrast to many higher eukaryotes, the enzymes catalyzing the conversion of pyruvate to oxaloacetate (Pyc1 and Pyc2) are not mitochondrial but cytoplasmic [
70]. In the oxidative branch of the pentose phosphate pathway (right, green arrows) glucose 6-phosphate is converted into ribulose 5-phosphate. This part includes two reactions that generate NADPH. In the non-oxidative branch of the pentose phosphate pathway (right, black arrows), pentoses such as ribulose 5-phosphate and the glycolytic/gluconeogenic intermediates fructose 6-phosphate and glyceraldehyde 3-phosphate are interconverted. Some reactions are catalyzed by multiple redundant isoenzymes. In contrast, Pfk1 and Pfk2 are two distinct subunits of the hetero-oligomeric phosphofructokinase [
71]. For simplification, ADP/ATP and NAD
+/NADH are not shown. Single-headed arrows designate irreversible reactions, and double-headed arrows indicate reversible reactions. Proteins that were identified in the Ste20 split-ubiquitin screen are shown in green, all other proteins in blue.
Figure 3.
Overview of glycolysis, gluconeogenesis and the pentose phosphate pathway. In glycolysis (left, red arrows), glucose is oxidized to pyruvate. In gluconeogenesis (left, red arrows), pyruvate is converted to glucose 6-phosphate. Most of the gluconeogenic enzymes are also used in glycolysis. Gluconeogenesis-specific enzymes are Pyc1, Pyc2, Pck1 and Fbp1. Unlike some higher eukaryotes, budding yeast has no glucose 6-phosphatase. In contrast to many higher eukaryotes, the enzymes catalyzing the conversion of pyruvate to oxaloacetate (Pyc1 and Pyc2) are not mitochondrial but cytoplasmic [
70]. In the oxidative branch of the pentose phosphate pathway (right, green arrows) glucose 6-phosphate is converted into ribulose 5-phosphate. This part includes two reactions that generate NADPH. In the non-oxidative branch of the pentose phosphate pathway (right, black arrows), pentoses such as ribulose 5-phosphate and the glycolytic/gluconeogenic intermediates fructose 6-phosphate and glyceraldehyde 3-phosphate are interconverted. Some reactions are catalyzed by multiple redundant isoenzymes. In contrast, Pfk1 and Pfk2 are two distinct subunits of the hetero-oligomeric phosphofructokinase [
71]. For simplification, ADP/ATP and NAD
+/NADH are not shown. Single-headed arrows designate irreversible reactions, and double-headed arrows indicate reversible reactions. Proteins that were identified in the Ste20 split-ubiquitin screen are shown in green, all other proteins in blue.
![Ijms 24 15916 g003]()
Figure 4.
Ste20 binds to glucose metabolism enzymes. (
A) Ste20 interacts with enzymes of glucose metabolism using the split-ubiquitin assay. Here, 10
4 cells of the indicated plasmid combinations were spotted onto plates lacking histidine and leucine (+uracil) to select for the plasmids, and onto plates lacking histidine, leucine and uracil (−uracil) to monitor protein–protein interactions. Ste20 and the control protein Rdi1 are both fused to the C-terminal half of ubiquitin and the reporter RUra3 (
Figure 1). The other proteins are fused to the N-terminal half of ubiquitin. Leu2, Twf1, Bem1 and Cdc42 were included as controls for Ste20 and Rdi1. (
B) Co-immunoprecipitation of Ste20 and Pgk1. Cells expressing
STE20-
3HA, and
STE20-
3HA with
PGK1-
9myc were lysed and equal amounts of protein were precipitated with anti-myc antibodies. Immunoblots were probed with antibodies raised against the myc and HA epitope.
Figure 4.
Ste20 binds to glucose metabolism enzymes. (
A) Ste20 interacts with enzymes of glucose metabolism using the split-ubiquitin assay. Here, 10
4 cells of the indicated plasmid combinations were spotted onto plates lacking histidine and leucine (+uracil) to select for the plasmids, and onto plates lacking histidine, leucine and uracil (−uracil) to monitor protein–protein interactions. Ste20 and the control protein Rdi1 are both fused to the C-terminal half of ubiquitin and the reporter RUra3 (
Figure 1). The other proteins are fused to the N-terminal half of ubiquitin. Leu2, Twf1, Bem1 and Cdc42 were included as controls for Ste20 and Rdi1. (
B) Co-immunoprecipitation of Ste20 and Pgk1. Cells expressing
STE20-
3HA, and
STE20-
3HA with
PGK1-
9myc were lysed and equal amounts of protein were precipitated with anti-myc antibodies. Immunoblots were probed with antibodies raised against the myc and HA epitope.
Figure 5.
STE20 levels affect diamide sensitivity. (A) Overexpression of STE20 increases diamide sensitivity. Serial dilutions (1:10) of the indicated strains were spotted on selective medium containing either no diamide or diamide at a concentration that is sublethal for the wild type (2 mM). The STE20 overexpression strain carried the multicopy plasmid pRS426 with STE20 under control of its endogenous promoter. The wild type carried the empty pRS426 plasmid. (B) STE20 deletion confers diamide resistance. Serial dilutions (1:10) of the indicated strains were spotted on YPD medium containing either no diamide or diamide at a concentration that is lethal for the wild type (3 mM).
Figure 5.
STE20 levels affect diamide sensitivity. (A) Overexpression of STE20 increases diamide sensitivity. Serial dilutions (1:10) of the indicated strains were spotted on selective medium containing either no diamide or diamide at a concentration that is sublethal for the wild type (2 mM). The STE20 overexpression strain carried the multicopy plasmid pRS426 with STE20 under control of its endogenous promoter. The wild type carried the empty pRS426 plasmid. (B) STE20 deletion confers diamide resistance. Serial dilutions (1:10) of the indicated strains were spotted on YPD medium containing either no diamide or diamide at a concentration that is lethal for the wild type (3 mM).
In summary, Ste20 binds to a large number of glucose metabolism enzymes either directly or indirectly, and these interactions are probably physiologically relevant.
2.3. Interactions between Ste20 and Nuclear Proteins
Ste20 can translocate to the nucleus where it has a range of functions, including the regulation of apoptosis through histone H2B [
13], the attenuation of stress gene expression via histone H4 [
47] and the modulation of the transcription of sterol uptake genes through the transcriptional regulator Sut1 [
17]. Using the split-ubiquitin screen, we identified another nine nuclear proteins as Ste20 interactors (
Table 1). In this study, we focus on three key regulators of various steps of gene expression: Sac3, Hmt1 and Ctk1.
Sac3 is a component of the transcription export complex 2 (TREX-2). This protein complex plays a central role in the integration of the nuclear export of mature mRNA with earlier steps in the gene expression pathway including transcription and RNA processing [
72]. Sac3 serves as a scaffold that binds the other TREX-2 components, Thp1, Sus1, Sem1 and Cdc31 [
73,
74,
75,
76] (
Figure 6A). The binding sites of Sac3 for mRNA, the principal mRNA export factors Mex67-Mtr2, the TREX-2 components Thp1, Sem1, Sus1 and Cdc31, the nuclear pore complexes and the Mediator complex, an essential regulator of RNA polymerase II, can all be found in Sac3 residues ~1–805 (
Figure 6A). In contrast, no proteins are known to interact with the C-terminal residues ~806–1301, and no function has been attributed to this region. It is noteworthy, that in our split-ubiquitin screen, the full-length Sac3 was not identified as a Ste20 interactor, but a C-terminal Sac3 fragment comprising only residues 968–1301. However, the full-length Sac3 also binds to Ste20 (
Figure 6B). Since the C-terminal 334 residues of Sac3 are sufficient for interaction with Ste20, more Sac3 fragments were generated to further narrow down the Ste20-binding site (
Figure 6A). Ste20 interacted with truncated Sac3 comprising C-terminal 400, 300, 200 and 100 residues (
Figure 6B). However, no binding was observed for a Sac3 fragment of the last 50 residues (
Figure 6B), indicating that the 100 C-terminal Sac3 residues are sufficient and necessary for the Ste20 interaction.
Ste20 lacking its nuclear localization signal (NLS) (Ste20
∆NLS) is functional but it no longer translocates to the nucleus [
17]. The full-length Sac3 did not bind to Ste20
∆NLS (
Figure 6B), suggesting that the Ste20–Sac3 interaction takes place in the nucleus. Ste20 that lacks its NLS still binds to Bem1 (
Figure 6B), an interaction that occurs at the plasma membrane [
26,
28], demonstrating that Ste20
∆NLS is suitable for monitoring protein–protein interactions outside the nucleus. All C-terminal fragments of Sac3, apart from the smallest one, also bound to Ste20
∆NLS (
Figure 6B). This is not unexpected because a Sac3 NLS has been predicted for residues 701–717 [
81] (
Figure 6A). The truncated Sac3 versions lack this putative NLS and, therefore, presumably can be found in the cytoplasm where they could interact with Ste20
∆NLS. Furthermore, Ste20 bound to the other TREX-2 subunits (Thp1, Sem1, Sus1 and Cdc31) (
Figure 6B). These interactions all required the Ste20 NLS (
Figure 6B).
For these experiments, Cla4 was included as a control because it is a PAK related to Ste20 [
51], and like Ste20, Cla4 localizes not only to the cytoplasm and plasma membrane but also the nucleus [
17,
82]. Cla4 bound to the control protein Bem1, confirming a well-established link [
30], but it did not interact with the full-length Sac3, any of the Sac3 fragments and the other TREX-2 components (
Figure 6B). This again demonstrates the high specificity of split-ubiquitin interactions.
Several large-scale screens revealed negative genetic interactions including synthetic lethality between
CLA4 and
SAC3 [
83,
84,
85,
86],
SUS1 [
84],
SEM1 [
83,
84,
85,
86,
87,
88,
89,
90] and
THP1 [
84] deletions. These interactions were confirmed here (
Figure 6C). In our hands, the deletion of either
SAC3,
SEM1,
SUS1 or
THP1 did not affect the growth rate of these strains (
Figure 6C). However, in all cases, we observed synthetic lethality when combined with the deletion of
CLA4 deletion (
Figure 6C). In contrast,
STE20 deletion has no effect on the growth of cells also lacking
SAC3,
SUS1,
SEM1 or
THP1 (
Figure 6C). As mentioned above (
Figure 2), such a high specificity of genetic and protein–protein interactions suggests that Sac3 and the other TREX-2 components act in the same pathway as Ste20, possibly as downstream effectors of Ste20.
The loss of
SAC3 results in a strong nuclear accumulation of mRNA, demonstrating a crucial role for Sac3 in nuclear mRNA export [
73,
91]. Therefore, it would seem reasonable to expect numerous serious defects in cells lacking
SAC3. However, under optimal lab conditions, the
SAC3 deletion strain grows like the wild type or only displays slightly reduced growth (
Figure 6C and
Figure 7A,C), and very few other obvious defects have been reported. In this study, we found that Sac3, like Ste20, has a role in filamentous growth, a differentiation process that is triggered by nutrient depletion. Filamentous growth can be observed in both haploids and diploids, but the stimuli that lead to it, the underlying signaling pathways and the morphological responses differ slightly in these cell types [
92]. In haploid cells, the lack of a fermentable carbon source such as glucose results in an invasion of the agar substratum. This form of filamentous growth is, therefore, also known as invasive growth. Diploids grown under nitrogen limitation elongate and move away from the colony, a process that is termed pseudohyphal growth. In the homozygous
SAC3 deletion strain, diploid pseudohyphal growth is completely absent (
Figure 7A), and cells overexpressing
SAC3 display increased pseudohyphal growth (
Figure 7B). In contrast, in haploid cells,
SAC3 deletion or overexpression does not affect invasive growth (
Figure 7C,D). These distinct phenotypes have also been observed for the
ste20∆NLS mutant. Wild type
STE20 is essential for both haploid invasive and diploid pseudohyphal growth [
7,
8] (
Figure 7A,C). Cells expressing
STE20 without its NLS exhibit normal haploid invasive growth (
Figure 7C) as previously shown [
17]. Here, we found that, unlike the
STE20 wild type allele,
STE20∆NLS does not complement the diploid pseudohyphal growth defect of the homozygous
STE20 deletion strain (
Figure 7E). An overexpression of wild type
STE20 and
STE20∆NLS had no effect on filamentous growth in either haploid or diploid cells (
Figure 7B,D). Taken together, this suggests that Ste20 has an essential nuclear function in diploid pseudohyphal growth but not in haploid invasive growth.
Next, we examined Hmt1, the predominant arginine methyltransferase in budding yeast [
93]. Through the methylation of proteins such as Hrp1, Nab2, Npl3, Snp1, Yra1 and histone H4, Hmt1 regulates a wide range of processes in gene expression. This includes transcription elongation and termination, the splicing of pre-mRNA, the nuclear export of mRNA and the formation of silent chromatin [
94,
95,
96,
97,
98,
99].
Hmt1 binds to Ste20, an interaction that requires the nuclear localization of Ste20 (
Figure 6B). The related PAK Cla4 does not associate with Hmt1 (
Figure 6B). Like the
ste20∆NLS mutant and the
SAC3 deletion strain,
hmt1∆ cells display normal haploid invasive growth (
Figure 7C), and diploid pseudohyphal growth is completely absent in
hmt1∆/
hmt1∆ cells (
Figure 7A). Furthermore,
HMT1 overexpression results in a strong increase in diploid pseudohyphal growth (
Figure 7B) but has no effect in haploid cells (
Figure 7D).
Finally, we examined the link between Ctk1 and Ste20. Ctk1 is the catalytic subunit of the heterotrimeric C-terminal domain (CTD) kinase I complex [
100]. It phosphorylates the CTD of Rpo21 (also known as Rpb1), the largest RNA polymerase II subunit. This CTD consists of tandem heptapeptide repeats whose dynamic phosphorylation coordinates transcription with co-transcriptional events [
101]. For example, CTD phosphorylation by Ctk1 plays an important role in pre-mRNA 3′ processing [
102,
103].
Like the other nuclear interactors of Ste20 tested here, the binding of Ste20 to Ctk1 requires the nuclear localization of Ste20, and Ctk1 does not interact with Cla4 (
Figure 6B). Since
CTK1 is essential in the Σ1278b background [
104], which is used for filamentous growth assays, the effect of
CTK1 deletion on invasive growth and pseudohyphal growth could not be tested. As for
SAC3 and
HMT1, the overexpression of
CTK1 increased diploid pseudohyphal growth (
Figure 7B) but had no effect on haploid invasive growth (
Figure 7D).
As reported previously, cells lacking
CTK1 grew extremely slowly (
Figure 8A) [
100]. The deletion of
CLA4 in the
ctk1∆ strain had no effect on the growth rate but, unexpectedly,
STE20 deletion rescued the severe growth defect of
ctk1∆ cells (
Figure 8A). When a copy of wild type
STE20 was brought back to the
ctk1∆
ste20∆ double mutant, growth was again greatly reduced as expected, but expression of
STE20∆NLS had no effect in the
CTK1 STE20 double deletion strain (
Figure 8A). These observations suggest that Ste20 and Ctk1 act in the same process in the nucleus, possibly the modification of RNA polymerase II CTD. Our observations can be explained by a model in which Ste20 functions as an inhibitor of an unknown protein that, like Ctk1, modifies the CTD (
Figure 8B). However, compared with Ctk1, CTD modification by this hypothetical protein is less important. Since Ctk1 plays such a crucial role in CTD phosphorylation, cells lacking
CTK1 display a severe growth defect but are still viable due to activity of other kinases including the hypothetical protein. Due to the lack of Ste20 inhibition in a
STE20 deletion strain, the hypothetical protein would be more active but this would not affect cell growth. However, in the
ctk1∆
ste20∆ double mutant, increased activity of the unknown protein could compensate for the absence of
CTK1. This model would suggest that
STE20 overexpression could inhibit the unknown protein even more (
Figure 8B). In the presence of
CTK1, reduced activity of the hypothetical protein probably would not affect the growth rate. However, in a
ctk1∆ strain,
STE20 overexpression could exacerbate the growth defect of
ctk1∆. To test whether this is the case, we overexpressed
STE20 and
STE20∆NLS using the strong, inducible
GAL1 promoter. Increased
STE20 levels did not affect the growth of wild type cells (
Figure 8C). In the
ctk1∆ strain,
STE20 overexpression reduced the growth even further. Notably, the overexpression of
STE20 lacking its NLS had no effect (
Figure 8C). These observations are in line with the proposed model, suggesting that Ste20 has a nuclear function acting in parallel with Ctk1.
In summary, Ste20 can translocate to the nucleus where it binds with high specificity to Sac3, Hmt1 and Ctk1. These proteins are key regulators integrating various aspects of gene expression from transcription to nuclear mRNA export. Nuclear Ste20, Sac3, Hmt1 and Ctk1 all play important roles in diploid pseudohyphal growth. This functional overlap and physical, as well as genetic, interactions suggest that nuclear Ste20 acts in the same processes as Sac3, Hmt1 and Ctk1.
2.4. Other Interactions
The ribosomal protein Rpl13B and seven other proteins involved in translation and protein folding were identified in the split-ubiquitin screen as Ste20 interactors (
Table 1). This was further analyzed to rule out that these interactions are just artefacts that occur when Ste20 is synthesized at ribosomes. Rpl13B, a protein of the large ribosomal subunit, and the ribosome-associated chaperones Ssb2 and Zuo1 all bound specifically to Ste20 but not to the control protein Rdi1 (
Figure 9A). Rpl13A, the highly similar paralog of Rpl13B, which was included here as another control, did not interact with either Ste20 or Rdi1.
CLA4 deletion has been reported to be synthetically lethal with the deletion of seven genes encoding ribosomal subunits (
RPL17A,
RPL19A,
RPL23A,
RPL24A,
RPL35A,
RPS21A,
RPS21B) [
83,
87,
88]. In contrast, no genetic interactions between
STE20 and any genes encoding ribosomal proteins have been published. Because of the high number of known genetic interactions between
CLA4 and ribosomal genes, we also analyzed
RPL13B and its paralog
RPL13A. The deletion of
RPL13B results in a severe growth defect, whereas
rpl13a∆ cells grow like the wild type (
Figure 9B).
STE20 deletion in these strains does not alter the growth rate. In contrast, the
rpl13b∆
cla4∆ double deletion strain is not viable (
Figure 9B). This synthetic lethal interaction is highly specific because
rpl13a∆
cla4∆ cells are indistinguishable from the wild type.
Another group of Ste20 interactors identified in the split-ubiquitin screen comprises five proteins with ubiquitin-related functions (
Table 1). Since the screen utilizes ubiquitin and ubiquitin-specific proteases (
Figure 1), it was important to find out whether the identified proteins were simply artefacts. Ubp3, a ubiquitin-specific protease [
105], and Ubx7, a ubiquitin regulatory X (UBX) domain-containing protein involved in ubiquitin-dependent protein degradation [
106,
107], both bound to Ste20 only and not to the control protein Rdi1. This suggests that these interactions are indeed specific and independent of the experimental approach employing ubiquitin (
Figure 9A).
Taken together, the high specificity of protein–protein and genetic interactions argues against the notion that the interactors identified in the Ste20 split-ubiquitin screen merely represent artefacts. Rather, the interactions seem physiological with a potential role for Ste20 modulating translation and protein folding as well as ubiquitin-dependent protein degradation.